zcy
/
solvespace
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
solvespace/src/drawconstraint.cpp

1233 lines
47 KiB
C++
Raw Normal View History

Changes in preparation for the release of SolveSpace under the GPL, to add that license, and change all copyright notices to me, not Useful Subset, LLC. [git-p4: depot-paths = "//depot/solvespace/": change = 2211]
2013-07-28 22:08:34 +00:00
//-----------------------------------------------------------------------------
// Given a constraint, draw a graphical and user-selectable representation
// of that constraint on-screen. We can either draw with gl, or compute the
// distance from a point (the location of the mouse pointer) to the lines
// that we would have drawn, for selection.
//
// Copyright 2008-2013 Jonathan Westhues.
//-----------------------------------------------------------------------------
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
#include "solvespace.h"
Add const qualifiers to functions where trivially possible. This will allow us in future to accept `const T &` anywhere it's necessary to reduce the amount of copying. This commit is quite conservative: it does not attempt very hard to refactor code that performs incidental mutation. In particular dogd and caches are not marked with the `mutable` keyword. dogd will be eliminated later, opening up more opportunities to add const qualifiers. This commit also doesn't introduce any uses of the newly added const qualifers. This will be done later.
2016-05-21 05:18:00 +00:00
std::string Constraint::Label() const {
Replace NameStr with std::string. This removes the arbitrary 64 byte restriction (which effectively limits us to as little as 16 Unicode characters with CJK encodings), makes classes smaller, and is easier to use. As a consequence of making the length of all ex-NameStr fields unbounded, all functions that returned a buffer derived from those were changed to return std::string. Then, functions that are contextually similar to the ones described above were changed to return std::string. Then, functions that now happened to mostly take an std::string argument converted to a C string were changed to accept std::string. This has produced a bit of churn, but is probably for the better.
2015-11-06 08:40:12 +00:00
std::string result;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(type == Type::ANGLE) {
Do not display trailing zeroes for integer angles. Angles that aren't an integral number of degrees are extremely rare, and the label is often in a tight space, which warrants special-casing this.
2016-04-16 01:13:06 +00:00
if(valA == floor(valA)) {
result = ssprintf("%.0f°", valA);
} else {
result = ssprintf("%.2f°", valA);
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
} else if(type == Type::LENGTH_RATIO) {
Replace NameStr with std::string. This removes the arbitrary 64 byte restriction (which effectively limits us to as little as 16 Unicode characters with CJK encodings), makes classes smaller, and is easier to use. As a consequence of making the length of all ex-NameStr fields unbounded, all functions that returned a buffer derived from those were changed to return std::string. Then, functions that are contextually similar to the ones described above were changed to return std::string. Then, functions that now happened to mostly take an std::string argument converted to a C string were changed to accept std::string. This has produced a bit of churn, but is probably for the better.
2015-11-06 08:40:12 +00:00
result = ssprintf("%.3f:1", valA);
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
} else if(type == Type::COMMENT) {
Replace NameStr with std::string. This removes the arbitrary 64 byte restriction (which effectively limits us to as little as 16 Unicode characters with CJK encodings), makes classes smaller, and is easier to use. As a consequence of making the length of all ex-NameStr fields unbounded, all functions that returned a buffer derived from those were changed to return std::string. Then, functions that are contextually similar to the ones described above were changed to return std::string. Then, functions that now happened to mostly take an std::string argument converted to a C string were changed to accept std::string. This has produced a bit of churn, but is probably for the better.
2015-11-06 08:40:12 +00:00
result = comment;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
} else if(type == Type::DIAMETER) {
Allow displaying and editing diameter constraints as radius. The savefile format is not changed; the display option is stored in Constraint::other.
2015-03-22 13:07:49 +00:00
if(!other) {
Use the Unicode diameter symbol ⌀ in diameter constraints. Before this commit, the diameter symbol was drawn using custom code, which is not necessary anymore.
2016-04-16 01:07:35 +00:00
result = "⌀" + SS.MmToString(valA);
Allow displaying and editing diameter constraints as radius. The savefile format is not changed; the display option is stored in Constraint::other.
2015-03-22 13:07:49 +00:00
} else {
Replace NameStr with std::string. This removes the arbitrary 64 byte restriction (which effectively limits us to as little as 16 Unicode characters with CJK encodings), makes classes smaller, and is easier to use. As a consequence of making the length of all ex-NameStr fields unbounded, all functions that returned a buffer derived from those were changed to return std::string. Then, functions that are contextually similar to the ones described above were changed to return std::string. Then, functions that now happened to mostly take an std::string argument converted to a C string were changed to accept std::string. This has produced a bit of churn, but is probably for the better.
2015-11-06 08:40:12 +00:00
result = "R" + SS.MmToString(valA / 2);
Allow displaying and editing diameter constraints as radius. The savefile format is not changed; the display option is stored in Constraint::other.
2015-03-22 13:07:49 +00:00
}
Add reference dimensions, add user-programmable lighting, and add a configuration screen where all this can be specified. [git-p4: depot-paths = "//depot/solvespace/": change = 1783]
2008-06-11 04:22:52 +00:00
} else {
Don't keep a dimension or group's user-entered numerical value as an Expr *, since that complicates multiple units and also memory management. Now it's just a double. [git-p4: depot-paths = "//depot/solvespace/": change = 1791]
2008-06-14 08:43:38 +00:00
// valA has units of distance
Replace NameStr with std::string. This removes the arbitrary 64 byte restriction (which effectively limits us to as little as 16 Unicode characters with CJK encodings), makes classes smaller, and is easier to use. As a consequence of making the length of all ex-NameStr fields unbounded, all functions that returned a buffer derived from those were changed to return std::string. Then, functions that are contextually similar to the ones described above were changed to return std::string. Then, functions that now happened to mostly take an std::string argument converted to a C string were changed to accept std::string. This has produced a bit of churn, but is probably for the better.
2015-11-06 08:40:12 +00:00
result = SS.MmToString(fabs(valA));
Add reference dimensions, add user-programmable lighting, and add a configuration screen where all this can be specified. [git-p4: depot-paths = "//depot/solvespace/": change = 1783]
2008-06-11 04:22:52 +00:00
}
if(reference) {
Replace NameStr with std::string. This removes the arbitrary 64 byte restriction (which effectively limits us to as little as 16 Unicode characters with CJK encodings), makes classes smaller, and is easier to use. As a consequence of making the length of all ex-NameStr fields unbounded, all functions that returned a buffer derived from those were changed to return std::string. Then, functions that are contextually similar to the ones described above were changed to return std::string. Then, functions that now happened to mostly take an std::string argument converted to a C string were changed to accept std::string. This has produced a bit of churn, but is probably for the better.
2015-11-06 08:40:12 +00:00
result += " REF";
Add reference dimensions, add user-programmable lighting, and add a configuration screen where all this can be specified. [git-p4: depot-paths = "//depot/solvespace/": change = 1783]
2008-06-11 04:22:52 +00:00
}
Replace NameStr with std::string. This removes the arbitrary 64 byte restriction (which effectively limits us to as little as 16 Unicode characters with CJK encodings), makes classes smaller, and is easier to use. As a consequence of making the length of all ex-NameStr fields unbounded, all functions that returned a buffer derived from those were changed to return std::string. Then, functions that are contextually similar to the ones described above were changed to return std::string. Then, functions that now happened to mostly take an std::string argument converted to a C string were changed to accept std::string. This has produced a bit of churn, but is probably for the better.
2015-11-06 08:40:12 +00:00
return result;
Add reference dimensions, add user-programmable lighting, and add a configuration screen where all this can be specified. [git-p4: depot-paths = "//depot/solvespace/": change = 1783]
2008-06-11 04:22:52 +00:00
}
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
void Constraint::DoLine(Canvas *canvas, Canvas::hStroke hcs, Vector a, Vector b) {
const Camera &camera = canvas->GetCamera();
a = camera.AlignToPixelGrid(a);
b = camera.AlignToPixelGrid(b);
canvas->DrawLine(a, b, hcs);
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::DoLabel(Canvas *canvas, Canvas::hStroke hcs,
Vector ref, Vector *labelPos, Vector gr, Vector gu) {
const Camera &camera = canvas->GetCamera();
Add ability to assign styles to cosmetic text (in the form of Constraint::COMMENTs), including line width and color, and text height and origin location. [git-p4: depot-paths = "//depot/solvespace/": change = 2033]
2009-09-24 15:52:48 +00:00
Replace NameStr with std::string. This removes the arbitrary 64 byte restriction (which effectively limits us to as little as 16 Unicode characters with CJK encodings), makes classes smaller, and is easier to use. As a consequence of making the length of all ex-NameStr fields unbounded, all functions that returned a buffer derived from those were changed to return std::string. Then, functions that are contextually similar to the ones described above were changed to return std::string. Then, functions that now happened to mostly take an std::string argument converted to a C string were changed to accept std::string. This has produced a bit of churn, but is probably for the better.
2015-11-06 08:40:12 +00:00
std::string s = Label();
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
double textHeight = Style::TextHeight(GetStyle()) / camera.scale;
double swidth = VectorFont::Builtin()->GetWidth(textHeight, s),
sheight = VectorFont::Builtin()->GetCapHeight(textHeight);
Add ability to assign styles to cosmetic text (in the form of Constraint::COMMENTs), including line width and color, and text height and origin location. [git-p4: depot-paths = "//depot/solvespace/": change = 2033]
2009-09-24 15:52:48 +00:00
// By default, the reference is from the center; but the style could
Add text angle for styles. Add ability to quickly change between perspective and parallel projections. Add a snap grid, for points and for text comments. Draw text comments in the plane of their workplane if they have one, otherwise always facing forward. And fix a few nasty bugs: the possibility of an extremely long animation onto a workplane, accidental use of the wrong style line width for constraints, misplaced text box in style screen for default styles, other little stuff. [git-p4: depot-paths = "//depot/solvespace/": change = 2037]
2009-09-29 11:35:19 +00:00
// specify otherwise if one is present, and it could also specify a
// rotation.
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(type == Type::COMMENT && disp.style.v) {
Add text angle for styles. Add ability to quickly change between perspective and parallel projections. Add a snap grid, for points and for text comments. Draw text comments in the plane of their workplane if they have one, otherwise always facing forward. And fix a few nasty bugs: the possibility of an extremely long animation onto a workplane, accidental use of the wrong style line width for constraints, misplaced text box in style screen for default styles, other little stuff. [git-p4: depot-paths = "//depot/solvespace/": change = 2037]
2009-09-29 11:35:19 +00:00
Style *st = Style::Get(disp.style);
// rotation first
double rads = st->textAngle*PI/180;
double c = cos(rads), s = sin(rads);
Vector pr = gr, pu = gu;
gr = pr.ScaledBy( c).Plus(pu.ScaledBy(s));
gu = pr.ScaledBy(-s).Plus(pu.ScaledBy(c));
// then origin
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
uint32_t o = (uint32_t)st->textOrigin;
if(o & (uint32_t)Style::TextOrigin::LEFT) ref = ref.Plus(gr.WithMagnitude(swidth/2));
if(o & (uint32_t)Style::TextOrigin::RIGHT) ref = ref.Minus(gr.WithMagnitude(swidth/2));
if(o & (uint32_t)Style::TextOrigin::BOT) ref = ref.Plus(gu.WithMagnitude(sheight/2));
if(o & (uint32_t)Style::TextOrigin::TOP) ref = ref.Minus(gu.WithMagnitude(sheight/2));
Add ability to assign styles to cosmetic text (in the form of Constraint::COMMENTs), including line width and color, and text height and origin location. [git-p4: depot-paths = "//depot/solvespace/": change = 2033]
2009-09-24 15:52:48 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Vector o = ref.Minus(gr.WithMagnitude(swidth/2)).Minus(
gu.WithMagnitude(sheight/2));
canvas->DrawVectorText(s, textHeight, o, gr.WithMagnitude(1), gu.WithMagnitude(1), hcs);
if(labelPos) *labelPos = o;
Add a diameter constraint, and add a `distance' entity that I can remap when I copy circle entities, in order to make the radius numerical somehow (analogy with the POINT_ and NORMAL_XFRMD) thing. [git-p4: depot-paths = "//depot/solvespace/": change = 1710]
2008-05-07 08:19:37 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::DoProjectedPoint(Canvas *canvas, Canvas::hStroke hcs, Vector *r) {
const Camera &camera = canvas->GetCamera();
Canvas::Stroke strokeStippled = *canvas->strokes.FindById(hcs);
strokeStippled.stipplePattern = StipplePattern::SHORT_DASH;
strokeStippled.stippleScale = 4.0 / camera.scale;
Canvas::hStroke hcsStippled = canvas->GetStroke(strokeStippled);
Add projected point-point distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 2115]
2010-01-27 18:15:06 +00:00
Vector p = r->ProjectInto(workplane);
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcsStippled, p, *r);
Add point-line distance, point-plane distance, and point-on-circle constraints. And generate the constraint equations for entities (e.g., that our unit quaternions have magnitude one). Numerical troubles there, but it sort of works. Also some stuff to draw projection lines with projected constraints, and to auto-insert more constraints as you draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1711]
2008-05-08 07:30:30 +00:00
*r = p;
}
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
//-----------------------------------------------------------------------------
// There is a rectangular box, aligned to our display axes (projRight, projUp)
// centered at ref. This is where a dimension label will be drawn. We want to
// draw a line from A to B. If that line would intersect the label box, then
// trim the line to leave a gap for it, and return zero. If not, then extend
// the line to almost meet the box, and return either positive or negative,
// depending whether that extension was from A or from B.
//-----------------------------------------------------------------------------
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
int Constraint::DoLineTrimmedAgainstBox(Canvas *canvas, Canvas::hStroke hcs,
Vector ref, Vector a, Vector b, bool extend) {
const Camera &camera = canvas->GetCamera();
double th = Style::TextHeight(GetStyle()) / camera.scale;
double pixels = 1.0 / camera.scale;
double swidth = VectorFont::Builtin()->GetWidth(th, Label()) + 4 * pixels,
sheight = VectorFont::Builtin()->GetCapHeight(th) + 8 * pixels;
Vector gu = camera.projUp.WithMagnitude(1),
gr = camera.projRight.WithMagnitude(1);
return DoLineTrimmedAgainstBox(canvas, hcs, ref, a, b, extend, gr, gu, swidth, sheight);
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
int Constraint::DoLineTrimmedAgainstBox(Canvas *canvas, Canvas::hStroke hcs,
Vector ref, Vector a, Vector b, bool extend,
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
Vector gr, Vector gu, double swidth, double sheight) {
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
struct {
Vector n;
double d;
} planes[4];
// reference pos is the center of box occupied by text; build a rectangle
// around that, aligned to axes gr and gu, from four planes will all four
// normals pointing inward
planes[0].n = gu.ScaledBy(-1); planes[0].d = -(gu.Dot(ref) + sheight/2);
planes[1].n = gu; planes[1].d = gu.Dot(ref) - sheight/2;
planes[2].n = gr; planes[2].d = gr.Dot(ref) - swidth/2;
planes[3].n = gr.ScaledBy(-1); planes[3].d = -(gr.Dot(ref) + swidth/2);
double tmin = VERY_POSITIVE, tmax = VERY_NEGATIVE;
Vector dl = b.Minus(a);
for(int i = 0; i < 4; i++) {
bool parallel;
Vector p = Vector::AtIntersectionOfPlaneAndLine(
planes[i].n, planes[i].d,
a, b, &parallel);
if(parallel) continue;
int j;
for(j = 0; j < 4; j++) {
double d = (planes[j].n).Dot(p) - planes[j].d;
if(d < -LENGTH_EPS) break;
}
if(j < 4) continue;
double t = (p.Minus(a)).DivPivoting(dl);
tmin = min(t, tmin);
tmax = max(t, tmax);
}
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
// Both in range; so there's pieces of the line on both sides of the label box.
if(tmin >= 0.0 && tmin <= 1.0 && tmax >= 0.0 && tmax <= 1.0) {
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, a, a.Plus(dl.ScaledBy(tmin)));
DoLine(canvas, hcs, a.Plus(dl.ScaledBy(tmax)), b);
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
return 0;
}
// Only one intersection in range; so the box is right on top of the endpoint
if(tmin >= 0.0 && tmin <= 1.0) {
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, a, a.Plus(dl.ScaledBy(tmin)));
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
return 0;
}
// Likewise.
if(tmax >= 0.0 && tmax <= 1.0) {
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, a.Plus(dl.ScaledBy(tmax)), b);
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
return 0;
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
}
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
// The line does not intersect the label; so the line should get
// extended to just barely meet the label.
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
// 0 means the label lies within the line, negative means it's outside
// and closer to b, positive means outside and closer to a.
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
if(tmax < 0.0) {
if(extend) a = a.Plus(dl.ScaledBy(tmax));
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, a, b);
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
return 1;
}
if(tmin > 1.0) {
if(extend) b = a.Plus(dl.ScaledBy(tmin));
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, a, b);
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
return -1;
}
// This will happen if the entire line lies within the box.
return 0;
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::DoArrow(Canvas *canvas, Canvas::hStroke hcs,
Vector p, Vector dir, Vector n, double width, double angle, double da) {
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
dir = dir.WithMagnitude(width / cos(angle));
dir = dir.RotatedAbout(n, da);
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, p, p.Plus(dir.RotatedAbout(n, angle)));
DoLine(canvas, hcs, p, p.Plus(dir.RotatedAbout(n, -angle)));
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
}
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
//-----------------------------------------------------------------------------
// Draw a line with arrows on both ends, and possibly a gap in the middle for
// the dimension. We will use these for most length dimensions. The length
// being dimensioned is from A to B; but those points get extended perpendicular
// to the line AB, until the line between the extensions crosses ref (the
// center of the label).
//-----------------------------------------------------------------------------
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::DoLineWithArrows(Canvas *canvas, Canvas::hStroke hcs,
Vector ref, Vector a, Vector b,
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
bool onlyOneExt)
{
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
const Camera &camera = canvas->GetCamera();
double pixels = 1.0 / camera.scale;
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
Vector ab = a.Minus(b);
Vector ar = a.Minus(ref);
// Normal to a plane containing the line and the label origin.
Vector n = ab.Cross(ar);
// Within that plane, and normal to the line AB; so that's our extension
// line.
Vector out = ab.Cross(n).WithMagnitude(1);
out = out.ScaledBy(-out.Dot(ar));
Vector ae = a.Plus(out), be = b.Plus(out);
// Extension lines extend 10 pixels beyond where the arrows get
// drawn (which is at the same offset perpendicular from AB as the
// label).
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, a, ae.Plus(out.WithMagnitude(10*pixels)));
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
if(!onlyOneExt) {
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, b, be.Plus(out.WithMagnitude(10*pixels)));
Draw projection lines for point-line distance constraints. This is only visible when the point and line lie within the workplane, but the constraint label does not.
2016-07-12 16:18:49 +00:00
} else {
Vector prj = be;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoProjectedPoint(canvas, hcs, &prj);
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
int within = DoLineTrimmedAgainstBox(canvas, hcs, ref, ae, be);
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
// Arrow heads are 13 pixels long, with an 18 degree half-angle.
double theta = 18*PI/180;
Vector arrow = (be.Minus(ae)).WithMagnitude(13*pixels);
if(within != 0) {
arrow = arrow.ScaledBy(-1);
Vector seg = (be.Minus(ae)).WithMagnitude(18*pixels);
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
if(within < 0) DoLine(canvas, hcs, ae, ae.Minus(seg));
if(within > 0) DoLine(canvas, hcs, be, be.Plus(seg));
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoArrow(canvas, hcs, ae, arrow, n, 13.0 * pixels, theta, 0.0);
DoArrow(canvas, hcs, be, arrow.Negated(), n, 13.0 * pixels, theta, 0.0);
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::DoEqualLenTicks(Canvas *canvas, Canvas::hStroke hcs,
Vector a, Vector b, Vector gn, Vector *refp) {
const Camera &camera = canvas->GetCamera();
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
Vector m = (a.ScaledBy(1.0/3)).Plus(b.ScaledBy(2.0/3));
Eliminate Constraint::dogd.refp. While we're at it, let's also emphasize both parts of a two-part constraint to make it easier to find.
2016-05-24 06:40:41 +00:00
if(refp) *refp = m;
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
Vector ab = a.Minus(b);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Vector n = (gn.Cross(ab)).WithMagnitude(10/camera.scale);
Trim trailing whitespace.
2015-03-29 00:30:52 +00:00
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, m.Minus(n), m.Plus(n));
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::DoEqualRadiusTicks(Canvas *canvas, Canvas::hStroke hcs,
hEntity he, Vector *refp) {
const Camera &camera = canvas->GetCamera();
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *circ = SK.GetEntity(he);
Add a constraint to make line length equal arc length. That's quite tricky; can't just use the dot product, since that blows up when you cross pi radians. A gear shift approach, use either sin or cos, same kind of thing as the 3d-parallel constraint. And report a NaN constraint as unconverged, of course. [git-p4: depot-paths = "//depot/solvespace/": change = 1890]
2009-01-08 17:22:59 +00:00
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector center = SK.GetEntity(circ->point[0])->PointGetNum();
Add a constraint to make line length equal arc length. That's quite tricky; can't just use the dot product, since that blows up when you cross pi radians. A gear shift approach, use either sin or cos, same kind of thing as the 3d-parallel constraint. And report a NaN constraint as unconverged, of course. [git-p4: depot-paths = "//depot/solvespace/": change = 1890]
2009-01-08 17:22:59 +00:00
double r = circ->CircleGetRadiusNum();
Quaternion q = circ->Normal()->NormalGetNum();
Vector u = q.RotationU(), v = q.RotationV();
double theta;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(circ->type == Entity::Type::CIRCLE) {
Add a constraint to make line length equal arc length. That's quite tricky; can't just use the dot product, since that blows up when you cross pi radians. A gear shift approach, use either sin or cos, same kind of thing as the 3d-parallel constraint. And report a NaN constraint as unconverged, of course. [git-p4: depot-paths = "//depot/solvespace/": change = 1890]
2009-01-08 17:22:59 +00:00
theta = PI/2;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
} else if(circ->type == Entity::Type::ARC_OF_CIRCLE) {
Add a constraint to make line length equal arc length. That's quite tricky; can't just use the dot product, since that blows up when you cross pi radians. A gear shift approach, use either sin or cos, same kind of thing as the 3d-parallel constraint. And report a NaN constraint as unconverged, of course. [git-p4: depot-paths = "//depot/solvespace/": change = 1890]
2009-01-08 17:22:59 +00:00
double thetaa, thetab, dtheta;
circ->ArcGetAngles(&thetaa, &thetab, &dtheta);
theta = thetaa + dtheta/2;
Replace all oops() checks with ssassert()s. This includes explanation and context for non-obvious cases and shortens debug cycles when just-in-time debugging is not available (like on Linux) by immediately printing description of the assert as well as symbolized backtrace.
2016-05-18 22:51:36 +00:00
} else ssassert(false, "Unexpected entity type");
Add a constraint to make line length equal arc length. That's quite tricky; can't just use the dot product, since that blows up when you cross pi radians. A gear shift approach, use either sin or cos, same kind of thing as the 3d-parallel constraint. And report a NaN constraint as unconverged, of course. [git-p4: depot-paths = "//depot/solvespace/": change = 1890]
2009-01-08 17:22:59 +00:00
Vector d = u.ScaledBy(cos(theta)).Plus(v.ScaledBy(sin(theta)));
d = d.ScaledBy(r);
Vector p = center.Plus(d);
Eliminate Constraint::dogd.refp. While we're at it, let's also emphasize both parts of a two-part constraint to make it easier to find.
2016-05-24 06:40:41 +00:00
if(refp) *refp = p;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Vector tick = d.WithMagnitude(10/camera.scale);
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, p.Plus(tick), p.Minus(tick));
Add a constraint to make line length equal arc length. That's quite tricky; can't just use the dot product, since that blows up when you cross pi radians. A gear shift approach, use either sin or cos, same kind of thing as the 3d-parallel constraint. And report a NaN constraint as unconverged, of course. [git-p4: depot-paths = "//depot/solvespace/": change = 1890]
2009-01-08 17:22:59 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::DoArcForAngle(Canvas *canvas, Canvas::hStroke hcs,
Vector a0, Vector da, Vector b0, Vector db,
Vector offset, Vector *ref, bool trim)
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
{
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
const Camera &camera = canvas->GetCamera();
double pixels = 1.0 / camera.scale;
Vector gr = camera.projRight.ScaledBy(1.0);
Vector gu = camera.projUp.ScaledBy(1.0);
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
if(workplane.v != Entity::FREE_IN_3D.v) {
a0 = a0.ProjectInto(workplane);
b0 = b0.ProjectInto(workplane);
da = da.ProjectVectorInto(workplane);
db = db.ProjectVectorInto(workplane);
}
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
Vector a1 = a0.Plus(da);
Vector b1 = b0.Plus(db);
Add a trim command. I can now do circles (or arcs) against lines, or lines against lines. The constraints get rather screwed up afterwards, of course. So make arcs with the endpoints coincident into circles, instead of nothings; since the first split of a circle produces that. And don't warn after deleting just point-coincident or horiz/vert constraints as a dependency; that's just a nuisance, because it happens too often. [git-p4: depot-paths = "//depot/solvespace/": change = 1884]
2009-01-03 12:27:33 +00:00
bool skew;
Trim trailing whitespace.
2015-03-29 00:30:52 +00:00
Vector pi = Vector::AtIntersectionOfLines(a0, a0.Plus(da),
Add a trim command. I can now do circles (or arcs) against lines, or lines against lines. The constraints get rather screwed up afterwards, of course. So make arcs with the endpoints coincident into circles, instead of nothings; since the first split of a circle produces that. And don't warn after deleting just point-coincident or horiz/vert constraints as a dependency; that's just a nuisance, because it happens too often. [git-p4: depot-paths = "//depot/solvespace/": change = 1884]
2009-01-03 12:27:33 +00:00
b0, b0.Plus(db), &skew);
if(!skew) {
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
*ref = pi.Plus(offset);
Add a trim command. I can now do circles (or arcs) against lines, or lines against lines. The constraints get rather screwed up afterwards, of course. So make arcs with the endpoints coincident into circles, instead of nothings; since the first split of a circle produces that. And don't warn after deleting just point-coincident or horiz/vert constraints as a dependency; that's just a nuisance, because it happens too often. [git-p4: depot-paths = "//depot/solvespace/": change = 1884]
2009-01-03 12:27:33 +00:00
// We draw in a coordinate system centered at the intersection point.
// One basis vector is da, and the other is normal to da and in
// the plane that contains our lines (so normal to its normal).
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
da = da.WithMagnitude(1);
db = db.WithMagnitude(1);
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
Vector norm = da.Cross(db);
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
Vector dna = norm.Cross(da).WithMagnitude(1.0);
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
Vector dnb = norm.Cross(db).WithMagnitude(1.0);
Add a trim command. I can now do circles (or arcs) against lines, or lines against lines. The constraints get rather screwed up afterwards, of course. So make arcs with the endpoints coincident into circles, instead of nothings; since the first split of a circle produces that. And don't warn after deleting just point-coincident or horiz/vert constraints as a dependency; that's just a nuisance, because it happens too often. [git-p4: depot-paths = "//depot/solvespace/": change = 1884]
2009-01-03 12:27:33 +00:00
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
// da and db magnitudes are 1.0
double thetaf = acos(da.Dot(db));
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
// Calculate median
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
Vector m = da.ScaledBy(cos(thetaf/2)).Plus(
dna.ScaledBy(sin(thetaf/2)));
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
Vector rm = (*ref).Minus(pi);
// Test which side we have to place an arc
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
if(m.Dot(rm) < 0) {
da = da.ScaledBy(-1); dna = dna.ScaledBy(-1);
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
db = db.ScaledBy(-1); dnb = dnb.ScaledBy(-1);
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
}
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
double rda = rm.Dot(da), rdna = rm.Dot(dna);
// Introduce minimal arc radius in pixels
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
double r = max(sqrt(rda*rda + rdna*rdna), 15.0 * pixels);
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
double th = Style::TextHeight(GetStyle()) / camera.scale;
double swidth = VectorFont::Builtin()->GetWidth(th, Label()) + 4*pixels,
sheight = VectorFont::Builtin()->GetCapHeight(th) + 8*pixels;
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
double textR = sqrt(swidth * swidth + sheight * sheight) / 2.0;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
*ref = pi.Plus(rm.WithMagnitude(std::max(rm.Magnitude(), 15 * pixels + textR)));
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
// Arrow points
Vector apa = da. ScaledBy(r).Plus(pi);
Vector apb = da. ScaledBy(r*cos(thetaf)).Plus(
dna.ScaledBy(r*sin(thetaf))).Plus(pi);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
double arrowW = 13 * pixels;
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
double arrowA = 18.0 * PI / 180.0;
bool arrowVisible = apb.Minus(apa).Magnitude() > 2.5 * arrowW;
// Arrow reversing indicator
bool arrowRev = false;
// The minimal extension length in angular representation
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
double extAngle = 18 * pixels / r;
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
// Arc additional angle
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
double addAngle = 0.0;
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
// Arc start angle
double startAngle = 0.0;
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
// Arc extension to db.
// We have just enlarge angle value.
if(HasLabel() && rm.Dot(dnb) > 0.0) {
// rm direction projected to plane with u = da, v = dna
Vector rmp = da.ScaledBy(rda).Plus(dna.ScaledBy(rdna)).WithMagnitude(1.0);
// rmp and db magnitudes are 1.0
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
addAngle = std::max(acos(rmp.Dot(db)), extAngle);
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
if(arrowVisible) {
startAngle = -extAngle;
addAngle += extAngle;
arrowRev = true;
}
}
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
// Arc extension to da.
// We are enlarge angle value and rewrite basis to align along rm projection.
if(HasLabel() && rm.Dot(dna) < 0.0) {
// rm direction projected to plane with u = da, v = dna
Vector rmp = da.ScaledBy(rda).Plus(dna.ScaledBy(rdna)).WithMagnitude(1.0);
// rmp and da magnitudes are 1.0
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
startAngle = -std::max(acos(rmp.Dot(da)), extAngle);
addAngle = -startAngle;
if(arrowVisible) {
addAngle += extAngle;
arrowRev = true;
}
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
}
Vector prev;
int n = 30;
for(int i = 0; i <= n; i++) {
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
double theta = startAngle + (i*(thetaf + addAngle))/n;
Vector p = da.ScaledBy(r*cos(theta)).Plus(
dna.ScaledBy(r*sin(theta))).Plus(pi);
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
if(i > 0) {
if(trim) {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLineTrimmedAgainstBox(canvas, hcs, *ref, prev, p,
/*extend=*/false, gr, gu, swidth, sheight);
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
} else {
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, prev, p);
Continue an extension arc towards angle constraint label.
2016-04-15 11:57:44 +00:00
}
Display angle in the middle of its dimension line, not adjacent. Less visual clutter and more consistency with distance constraints.
2016-04-13 02:45:52 +00:00
}
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
prev = p;
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLineExtend(canvas, hcs, a0, a1, apa, 5.0 * pixels);
DoLineExtend(canvas, hcs, b0, b1, apb, 5.0 * pixels);
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
// Draw arrows only when we have enough space.
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
if(arrowVisible) {
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
double angleCorr = arrowW / (2.0 * r);
Reverse arrows on an angle constraint when the label is outside.
2016-04-16 11:58:23 +00:00
if(arrowRev) {
dna = dna.ScaledBy(-1.0);
angleCorr = -angleCorr;
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoArrow(canvas, hcs, apa, dna, norm, arrowW, arrowA, angleCorr);
DoArrow(canvas, hcs, apb, dna, norm, arrowW, arrowA, thetaf + PI - angleCorr);
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
}
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
} else {
// The lines are skew; no wonderful way to illustrate that.
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
*ref = a0.Plus(b0);
*ref = (*ref).ScaledBy(0.5).Plus(disp.offset);
Add a mechanism to record the lines drawn when we display a constraint, so that we can export that too. This includes the lines for the vector font. A little ugly; it needs some kind of line style (color or width) to distinguish those lines from the real geometry in the exported file. [git-p4: depot-paths = "//depot/solvespace/": change = 2007]
2009-07-03 20:55:57 +00:00
gu = gu.WithMagnitude(1);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
double textHeight = Style::TextHeight(GetStyle()) / camera.scale;
Add ability to assign styles to cosmetic text (in the form of Constraint::COMMENTs), including line width and color, and text height and origin location. [git-p4: depot-paths = "//depot/solvespace/": change = 2033]
2009-09-24 15:52:48 +00:00
Vector trans =
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
(*ref).Plus(gu.ScaledBy(-1.5*VectorFont::Builtin()->GetCapHeight(textHeight)));
canvas->DrawVectorText("angle between skew lines", textHeight,
trans, gr.WithMagnitude(pixels), gu.WithMagnitude(pixels),
hcs);
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
}
}
DXF: only export visible constraints.
2016-03-24 13:55:36 +00:00
bool Constraint::IsVisible() const {
if(!SS.GW.showConstraints) return false;
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Group *g = SK.GetGroup(group);
Tear apart the text window, mostly to beautify things. The foreground and background colours are now specified separately, and it's possible to insert half-line spaces. So now I have a window that lets me show/hide groups, and select the active one. [git-p4: depot-paths = "//depot/solvespace/": change = 1695]
2008-04-28 07:18:39 +00:00
// If the group is hidden, then the constraints are hidden and not
// able to be selected.
DXF: only export visible constraints.
2016-03-24 13:55:36 +00:00
if(!(g->visible)) return false;
Add text angle for styles. Add ability to quickly change between perspective and parallel projections. Add a snap grid, for points and for text comments. Draw text comments in the plane of their workplane if they have one, otherwise always facing forward. And fix a few nasty bugs: the possibility of an extremely long animation onto a workplane, accidental use of the wrong style line width for constraints, misplaced text box in style screen for default styles, other little stuff. [git-p4: depot-paths = "//depot/solvespace/": change = 2037]
2009-09-29 11:35:19 +00:00
// And likewise if the group is not the active group; except for comments
// with an assigned style.
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(g->h.v != SS.GW.activeGroup.v && !(type == Type::COMMENT && disp.style.v)) {
DXF: only export visible constraints.
2016-03-24 13:55:36 +00:00
return false;
Add text angle for styles. Add ability to quickly change between perspective and parallel projections. Add a snap grid, for points and for text comments. Draw text comments in the plane of their workplane if they have one, otherwise always facing forward. And fix a few nasty bugs: the possibility of an extremely long animation onto a workplane, accidental use of the wrong style line width for constraints, misplaced text box in style screen for default styles, other little stuff. [git-p4: depot-paths = "//depot/solvespace/": change = 2037]
2009-09-29 11:35:19 +00:00
}
if(disp.style.v) {
Style *s = Style::Get(disp.style);
DXF: only export visible constraints.
2016-03-24 13:55:36 +00:00
if(!s->visible) return false;
Add text angle for styles. Add ability to quickly change between perspective and parallel projections. Add a snap grid, for points and for text comments. Draw text comments in the plane of their workplane if they have one, otherwise always facing forward. And fix a few nasty bugs: the possibility of an extremely long animation onto a workplane, accidental use of the wrong style line width for constraints, misplaced text box in style screen for default styles, other little stuff. [git-p4: depot-paths = "//depot/solvespace/": change = 2037]
2009-09-29 11:35:19 +00:00
}
DXF: only export visible constraints.
2016-03-24 13:55:36 +00:00
return true;
}
Start to add some constraint stuff. I now have point-coincident, and point-in-plane. These work, but the equation is still stupid, solving everything at once and not substituting trivial equations. [git-p4: depot-paths = "//depot/solvespace/": change = 1677]
2008-04-21 08:16:38 +00:00
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
bool Constraint::DoLineExtend(Canvas *canvas, Canvas::hStroke hcs,
Vector p0, Vector p1, Vector pt, double salient) {
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
Vector dir = p1.Minus(p0);
double k = dir.Dot(pt.Minus(p0)) / dir.Dot(dir);
Vector ptOnLine = p0.Plus(dir.ScaledBy(k));
// Draw projection line.
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, pt, ptOnLine);
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
// Calculate salient direction.
Vector sd = dir.WithMagnitude(1.0).ScaledBy(salient);
Vector from;
Vector to;
if(k < 0.0) {
from = p0;
to = ptOnLine.Minus(sd);
} else if(k > 1.0) {
from = p1;
to = ptOnLine.Plus(sd);
} else {
return false;
}
// Draw extension line.
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, from, to);
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
return true;
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::DoLayout(DrawAs how, Canvas *canvas,
Vector *labelPos, std::vector<Vector> *refs) {
DXF: only export visible constraints.
2016-03-24 13:55:36 +00:00
if(!IsVisible()) return;
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
Add point on line constraints, in 2d and 3d. The 3d equations do not have much motivation behind them, but they seem to work. And make sure that we don't solve multiple times without repainting in between, and tweak the text window a bit more. [git-p4: depot-paths = "//depot/solvespace/": change = 1696]
2008-04-28 09:40:02 +00:00
// Unit vectors that describe our current view of the scene. One pixel
// long, not one actual unit.
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
const Camera &camera = canvas->GetCamera();
Vector gr = camera.projRight.ScaledBy(1/camera.scale);
Vector gu = camera.projUp.ScaledBy(1/camera.scale);
Vector gn = (gr.Cross(gu)).WithMagnitude(1/camera.scale);
double textHeight = Style::TextHeight(GetStyle()) / camera.scale;
RgbaColor color = {};
switch(how) {
case DrawAs::DEFAULT: color = Style::Color(GetStyle()); break;
case DrawAs::HOVERED: color = Style::Color(Style::HOVERED); break;
case DrawAs::SELECTED: color = Style::Color(Style::SELECTED); break;
}
Canvas::Stroke stroke = {};
stroke.layer = Canvas::Layer::FRONT;
stroke.color = color;
stroke.width = Style::Width(GetStyle());
stroke.zIndex = 4;
Canvas::hStroke hcs = canvas->GetStroke(stroke);
Canvas::Fill fill = {};
fill.layer = Canvas::Layer::FRONT;
fill.color = color;
fill.zIndex = stroke.zIndex;
Canvas::hFill hcf = canvas->GetFill(fill);
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
switch(type) {
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::PT_PT_DISTANCE: {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector ap = SK.GetEntity(ptA)->PointGetNum();
Vector bp = SK.GetEntity(ptB)->PointGetNum();
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
Add point-line distance, point-plane distance, and point-on-circle constraints. And generate the constraint equations for entities (e.g., that our unit quaternions have magnitude one). Numerical troubles there, but it sort of works. Also some stuff to draw projection lines with projected constraints, and to auto-insert more constraints as you draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1711]
2008-05-08 07:30:30 +00:00
if(workplane.v != Entity::FREE_IN_3D.v) {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoProjectedPoint(canvas, hcs, &ap);
DoProjectedPoint(canvas, hcs, &bp);
Add point-line distance, point-plane distance, and point-on-circle constraints. And generate the constraint equations for entities (e.g., that our unit quaternions have magnitude one). Numerical troubles there, but it sort of works. Also some stuff to draw projection lines with projected constraints, and to auto-insert more constraints as you draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1711]
2008-05-08 07:30:30 +00:00
}
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
Vector ref = ((ap.Plus(bp)).ScaledBy(0.5)).Plus(disp.offset);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(ref);
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLineWithArrows(canvas, hcs, ref, ap, bp, /*onlyOneExt=*/false);
DoLabel(canvas, hcs, ref, labelPos, gr, gu);
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add a diameter constraint, and add a `distance' entity that I can remap when I copy circle entities, in order to make the radius numerical somehow (analogy with the POINT_ and NORMAL_XFRMD) thing. [git-p4: depot-paths = "//depot/solvespace/": change = 1710]
2008-05-07 08:19:37 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::PROJ_PT_DISTANCE: {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Canvas::Stroke strokeStippled = stroke;
strokeStippled.stipplePattern = StipplePattern::SHORT_DASH;
strokeStippled.stippleScale = 4.0 / camera.scale;
Canvas::hStroke hcsStippled = canvas->GetStroke(strokeStippled);
Add projected point-point distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 2115]
2010-01-27 18:15:06 +00:00
Vector ap = SK.GetEntity(ptA)->PointGetNum(),
bp = SK.GetEntity(ptB)->PointGetNum(),
dp = (bp.Minus(ap)),
pp = SK.GetEntity(entityA)->VectorGetNum();
Vector ref = ((ap.Plus(bp)).ScaledBy(0.5)).Plus(disp.offset);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(ref);
Add projected point-point distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 2115]
2010-01-27 18:15:06 +00:00
pp = pp.WithMagnitude(1);
double d = dp.Dot(pp);
Vector bpp = ap.Plus(pp.ScaledBy(d));
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcsStippled, ap, bpp);
DoLine(canvas, hcsStippled, bp, bpp);
Add projected point-point distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 2115]
2010-01-27 18:15:06 +00:00
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLineWithArrows(canvas, hcs, ref, ap, bpp, /*onlyOneExt=*/false);
DoLabel(canvas, hcs, ref, labelPos, gr, gu);
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add projected point-point distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 2115]
2010-01-27 18:15:06 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::PT_FACE_DISTANCE:
case Type::PT_PLANE_DISTANCE: {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector pt = SK.GetEntity(ptA)->PointGetNum();
Entity *enta = SK.GetEntity(entityA);
Add point-face distance constraints. [git-p4: depot-paths = "//depot/solvespace/": change = 1778]
2008-06-06 08:46:55 +00:00
Vector n, p;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(type == Type::PT_PLANE_DISTANCE) {
Add point-face distance constraints. [git-p4: depot-paths = "//depot/solvespace/": change = 1778]
2008-06-06 08:46:55 +00:00
n = enta->Normal()->NormalN();
p = enta->WorkplaneGetOffset();
} else {
n = enta->FaceGetNormalNum();
p = enta->FaceGetPointNum();
}
Add point-line distance, point-plane distance, and point-on-circle constraints. And generate the constraint equations for entities (e.g., that our unit quaternions have magnitude one). Numerical troubles there, but it sort of works. Also some stuff to draw projection lines with projected constraints, and to auto-insert more constraints as you draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1711]
2008-05-08 07:30:30 +00:00
double d = (p.Minus(pt)).Dot(n);
Vector closest = pt.Plus(n.WithMagnitude(d));
Vector ref = ((closest.Plus(pt)).ScaledBy(0.5)).Plus(disp.offset);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(ref);
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
if(!pt.Equals(closest)) {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLineWithArrows(canvas, hcs, ref, pt, closest, /*onlyOneExt=*/true);
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLabel(canvas, hcs, ref, labelPos, gr, gu);
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add point-line distance, point-plane distance, and point-on-circle constraints. And generate the constraint equations for entities (e.g., that our unit quaternions have magnitude one). Numerical troubles there, but it sort of works. Also some stuff to draw projection lines with projected constraints, and to auto-insert more constraints as you draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1711]
2008-05-08 07:30:30 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::PT_LINE_DISTANCE: {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector pt = SK.GetEntity(ptA)->PointGetNum();
Entity *line = SK.GetEntity(entityA);
Vector lA = SK.GetEntity(line->point[0])->PointGetNum();
Vector lB = SK.GetEntity(line->point[1])->PointGetNum();
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
Vector dl = lB.Minus(lA);
Add point-line distance, point-plane distance, and point-on-circle constraints. And generate the constraint equations for entities (e.g., that our unit quaternions have magnitude one). Numerical troubles there, but it sort of works. Also some stuff to draw projection lines with projected constraints, and to auto-insert more constraints as you draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1711]
2008-05-08 07:30:30 +00:00
if(workplane.v != Entity::FREE_IN_3D.v) {
lA = lA.ProjectInto(workplane);
lB = lB.ProjectInto(workplane);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoProjectedPoint(canvas, hcs, &pt);
Add point-line distance, point-plane distance, and point-on-circle constraints. And generate the constraint equations for entities (e.g., that our unit quaternions have magnitude one). Numerical troubles there, but it sort of works. Also some stuff to draw projection lines with projected constraints, and to auto-insert more constraints as you draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1711]
2008-05-08 07:30:30 +00:00
}
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
// Find the closest point on the line
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
Vector closest = pt.ClosestPointOnLine(lA, dl);
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
Add point-line distance, point-plane distance, and point-on-circle constraints. And generate the constraint equations for entities (e.g., that our unit quaternions have magnitude one). Numerical troubles there, but it sort of works. Also some stuff to draw projection lines with projected constraints, and to auto-insert more constraints as you draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1711]
2008-05-08 07:30:30 +00:00
Vector ref = ((closest.Plus(pt)).ScaledBy(0.5)).Plus(disp.offset);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(ref);
DoLabel(canvas, hcs, ref, labelPos, gr, gu);
Draw an additional projection line, when showing a point-line projected distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 1712]
2008-05-08 08:12:23 +00:00
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
if(!pt.Equals(closest)) {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLineWithArrows(canvas, hcs, ref, pt, closest, /*onlyOneExt=*/true);
Eliminate DEFAULT_TEXT_HEIGHT from drawing code. Instead, query the text height for constraint style.
2016-03-02 17:15:28 +00:00
Draw extension lines from pt-line-distance to line ends. Before this commit, if a pt-line-distance constraint is placed so that the dimension line doesn't touch the line, no extension is drawn. After this commit, an extension line will be drawn towards the nearest end of the line.
2016-01-08 05:20:37 +00:00
// Extensions to line
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
double pixels = 1.0 / camera.scale;
Draw extension lines from pt-line-distance to line ends. Before this commit, if a pt-line-distance constraint is placed so that the dimension line doesn't touch the line, no extension is drawn. After this commit, an extension line will be drawn towards the nearest end of the line.
2016-01-08 05:20:37 +00:00
Vector refClosest = ref.ClosestPointOnLine(lA, dl);
double ddl = dl.Dot(dl);
if(fabs(ddl) > LENGTH_EPS * LENGTH_EPS) {
double t = refClosest.Minus(lA).Dot(dl) / ddl;
if(t < 0.0) {
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, refClosest.Minus(dl.WithMagnitude(10.0 * pixels)), lA);
Draw extension lines from pt-line-distance to line ends. Before this commit, if a pt-line-distance constraint is placed so that the dimension line doesn't touch the line, no extension is drawn. After this commit, an extension line will be drawn towards the nearest end of the line.
2016-01-08 05:20:37 +00:00
} else if(t > 1.0) {
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, refClosest.Plus(dl.WithMagnitude(10.0 * pixels)), lB);
Draw extension lines from pt-line-distance to line ends. Before this commit, if a pt-line-distance constraint is placed so that the dimension line doesn't touch the line, no extension is drawn. After this commit, an extension line will be drawn towards the nearest end of the line.
2016-01-08 05:20:37 +00:00
}
}
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
}
Draw an additional projection line, when showing a point-line projected distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 1712]
2008-05-08 08:12:23 +00:00
if(workplane.v != Entity::FREE_IN_3D.v) {
// Draw the projection marker from the closest point on the
// projected line to the projected point on the real line.
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
Vector lAB = (lA.Minus(lB));
Draw an additional projection line, when showing a point-line projected distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 1712]
2008-05-08 08:12:23 +00:00
double t = (lA.Minus(closest)).DivPivoting(lAB);
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector lA = SK.GetEntity(line->point[0])->PointGetNum();
Vector lB = SK.GetEntity(line->point[1])->PointGetNum();
Draw an additional projection line, when showing a point-line projected distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 1712]
2008-05-08 08:12:23 +00:00
Vector c2 = (lA.ScaledBy(1-t)).Plus(lB.ScaledBy(t));
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoProjectedPoint(canvas, hcs, &c2);
Draw an additional projection line, when showing a point-line projected distance constraint. [git-p4: depot-paths = "//depot/solvespace/": change = 1712]
2008-05-08 08:12:23 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add point-line distance, point-plane distance, and point-on-circle constraints. And generate the constraint equations for entities (e.g., that our unit quaternions have magnitude one). Numerical troubles there, but it sort of works. Also some stuff to draw projection lines with projected constraints, and to auto-insert more constraints as you draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1711]
2008-05-08 07:30:30 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::DIAMETER: {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *circle = SK.GetEntity(entityA);
Vector center = SK.GetEntity(circle->point[0])->PointGetNum();
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
Quaternion q = SK.GetEntity(circle->normal)->NormalGetNum();
Vector n = q.RotationN().WithMagnitude(1);
Add an arc entity. That's not allowed to exist in free space (since we need something to force the points into plane, and the workplane supplies that), but otherwise straightforward. And add diameter and equal radius constraints for the arc. [git-p4: depot-paths = "//depot/solvespace/": change = 1718]
2008-05-12 10:01:44 +00:00
double r = circle->CircleGetRadiusNum();
Add a diameter constraint, and add a `distance' entity that I can remap when I copy circle entities, in order to make the radius numerical somehow (analogy with the POINT_ and NORMAL_XFRMD) thing. [git-p4: depot-paths = "//depot/solvespace/": change = 1710]
2008-05-07 08:19:37 +00:00
Add ability to set export canvas size (paper size for PDF, bbox size for EPS, etc.). This can either be fixed, with a given width and height and offset, or automatic, by the left right bottom top margins. And draw nicer dimensions for length, with arrows and more extension lines. Add code to trim those lines against the (rectangular, axis-aligned) box that contains the actual number, and use that (instead of the elliptical interpolation, which was only approximately right) for diameter dimensions too. [git-p4: depot-paths = "//depot/solvespace/": change = 2027]
2009-09-03 08:13:09 +00:00
Vector ref = center.Plus(disp.offset);
// Force the label into the same plane as the circle.
ref = ref.Minus(n.ScaledBy(n.Dot(ref) - n.Dot(center)));
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(ref);
Add a diameter constraint, and add a `distance' entity that I can remap when I copy circle entities, in order to make the radius numerical somehow (analogy with the POINT_ and NORMAL_XFRMD) thing. [git-p4: depot-paths = "//depot/solvespace/": change = 1710]
2008-05-07 08:19:37 +00:00
Vector mark = ref.Minus(center);
mark = mark.WithMagnitude(mark.Magnitude()-r);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLineTrimmedAgainstBox(canvas, hcs, ref, ref, ref.Minus(mark));
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
Indicate diameter symbol on diameter dimensions, instead of just a bare number. [git-p4: depot-paths = "//depot/solvespace/": change = 2157]
2010-05-23 06:21:42 +00:00
Vector topLeft;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLabel(canvas, hcs, ref, &topLeft, gr, gu);
Indicate diameter symbol on diameter dimensions, instead of just a bare number. [git-p4: depot-paths = "//depot/solvespace/": change = 2157]
2010-05-23 06:21:42 +00:00
if(labelPos) *labelPos = topLeft;
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::POINTS_COINCIDENT: {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(how == DrawAs::DEFAULT) {
Make sure all DrawOrGetDistance style code only draws in Paint(); Otherwise, GL reports errors when used with FBOs, as well as creates visual artifacts, especially when UI is also drawn with OpenGL.
2015-03-26 00:34:28 +00:00
// Let's adjust the color of this constraint to have the same
// rough luma as the point color, so that the constraint does not
// stand out in an ugly way.
Rename RgbColor to RgbaColor.
2015-07-10 11:54:39 +00:00
RgbaColor cd = Style::Color(Style::DATUM),
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
cc = Style::Color(Style::CONSTRAINT);
Make sure all DrawOrGetDistance style code only draws in Paint(); Otherwise, GL reports errors when used with FBOs, as well as creates visual artifacts, especially when UI is also drawn with OpenGL.
2015-03-26 00:34:28 +00:00
// convert from 8-bit color to a vector
Vector vd = Vector::From(cd.redF(), cd.greenF(), cd.blueF()),
vc = Vector::From(cc.redF(), cc.greenF(), cc.blueF());
// and scale the constraint color to have the same magnitude as
// the datum color, maybe a bit dimmer
vc = vc.WithMagnitude(vd.Magnitude()*0.9);
// and set the color to that.
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
fill.color = RGBf(vc.x, vc.y, vc.z);
hcf = canvas->GetFill(fill);
}
Some graphics tweaks, to the order in which stuff gets drawn, to determine what goes in front (e.g. put a drawn line in front of the reference plane, even if the z order would want the opposite), and some tweaks to the mouse behaviour, and a function to modify constraints like dimensions so that the are initially satisfied. [git-p4: depot-paths = "//depot/solvespace/": change = 1681]
2008-04-22 10:53:42 +00:00
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
for(int a = 0; a < 2; a++) {
Vector r = camera.projRight.ScaledBy((a+1)/camera.scale);
Vector d = camera.projUp.ScaledBy((2-a)/camera.scale);
for(int i = 0; i < 2; i++) {
Vector p = SK.GetEntity(i == 0 ? ptA : ptB)-> PointGetNum();
if(refs) refs->push_back(p);
canvas->DrawQuad(p.Plus (r).Plus (d),
p.Plus (r).Minus(d),
p.Minus(r).Minus(d),
p.Minus(r).Plus (d),
hcf);
Make sure all DrawOrGetDistance style code only draws in Paint(); Otherwise, GL reports errors when used with FBOs, as well as creates visual artifacts, especially when UI is also drawn with OpenGL.
2015-03-26 00:34:28 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Start to add some constraint stuff. I now have point-coincident, and point-in-plane. These work, but the equation is still stupid, solving everything at once and not substituting trivial equations. [git-p4: depot-paths = "//depot/solvespace/": change = 1677]
2008-04-21 08:16:38 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Start to add some constraint stuff. I now have point-coincident, and point-in-plane. These work, but the equation is still stupid, solving everything at once and not substituting trivial equations. [git-p4: depot-paths = "//depot/solvespace/": change = 1677]
2008-04-21 08:16:38 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::PT_ON_CIRCLE:
case Type::PT_ON_LINE:
case Type::PT_ON_FACE:
case Type::PT_IN_PLANE: {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
double s = 8/camera.scale;
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector p = SK.GetEntity(ptA)->PointGetNum();
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(p);
Add selectable faces, by associating an hEntity with the triangle's metadata. And add point-on-face constraints to go with that. Still needs some cleanup for the user interface. [git-p4: depot-paths = "//depot/solvespace/": change = 1766]
2008-06-02 03:31:37 +00:00
Vector r, d;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(type == Type::PT_ON_FACE) {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector n = SK.GetEntity(entityA)->FaceGetNormalNum();
Add selectable faces, by associating an hEntity with the triangle's metadata. And add point-on-face constraints to go with that. Still needs some cleanup for the user interface. [git-p4: depot-paths = "//depot/solvespace/": change = 1766]
2008-06-02 03:31:37 +00:00
r = n.Normal(0);
d = n.Normal(1);
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
} else if(type == Type::PT_IN_PLANE) {
The constraint solver now compiles as a library, and I have a little test app that links against it. I still need to polish a few things, but this is more or less as it should be. [git-p4: depot-paths = "//depot/solvespace/": change = 1944]
2009-04-20 07:30:09 +00:00
EntityBase *n = SK.GetEntity(entityA)->Normal();
Add selectable faces, by associating an hEntity with the triangle's metadata. And add point-on-face constraints to go with that. Still needs some cleanup for the user interface. [git-p4: depot-paths = "//depot/solvespace/": change = 1766]
2008-06-02 03:31:37 +00:00
r = n->NormalU();
d = n->NormalV();
} else {
r = gr;
d = gu;
s *= (6.0/8); // draw these a little smaller
}
r = r.WithMagnitude(s); d = d.WithMagnitude(s);
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, p.Plus (r).Plus (d), p.Plus (r).Minus(d));
DoLine(canvas, hcs, p.Plus (r).Minus(d), p.Minus(r).Minus(d));
DoLine(canvas, hcs, p.Minus(r).Minus(d), p.Minus(r).Plus (d));
DoLine(canvas, hcs, p.Minus(r).Plus (d), p.Plus (r).Plus (d));
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Start to add some constraint stuff. I now have point-coincident, and point-in-plane. These work, but the equation is still stupid, solving everything at once and not substituting trivial equations. [git-p4: depot-paths = "//depot/solvespace/": change = 1677]
2008-04-21 08:16:38 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::WHERE_DRAGGED: {
Eliminate Constraint::dogd.refp. While we're at it, let's also emphasize both parts of a two-part constraint to make it easier to find.
2016-05-24 06:40:41 +00:00
Vector p = SK.GetEntity(ptA)->PointGetNum();
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(p);
Vector u = p.Plus(gu.WithMagnitude(8/camera.scale)).Plus(
gr.WithMagnitude(8/camera.scale)),
uu = u.Minus(gu.WithMagnitude(5/camera.scale)),
ur = u.Minus(gr.WithMagnitude(5/camera.scale));
Add "lock point where dragged" constraint, in either 2d or 3d. And call the cubic things splines, not segments, since that's what they are now. [git-p4: depot-paths = "//depot/solvespace/": change = 2135]
2010-05-04 05:11:52 +00:00
// Draw four little crop marks, uniformly spaced (by ninety
// degree rotations) around the point.
int i;
for(i = 0; i < 4; i++) {
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, u, uu);
DoLine(canvas, hcs, u, ur);
Add "lock point where dragged" constraint, in either 2d or 3d. And call the cubic things splines, not segments, since that's what they are now. [git-p4: depot-paths = "//depot/solvespace/": change = 2135]
2010-05-04 05:11:52 +00:00
u = u.RotatedAbout(p, gn, PI/2);
ur = ur.RotatedAbout(p, gn, PI/2);
uu = uu.RotatedAbout(p, gn, PI/2);
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add "lock point where dragged" constraint, in either 2d or 3d. And call the cubic things splines, not segments, since that's what they are now. [git-p4: depot-paths = "//depot/solvespace/": change = 2135]
2010-05-04 05:11:52 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::SAME_ORIENTATION: {
Add parallel and same orientation constraints, that work on line segments and normals. [git-p4: depot-paths = "//depot/solvespace/": change = 1713]
2008-05-09 05:33:23 +00:00
for(int i = 0; i < 2; i++) {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *e = SK.GetEntity(i == 0 ? entityA : entityB);
Add parallel and same orientation constraints, that work on line segments and normals. [git-p4: depot-paths = "//depot/solvespace/": change = 1713]
2008-05-09 05:33:23 +00:00
Quaternion q = e->NormalGetNum();
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Vector n = q.RotationN().WithMagnitude(25/camera.scale);
Vector u = q.RotationU().WithMagnitude(6/camera.scale);
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector p = SK.GetEntity(e->point[0])->PointGetNum();
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
p = p.Plus(n.WithMagnitude(10/camera.scale));
if(refs) refs->push_back(p);
Add parallel and same orientation constraints, that work on line segments and normals. [git-p4: depot-paths = "//depot/solvespace/": change = 1713]
2008-05-09 05:33:23 +00:00
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, p.Plus(u), p.Minus(u).Plus(n));
DoLine(canvas, hcs, p.Minus(u), p.Plus(u).Plus(n));
Add parallel and same orientation constraints, that work on line segments and normals. [git-p4: depot-paths = "//depot/solvespace/": change = 1713]
2008-05-09 05:33:23 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add parallel and same orientation constraints, that work on line segments and normals. [git-p4: depot-paths = "//depot/solvespace/": change = 1713]
2008-05-09 05:33:23 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::EQUAL_ANGLE: {
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
Vector ref;
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *a = SK.GetEntity(entityA);
Entity *b = SK.GetEntity(entityB);
Entity *c = SK.GetEntity(entityC);
Entity *d = SK.GetEntity(entityD);
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
Vector a0 = a->VectorGetStartPoint();
Vector b0 = b->VectorGetStartPoint();
Vector c0 = c->VectorGetStartPoint();
Vector d0 = d->VectorGetStartPoint();
Add angle constraints. I'm doing these differently from SketchFlat, as a constraint on the direction cosine, rather than driving the dot product against a rotated vector to zero. The drawing is the ugly part; to do that for skew lines, I gave up. Also add a function to clear non-existent items on the selection after solving, since that could have caused an oops(). [git-p4: depot-paths = "//depot/solvespace/": change = 1727]
2008-05-17 11:15:14 +00:00
Vector da = a->VectorGetNum();
Vector db = b->VectorGetNum();
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
Vector dc = c->VectorGetNum();
Vector dd = d->VectorGetNum();
Add angle constraints. I'm doing these differently from SketchFlat, as a constraint on the direction cosine, rather than driving the dot product against a rotated vector to zero. The drawing is the ugly part; to do that for skew lines, I gave up. Also add a function to clear non-existent items on the selection after solving, since that could have caused an oops(). [git-p4: depot-paths = "//depot/solvespace/": change = 1727]
2008-05-17 11:15:14 +00:00
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
if(other) {
a0 = a0.Plus(da);
da = da.ScaledBy(-1);
}
Add angle constraints. I'm doing these differently from SketchFlat, as a constraint on the direction cosine, rather than driving the dot product against a rotated vector to zero. The drawing is the ugly part; to do that for skew lines, I gave up. Also add a function to clear non-existent items on the selection after solving, since that could have caused an oops(). [git-p4: depot-paths = "//depot/solvespace/": change = 1727]
2008-05-17 11:15:14 +00:00
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoArcForAngle(canvas, hcs, a0, da, b0, db,
da.WithMagnitude(40/camera.scale), &ref, /*trim=*/false);
if(refs) refs->push_back(ref);
DoArcForAngle(canvas, hcs, c0, dc, d0, dd,
dc.WithMagnitude(40/camera.scale), &ref, /*trim=*/false);
if(refs) refs->push_back(ref);
Add angle constraints. I'm doing these differently from SketchFlat, as a constraint on the direction cosine, rather than driving the dot product against a rotated vector to zero. The drawing is the ugly part; to do that for skew lines, I gave up. Also add a function to clear non-existent items on the selection after solving, since that could have caused an oops(). [git-p4: depot-paths = "//depot/solvespace/": change = 1727]
2008-05-17 11:15:14 +00:00
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
}
Add angle constraints. I'm doing these differently from SketchFlat, as a constraint on the direction cosine, rather than driving the dot product against a rotated vector to zero. The drawing is the ugly part; to do that for skew lines, I gave up. Also add a function to clear non-existent items on the selection after solving, since that could have caused an oops(). [git-p4: depot-paths = "//depot/solvespace/": change = 1727]
2008-05-17 11:15:14 +00:00
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::ANGLE: {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *a = SK.GetEntity(entityA);
Entity *b = SK.GetEntity(entityB);
Trim trailing whitespace.
2015-03-29 00:30:52 +00:00
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
Vector a0 = a->VectorGetStartPoint();
Vector b0 = b->VectorGetStartPoint();
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
Vector da = a->VectorGetNum();
Vector db = b->VectorGetNum();
Draw extension lines for angle constraints between line segments.
2016-04-11 14:21:41 +00:00
if(other) {
a0 = a0.Plus(da);
da = da.ScaledBy(-1);
}
Add equal-angle constraints. These are implemented by brute force, as the difference between the cosines of the two angles. All of the angle stuff generates huge expressions (Expr *), but doesn't seem noticeably slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1847]
2008-07-20 12:24:43 +00:00
Vector ref;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoArcForAngle(canvas, hcs, a0, da, b0, db, disp.offset, &ref, /*trim=*/true);
DoLabel(canvas, hcs, ref, labelPos, gr, gu);
if(refs) refs->push_back(ref);
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add a perpendicular constraint; identical to constraining a ninety degree angle, but shows with a right angle symbol instead of a numerical angle you can edit. [git-p4: depot-paths = "//depot/solvespace/": change = 1819]
2008-07-02 09:21:29 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::PERPENDICULAR: {
Quash "variable may be used uninitialized" warnings Whether or not there is any actual danger of these variables being used without initialization, the warnings are noise, and getting rid of them is trivial.
2013-08-26 19:36:00 +00:00
Vector u = Vector::From(0, 0, 0), v = Vector::From(0, 0, 0);
Add a perpendicular constraint; identical to constraining a ninety degree angle, but shows with a right angle symbol instead of a numerical angle you can edit. [git-p4: depot-paths = "//depot/solvespace/": change = 1819]
2008-07-02 09:21:29 +00:00
Vector rn, ru;
if(workplane.v == Entity::FREE_IN_3D.v) {
rn = gn;
ru = gu;
} else {
The constraint solver now compiles as a library, and I have a little test app that links against it. I still need to polish a few things, but this is more or less as it should be. [git-p4: depot-paths = "//depot/solvespace/": change = 1944]
2009-04-20 07:30:09 +00:00
EntityBase *normal = SK.GetEntity(workplane)->Normal();
Add a perpendicular constraint; identical to constraining a ninety degree angle, but shows with a right angle symbol instead of a numerical angle you can edit. [git-p4: depot-paths = "//depot/solvespace/": change = 1819]
2008-07-02 09:21:29 +00:00
rn = normal->NormalN();
ru = normal->NormalV(); // ru meaning r_up, not u/v
}
for(int i = 0; i < 2; i++) {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *e = SK.GetEntity(i == 0 ? entityA : entityB);
Add a perpendicular constraint; identical to constraining a ninety degree angle, but shows with a right angle symbol instead of a numerical angle you can edit. [git-p4: depot-paths = "//depot/solvespace/": change = 1819]
2008-07-02 09:21:29 +00:00
if(i == 0) {
// Calculate orientation of perpendicular sign only
// once, so that it's the same both times it's drawn
u = e->VectorGetNum();
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
u = u.WithMagnitude(16/camera.scale);
v = (rn.Cross(u)).WithMagnitude(16/camera.scale);
Stop the perpendicular signs from flickering between the two drawn orientations when u and v are equally good choices. [git-p4: depot-paths = "//depot/solvespace/": change = 2080]
2009-11-29 02:42:50 +00:00
// a bit of bias to stop it from flickering between the
// two possibilities
if(fabs(u.Dot(ru)) < fabs(v.Dot(ru)) + LENGTH_EPS) {
Use C++ std::{min,max,swap} instead of custom ones. The main benefit is that std::swap will ensure that the type of arguments is copy-constructible and move-constructible. It is more concise as well. When min and max are defined as macros, they will conflict with STL header files included by other C++ libraries; in this case STL will #undef any other definition.
2015-03-27 15:43:28 +00:00
swap(u, v);
Add a perpendicular constraint; identical to constraining a ninety degree angle, but shows with a right angle symbol instead of a numerical angle you can edit. [git-p4: depot-paths = "//depot/solvespace/": change = 1819]
2008-07-02 09:21:29 +00:00
}
if(u.Dot(ru) < 0) u = u.ScaledBy(-1);
}
Vector p = e->VectorGetRefPoint();
Vector s = p.Plus(u).Plus(v);
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, s, s.Plus(v));
Add a perpendicular constraint; identical to constraining a ninety degree angle, but shows with a right angle symbol instead of a numerical angle you can edit. [git-p4: depot-paths = "//depot/solvespace/": change = 1819]
2008-07-02 09:21:29 +00:00
Vector m = s.Plus(v.ScaledBy(0.5));
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, m, m.Plus(u));
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(m);
Add a perpendicular constraint; identical to constraining a ninety degree angle, but shows with a right angle symbol instead of a numerical angle you can edit. [git-p4: depot-paths = "//depot/solvespace/": change = 1819]
2008-07-02 09:21:29 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add tangency constraints, for line segments against arcs or cubics. These are just a convenience, since it would be possible to get the same result by drawing a construction line. [git-p4: depot-paths = "//depot/solvespace/": change = 1836]
2008-07-13 12:44:05 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::CURVE_CURVE_TANGENT:
case Type::CUBIC_LINE_TANGENT:
case Type::ARC_LINE_TANGENT: {
Add tangency constraints, for line segments against arcs or cubics. These are just a convenience, since it would be possible to get the same result by drawing a construction line. [git-p4: depot-paths = "//depot/solvespace/": change = 1836]
2008-07-13 12:44:05 +00:00
Vector textAt, u, v;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(type == Type::ARC_LINE_TANGENT) {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *arc = SK.GetEntity(entityA);
Entity *norm = SK.GetEntity(arc->normal);
Vector c = SK.GetEntity(arc->point[0])->PointGetNum();
Trim trailing whitespace.
2015-03-29 00:30:52 +00:00
Vector p =
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
SK.GetEntity(arc->point[other ? 2 : 1])->PointGetNum();
Add tangency constraints, for line segments against arcs or cubics. These are just a convenience, since it would be possible to get the same result by drawing a construction line. [git-p4: depot-paths = "//depot/solvespace/": change = 1836]
2008-07-13 12:44:05 +00:00
Vector r = p.Minus(c);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
textAt = p.Plus(r.WithMagnitude(14/camera.scale));
Add tangency constraints, for line segments against arcs or cubics. These are just a convenience, since it would be possible to get the same result by drawing a construction line. [git-p4: depot-paths = "//depot/solvespace/": change = 1836]
2008-07-13 12:44:05 +00:00
u = norm->NormalU();
v = norm->NormalV();
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
} else if(type == Type::CUBIC_LINE_TANGENT) {
Add tangency constraints, for line segments against arcs or cubics. These are just a convenience, since it would be possible to get the same result by drawing a construction line. [git-p4: depot-paths = "//depot/solvespace/": change = 1836]
2008-07-13 12:44:05 +00:00
Vector n;
if(workplane.v == Entity::FREE_IN_3D.v) {
u = gr;
v = gu;
n = gn;
} else {
The constraint solver now compiles as a library, and I have a little test app that links against it. I still need to polish a few things, but this is more or less as it should be. [git-p4: depot-paths = "//depot/solvespace/": change = 1944]
2009-04-20 07:30:09 +00:00
EntityBase *wn = SK.GetEntity(workplane)->Normal();
Add tangency constraints, for line segments against arcs or cubics. These are just a convenience, since it would be possible to get the same result by drawing a construction line. [git-p4: depot-paths = "//depot/solvespace/": change = 1836]
2008-07-13 12:44:05 +00:00
u = wn->NormalU();
v = wn->NormalV();
n = wn->NormalN();
}
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *cubic = SK.GetEntity(entityA);
Add interpolating splines: both periodic splines (that form a loop), and open-ended splines, with their tangents specified at their endpoints. Also change constraint solver matrix size to 1024, on the theory that a power of two will generate better array indexing, and replace fabs() with my own function that for some reason is faster. [git-p4: depot-paths = "//depot/solvespace/": change = 2055]
2009-10-21 04:46:01 +00:00
Vector p = other ? cubic->CubicGetFinishNum() :
Trim trailing whitespace.
2015-03-29 00:30:52 +00:00
cubic->CubicGetStartNum();
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector dir = SK.GetEntity(entityB)->VectorGetNum();
Add tangency constraints, for line segments against arcs or cubics. These are just a convenience, since it would be possible to get the same result by drawing a construction line. [git-p4: depot-paths = "//depot/solvespace/": change = 1836]
2008-07-13 12:44:05 +00:00
Vector out = n.Cross(dir);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
textAt = p.Plus(out.WithMagnitude(14/camera.scale));
Add a constraint for tangency between any combination of arcs and cubics. Also add that to the library interface. It might have been better to use a single constraint for that, plus all the line-curve or line-line cases, but it would break backwards compatibility if I did that now, and perhaps be confusing with the 'other' member (which is meaningless for lines) anyways. [git-p4: depot-paths = "//depot/solvespace/": change = 2141]
2010-05-10 04:14:06 +00:00
} else {
Vector n, dir;
EntityBase *wn = SK.GetEntity(workplane)->Normal();
u = wn->NormalU();
v = wn->NormalV();
n = wn->NormalN();
EntityBase *eA = SK.GetEntity(entityA);
// Big pain; we have to get a vector tangent to the curve
// at the shared point, which could be from either a cubic
// or an arc.
if(other) {
textAt = eA->EndpointFinish();
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(eA->type == Entity::Type::CUBIC) {
Add a constraint for tangency between any combination of arcs and cubics. Also add that to the library interface. It might have been better to use a single constraint for that, plus all the line-curve or line-line cases, but it would break backwards compatibility if I did that now, and perhaps be confusing with the 'other' member (which is meaningless for lines) anyways. [git-p4: depot-paths = "//depot/solvespace/": change = 2141]
2010-05-10 04:14:06 +00:00
dir = eA->CubicGetFinishTangentNum();
} else {
dir = SK.GetEntity(eA->point[0])->PointGetNum().Minus(
SK.GetEntity(eA->point[2])->PointGetNum());
dir = n.Cross(dir);
}
} else {
textAt = eA->EndpointStart();
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(eA->type == Entity::Type::CUBIC) {
Add a constraint for tangency between any combination of arcs and cubics. Also add that to the library interface. It might have been better to use a single constraint for that, plus all the line-curve or line-line cases, but it would break backwards compatibility if I did that now, and perhaps be confusing with the 'other' member (which is meaningless for lines) anyways. [git-p4: depot-paths = "//depot/solvespace/": change = 2141]
2010-05-10 04:14:06 +00:00
dir = eA->CubicGetStartTangentNum();
} else {
dir = SK.GetEntity(eA->point[0])->PointGetNum().Minus(
SK.GetEntity(eA->point[1])->PointGetNum());
dir = n.Cross(dir);
}
}
dir = n.Cross(dir);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
textAt = textAt.Plus(dir.WithMagnitude(14/camera.scale));
Add tangency constraints, for line segments against arcs or cubics. These are just a convenience, since it would be possible to get the same result by drawing a construction line. [git-p4: depot-paths = "//depot/solvespace/": change = 1836]
2008-07-13 12:44:05 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Vector ex = VectorFont::Builtin()->GetExtents(textHeight, "T");
canvas->DrawVectorText("T", textHeight, textAt.Minus(ex.ScaledBy(0.5)), u, v, hcs);
if(refs) refs->push_back(textAt);
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add angle constraints. I'm doing these differently from SketchFlat, as a constraint on the direction cosine, rather than driving the dot product against a rotated vector to zero. The drawing is the ugly part; to do that for skew lines, I gave up. Also add a function to clear non-existent items on the selection after solving, since that could have caused an oops(). [git-p4: depot-paths = "//depot/solvespace/": change = 1727]
2008-05-17 11:15:14 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::PARALLEL: {
Add parallel and same orientation constraints, that work on line segments and normals. [git-p4: depot-paths = "//depot/solvespace/": change = 1713]
2008-05-09 05:33:23 +00:00
for(int i = 0; i < 2; i++) {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *e = SK.GetEntity(i == 0 ? entityA : entityB);
Find a memory corruption! I was getting a pointer into the entity list, and then adding a new entity to that list, and then looking at that pointer again. Not okay; the add operation might have forced a realloc. I have to watch for that. And add a "distance ratio" constraint, plus a new kind of group that comes with its own workplane. The workplane is not solved for; it's generated explicitly in terms of elements that are already solved. [git-p4: depot-paths = "//depot/solvespace/": change = 1716]
2008-05-11 10:40:37 +00:00
Vector n = e->VectorGetNum();
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
n = n.WithMagnitude(25/camera.scale);
Vector u = (gn.Cross(n)).WithMagnitude(4/camera.scale);
Add parallel and same orientation constraints, that work on line segments and normals. [git-p4: depot-paths = "//depot/solvespace/": change = 1713]
2008-05-09 05:33:23 +00:00
Vector p = e->VectorGetRefPoint();
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, p.Plus(u), p.Plus(u).Plus(n));
DoLine(canvas, hcs, p.Minus(u), p.Minus(u).Plus(n));
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(p.Plus(n.ScaledBy(0.5)));
Add parallel and same orientation constraints, that work on line segments and normals. [git-p4: depot-paths = "//depot/solvespace/": change = 1713]
2008-05-09 05:33:23 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add parallel and same orientation constraints, that work on line segments and normals. [git-p4: depot-paths = "//depot/solvespace/": change = 1713]
2008-05-09 05:33:23 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::EQUAL_RADIUS: {
Add an arc entity. That's not allowed to exist in free space (since we need something to force the points into plane, and the workplane supplies that), but otherwise straightforward. And add diameter and equal radius constraints for the arc. [git-p4: depot-paths = "//depot/solvespace/": change = 1718]
2008-05-12 10:01:44 +00:00
for(int i = 0; i < 2; i++) {
Eliminate Constraint::dogd.refp. While we're at it, let's also emphasize both parts of a two-part constraint to make it easier to find.
2016-05-24 06:40:41 +00:00
Vector ref;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoEqualRadiusTicks(canvas, hcs, i == 0 ? entityA : entityB, &ref);
if(refs) refs->push_back(ref);
Add an arc entity. That's not allowed to exist in free space (since we need something to force the points into plane, and the workplane supplies that), but otherwise straightforward. And add diameter and equal radius constraints for the arc. [git-p4: depot-paths = "//depot/solvespace/": change = 1718]
2008-05-12 10:01:44 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add an arc entity. That's not allowed to exist in free space (since we need something to force the points into plane, and the workplane supplies that), but otherwise straightforward. And add diameter and equal radius constraints for the arc. [git-p4: depot-paths = "//depot/solvespace/": change = 1718]
2008-05-12 10:01:44 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::EQUAL_LINE_ARC_LEN: {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *line = SK.GetEntity(entityA);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Vector ref;
DoEqualLenTicks(canvas, hcs,
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
SK.GetEntity(line->point[0])->PointGetNum(),
SK.GetEntity(line->point[1])->PointGetNum(),
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
gn, &ref);
if(refs) refs->push_back(ref);
DoEqualRadiusTicks(canvas, hcs, entityB, &ref);
if(refs) refs->push_back(ref);
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add a constraint to make line length equal arc length. That's quite tricky; can't just use the dot product, since that blows up when you cross pi radians. A gear shift approach, use either sin or cos, same kind of thing as the 3d-parallel constraint. And report a NaN constraint as unconverged, of course. [git-p4: depot-paths = "//depot/solvespace/": change = 1890]
2009-01-08 17:22:59 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::LENGTH_RATIO:
case Type::LENGTH_DIFFERENCE:
case Type::EQUAL_LENGTH_LINES: {
Quash "variable may be used uninitialized" warnings Whether or not there is any actual danger of these variables being used without initialization, the warnings are noise, and getting rid of them is trivial.
2013-08-26 19:36:00 +00:00
Vector a, b = Vector::From(0, 0, 0);
Add an equal length constraint for line segments. [git-p4: depot-paths = "//depot/solvespace/": change = 1680]
2008-04-22 05:00:49 +00:00
for(int i = 0; i < 2; i++) {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *e = SK.GetEntity(i == 0 ? entityA : entityB);
a = SK.GetEntity(e->point[0])->PointGetNum();
b = SK.GetEntity(e->point[1])->PointGetNum();
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
if(workplane.v != Entity::FREE_IN_3D.v) {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoProjectedPoint(canvas, hcs, &a);
DoProjectedPoint(canvas, hcs, &b);
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
}
Eliminate Constraint::dogd.refp. While we're at it, let's also emphasize both parts of a two-part constraint to make it easier to find.
2016-05-24 06:40:41 +00:00
Vector ref;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoEqualLenTicks(canvas, hcs, a, b, gn, &ref);
if(refs) refs->push_back(ref);
Add an equal length constraint for line segments. [git-p4: depot-paths = "//depot/solvespace/": change = 1680]
2008-04-22 05:00:49 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if((type == Type::LENGTH_RATIO) || (type == Type::LENGTH_DIFFERENCE)) {
Find a memory corruption! I was getting a pointer into the entity list, and then adding a new entity to that list, and then looking at that pointer again. Not okay; the add operation might have forced a realloc. I have to watch for that. And add a "distance ratio" constraint, plus a new kind of group that comes with its own workplane. The workplane is not solved for; it's generated explicitly in terms of elements that are already solved. [git-p4: depot-paths = "//depot/solvespace/": change = 1716]
2008-05-11 10:40:37 +00:00
Vector ref = ((a.Plus(b)).ScaledBy(0.5)).Plus(disp.offset);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLabel(canvas, hcs, ref, labelPos, gr, gu);
Find a memory corruption! I was getting a pointer into the entity list, and then adding a new entity to that list, and then looking at that pointer again. Not okay; the add operation might have forced a realloc. I have to watch for that. And add a "distance ratio" constraint, plus a new kind of group that comes with its own workplane. The workplane is not solved for; it's generated explicitly in terms of elements that are already solved. [git-p4: depot-paths = "//depot/solvespace/": change = 1716]
2008-05-11 10:40:37 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add an equal length constraint for line segments. [git-p4: depot-paths = "//depot/solvespace/": change = 1680]
2008-04-22 05:00:49 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::EQ_LEN_PT_LINE_D: {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *forLen = SK.GetEntity(entityA);
Vector a = SK.GetEntity(forLen->point[0])->PointGetNum(),
b = SK.GetEntity(forLen->point[1])->PointGetNum();
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
if(workplane.v != Entity::FREE_IN_3D.v) {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoProjectedPoint(canvas, hcs, &a);
DoProjectedPoint(canvas, hcs, &b);
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
}
Eliminate Constraint::dogd.refp. While we're at it, let's also emphasize both parts of a two-part constraint to make it easier to find.
2016-05-24 06:40:41 +00:00
Vector refa;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoEqualLenTicks(canvas, hcs, a, b, gn, &refa);
if(refs) refs->push_back(refa);
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *ln = SK.GetEntity(entityB);
Vector la = SK.GetEntity(ln->point[0])->PointGetNum(),
lb = SK.GetEntity(ln->point[1])->PointGetNum();
Vector pt = SK.GetEntity(ptA)->PointGetNum();
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
if(workplane.v != Entity::FREE_IN_3D.v) {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoProjectedPoint(canvas, hcs, &pt);
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
la = la.ProjectInto(workplane);
lb = lb.ProjectInto(workplane);
}
Vector closest = pt.ClosestPointOnLine(la, lb.Minus(la));
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, pt, closest);
Eliminate Constraint::dogd.refp. While we're at it, let's also emphasize both parts of a two-part constraint to make it easier to find.
2016-05-24 06:40:41 +00:00
Vector refb;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoEqualLenTicks(canvas, hcs, pt, closest, gn, &refb);
if(refs) refs->push_back(refb);
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::EQ_PT_LN_DISTANCES: {
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
for(int i = 0; i < 2; i++) {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *ln = SK.GetEntity(i == 0 ? entityA : entityB);
Vector la = SK.GetEntity(ln->point[0])->PointGetNum(),
lb = SK.GetEntity(ln->point[1])->PointGetNum();
Entity *pte = SK.GetEntity(i == 0 ? ptA : ptB);
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
Vector pt = pte->PointGetNum();
if(workplane.v != Entity::FREE_IN_3D.v) {
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoProjectedPoint(canvas, hcs, &pt);
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
la = la.ProjectInto(workplane);
lb = lb.ProjectInto(workplane);
}
Vector closest = pt.ClosestPointOnLine(la, lb.Minus(la));
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, pt, closest);
Eliminate Constraint::dogd.refp. While we're at it, let's also emphasize both parts of a two-part constraint to make it easier to find.
2016-05-24 06:40:41 +00:00
Vector ref;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoEqualLenTicks(canvas, hcs, pt, closest, gn, &ref);
if(refs) refs->push_back(ref);
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add two more exotic constraints: equal point-line distances, and point-line distance equal to line segment length. These are available in both normal and projected versions, with fancy display for all of these. [git-p4: depot-paths = "//depot/solvespace/": change = 1793]
2008-06-14 11:16:14 +00:00
}
Add symmetry constraints without an explicit symmetry plane; these use the horizontal or vertical axes of the active workplane. The equations that we write from these are very fast (of the form u + v = 0, or u - v = 0). [git-p4: depot-paths = "//depot/solvespace/": change = 1720]
2008-05-13 10:38:21 +00:00
{
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::SYMMETRIC:
Various fixes for warnings and minutia This commit consists of numerous small changes, none significant enough to merit a commit on their own: * Added extra braces to quash for-loop variable scoping issues for older compilers (or "g++ -fno-for-scope") * Appeased "unreachable code" warnings, spurious or otherwise * Added casts to fix integer-variable signedness warnings * Added a dummy virtual method to the VectorFileWriter class to silence the -Wweak-vtables warning from Clang++ * Renamed some parameters in the Expr and GraphicsWindow classes to eliminate "parameter shadows a field" warnings * Removed an inert "0 ||" from a conditional, and changed a "&& 0" into an "#if 0" * Added missing elements to array/struct/class initializers to zap further warnings * Indented some cpp conditionals where appropriate * Qualified some variables and functions as static to quiet "no previous declaration" warnings * toolbar.cpp needed to #include<icons-proto.h> to fix those same "no previous declaration" warnings from icons.h * Added some casts and const qualifiers to the Win32 code to address warnings produced by g++ when compiling under MinGW * Rewrote Cnf{Freeze,Thaw}Float() to use a union rather than pointer aliasing; this makes g++ a lot happier * Removed redundant #includes from win32/w32util.cpp * With Jonathan's blessing, shortened the last line of the About dialog text to better match the preceding lines
2013-10-19 05:36:45 +00:00
Vector n;
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
n = SK.GetEntity(entityA)->Normal()->NormalN(); goto s;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::SYMMETRIC_HORIZ:
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
n = SK.GetEntity(workplane)->Normal()->NormalU(); goto s;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::SYMMETRIC_VERT:
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
n = SK.GetEntity(workplane)->Normal()->NormalV(); goto s;
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::SYMMETRIC_LINE: {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *ln = SK.GetEntity(entityA);
Vector la = SK.GetEntity(ln->point[0])->PointGetNum(),
lb = SK.GetEntity(ln->point[1])->PointGetNum();
Add symmetric about line constraint. This applies only when locked in a workplane, since it would otherwise be ambiguous. [git-p4: depot-paths = "//depot/solvespace/": change = 1817]
2008-07-02 06:59:49 +00:00
la = la.ProjectInto(workplane);
lb = lb.ProjectInto(workplane);
n = lb.Minus(la);
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector nw = SK.GetEntity(workplane)->Normal()->NormalN();
Add symmetric about line constraint. This applies only when locked in a workplane, since it would otherwise be ambiguous. [git-p4: depot-paths = "//depot/solvespace/": change = 1817]
2008-07-02 06:59:49 +00:00
n = n.RotatedAbout(nw, PI/2);
goto s;
}
Add symmetry constraints without an explicit symmetry plane; these use the horizontal or vertical axes of the active workplane. The equations that we write from these are very fast (of the form u + v = 0, or u - v = 0). [git-p4: depot-paths = "//depot/solvespace/": change = 1720]
2008-05-13 10:38:21 +00:00
s:
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector a = SK.GetEntity(ptA)->PointGetNum();
Vector b = SK.GetEntity(ptB)->PointGetNum();
Add symmetric about line constraint. This applies only when locked in a workplane, since it would otherwise be ambiguous. [git-p4: depot-paths = "//depot/solvespace/": change = 1817]
2008-07-02 06:59:49 +00:00
Add symmetric and at midpoint constraints. The symmetry constraint turned out straightforward, in great part because the planes are workplanes (6 DOF, represented by a unit quaternion and a point), and therefore make it easy to get a vector in the plane, as well as a normal. And on that subject, replace the previous hack for parallel vector constraints with a better hack: pivot on the initial numerical guess, to choose which components of the cross product to drive to zero. Ugly, but I think that will be as robust as I can get. [git-p4: depot-paths = "//depot/solvespace/": change = 1699]
2008-04-30 08:14:32 +00:00
for(int i = 0; i < 2; i++) {
Vector tail = (i == 0) ? a : b;
Vector d = (i == 0) ? b : a;
d = d.Minus(tail);
// Project the direction in which the arrow is drawn normal
// to the symmetry plane; for projected symmetry constraints,
// they might not be in the same direction, even when the
// constraint is fully solved.
d = n.ScaledBy(d.Dot(n));
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
d = d.WithMagnitude(20/camera.scale);
Add symmetric and at midpoint constraints. The symmetry constraint turned out straightforward, in great part because the planes are workplanes (6 DOF, represented by a unit quaternion and a point), and therefore make it easy to get a vector in the plane, as well as a normal. And on that subject, replace the previous hack for parallel vector constraints with a better hack: pivot on the initial numerical guess, to choose which components of the cross product to drive to zero. Ugly, but I think that will be as robust as I can get. [git-p4: depot-paths = "//depot/solvespace/": change = 1699]
2008-04-30 08:14:32 +00:00
Vector tip = tail.Plus(d);
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, tail, tip);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
d = d.WithMagnitude(9/camera.scale);
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, tip, tip.Minus(d.RotatedAbout(gn, 0.6)));
DoLine(canvas, hcs, tip, tip.Minus(d.RotatedAbout(gn, -0.6)));
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(refs) refs->push_back(tip);
Add symmetric and at midpoint constraints. The symmetry constraint turned out straightforward, in great part because the planes are workplanes (6 DOF, represented by a unit quaternion and a point), and therefore make it easy to get a vector in the plane, as well as a normal. And on that subject, replace the previous hack for parallel vector constraints with a better hack: pivot on the initial numerical guess, to choose which components of the cross product to drive to zero. Ugly, but I think that will be as robust as I can get. [git-p4: depot-paths = "//depot/solvespace/": change = 1699]
2008-04-30 08:14:32 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add symmetric and at midpoint constraints. The symmetry constraint turned out straightforward, in great part because the planes are workplanes (6 DOF, represented by a unit quaternion and a point), and therefore make it easy to get a vector in the plane, as well as a normal. And on that subject, replace the previous hack for parallel vector constraints with a better hack: pivot on the initial numerical guess, to choose which components of the cross product to drive to zero. Ugly, but I think that will be as robust as I can get. [git-p4: depot-paths = "//depot/solvespace/": change = 1699]
2008-04-30 08:14:32 +00:00
}
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::AT_MIDPOINT:
case Type::HORIZONTAL:
case Type::VERTICAL:
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
if(entityA.v) {
Fix my rotation-matrix-to-quaternion code: it broke when the diagonal elements of the matrix summed to -1. Now it's ugly, but I think that it's correct. And add a command to flip the view to the other side, which is what started my problems. And tweak display of H and V and M for constraints: put them in the constraint plane, so that they're stationary as you rotate around. [git-p4: depot-paths = "//depot/solvespace/": change = 1704]
2008-05-03 03:33:35 +00:00
Vector r, u, n;
if(workplane.v == Entity::FREE_IN_3D.v) {
r = gr; u = gu; n = gn;
} else {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
r = SK.GetEntity(workplane)->Normal()->NormalU();
u = SK.GetEntity(workplane)->Normal()->NormalV();
Fix my rotation-matrix-to-quaternion code: it broke when the diagonal elements of the matrix summed to -1. Now it's ugly, but I think that it's correct. And add a command to flip the view to the other side, which is what started my problems. And tweak display of H and V and M for constraints: put them in the constraint plane, so that they're stationary as you rotate around. [git-p4: depot-paths = "//depot/solvespace/": change = 1704]
2008-05-03 03:33:35 +00:00
n = r.Cross(u);
}
Add symmetric and at midpoint constraints. The symmetry constraint turned out straightforward, in great part because the planes are workplanes (6 DOF, represented by a unit quaternion and a point), and therefore make it easy to get a vector in the plane, as well as a normal. And on that subject, replace the previous hack for parallel vector constraints with a better hack: pivot on the initial numerical guess, to choose which components of the cross product to drive to zero. Ugly, but I think that will be as robust as I can get. [git-p4: depot-paths = "//depot/solvespace/": change = 1699]
2008-04-30 08:14:32 +00:00
// For "at midpoint", this branch is always taken.
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *e = SK.GetEntity(entityA);
Vector a = SK.GetEntity(e->point[0])->PointGetNum();
Vector b = SK.GetEntity(e->point[1])->PointGetNum();
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
Vector m = (a.ScaledBy(0.5)).Plus(b.ScaledBy(0.5));
Fix my rotation-matrix-to-quaternion code: it broke when the diagonal elements of the matrix summed to -1. Now it's ugly, but I think that it's correct. And add a command to flip the view to the other side, which is what started my problems. And tweak display of H and V and M for constraints: put them in the constraint plane, so that they're stationary as you rotate around. [git-p4: depot-paths = "//depot/solvespace/": change = 1704]
2008-05-03 03:33:35 +00:00
Vector offset = (a.Minus(b)).Cross(n);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
offset = offset.WithMagnitude(textHeight);
Make at-midpoint constraint also work to constrain midpoint of a line segment on a plane. And simplify that code a bit, and improve display. [git-p4: depot-paths = "//depot/solvespace/": change = 1702]
2008-05-01 06:53:50 +00:00
// Draw midpoint constraint on other side of line, so that
// a line can be midpoint and horizontal at same time.
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(type == Type::AT_MIDPOINT) offset = offset.ScaledBy(-1);
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
std::string s;
switch(type) {
case Type::HORIZONTAL: s = "H"; break;
case Type::VERTICAL: s = "V"; break;
case Type::AT_MIDPOINT: s = "M"; break;
default: ssassert(false, "Unexpected constraint type");
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Vector o = m.Plus(offset).Plus(u.WithMagnitude(textHeight/5)),
ex = VectorFont::Builtin()->GetExtents(textHeight, s);
canvas->DrawVectorText(s, textHeight, o.Minus(ex.ScaledBy(0.5)), r, u, hcs);
if(refs) refs->push_back(o);
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
} else {
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Vector a = SK.GetEntity(ptA)->PointGetNum();
Vector b = SK.GetEntity(ptB)->PointGetNum();
Draw the horizontal/vertical markers in the direction of the actual basis vectors, not in the direction of the line connecting the points. [git-p4: depot-paths = "//depot/solvespace/": change = 1684]
2008-04-23 07:46:24 +00:00
A monster change; move the parameter, entity, group, and constraint tables from SolveSpace to their own class. This is intended to simplify use of the constraint solver in a library. [git-p4: depot-paths = "//depot/solvespace/": change = 1942]
2009-04-19 05:53:16 +00:00
Entity *w = SK.GetEntity(workplane);
Now we can add new workplanes. So fix all the mistakes that I hadn't previously noticed, because I didn't use to have workplanes with non-zero offsets. And clean up the interface to normals a bit. [git-p4: depot-paths = "//depot/solvespace/": change = 1707]
2008-05-05 11:17:00 +00:00
Vector cu = w->Normal()->NormalU();
Vector cv = w->Normal()->NormalV();
Vector cn = w->Normal()->NormalN();
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
int i;
for(i = 0; i < 2; i++) {
Vector o = (i == 0) ? a : b;
Draw the horizontal/vertical markers in the direction of the actual basis vectors, not in the direction of the line connecting the points. [git-p4: depot-paths = "//depot/solvespace/": change = 1684]
2008-04-23 07:46:24 +00:00
Vector oo = (i == 0) ? a.Minus(b) : b.Minus(a);
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
Vector d = (type == Type::HORIZONTAL) ? cu : cv;
Draw the horizontal/vertical markers in the direction of the actual basis vectors, not in the direction of the line connecting the points. [git-p4: depot-paths = "//depot/solvespace/": change = 1684]
2008-04-23 07:46:24 +00:00
if(oo.Dot(d) < 0) d = d.ScaledBy(-1);
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
Vector dp = cn.Cross(d);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
d = d.WithMagnitude(14/camera.scale);
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
Vector c = o.Minus(d);
Align constraint lines to pixel grid. This ensures that constraints always look crisp, no matter the camera position.
2016-08-02 11:48:16 +00:00
DoLine(canvas, hcs, o, c);
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
d = d.WithMagnitude(3/camera.scale);
dp = dp.WithMagnitude(2/camera.scale);
canvas->DrawQuad((c.Plus(d)).Plus(dp),
(c.Minus(d)).Plus(dp),
(c.Minus(d)).Minus(dp),
(c.Plus(d)).Minus(dp),
hcf);
if(refs) refs->push_back(c);
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
}
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add horizontal and vertical constraints. Those have their own 2d/3d issues, when the points are not all in the same coordinate system. All painful, of course. Also add continuous line drawing, and auto-constraining of line segments as I draw. [git-p4: depot-paths = "//depot/solvespace/": change = 1683]
2008-04-23 07:29:19 +00:00
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::COMMENT: {
Add text angle for styles. Add ability to quickly change between perspective and parallel projections. Add a snap grid, for points and for text comments. Draw text comments in the plane of their workplane if they have one, otherwise always facing forward. And fix a few nasty bugs: the possibility of an extremely long animation onto a workplane, accidental use of the wrong style line width for constraints, misplaced text box in style screen for default styles, other little stuff. [git-p4: depot-paths = "//depot/solvespace/": change = 2037]
2009-09-29 11:35:19 +00:00
Vector u, v;
if(workplane.v == Entity::FREE_IN_3D.v) {
u = gr;
v = gu;
} else {
EntityBase *norm = SK.GetEntity(workplane)->Normal();
u = norm->NormalU();
v = norm->NormalV();
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
if(disp.style.v != 0) {
stroke.width = Style::Width(disp.style);
if(how == DrawAs::DEFAULT) {
stroke.color = Style::Color(disp.style);
}
hcs = canvas->GetStroke(stroke);
}
DoLabel(canvas, hcs, disp.offset, labelPos, u, v);
if(refs) refs->push_back(disp.offset);
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
return;
Add text angle for styles. Add ability to quickly change between perspective and parallel projections. Add a snap grid, for points and for text comments. Draw text comments in the plane of their workplane if they have one, otherwise always facing forward. And fix a few nasty bugs: the possibility of an extremely long animation onto a workplane, accidental use of the wrong style line width for constraints, misplaced text box in style screen for default styles, other little stuff. [git-p4: depot-paths = "//depot/solvespace/": change = 2037]
2009-09-29 11:35:19 +00:00
}
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
}
Enable exhaustive switch coverage warnings as an error, and use them. Specifically, this enables -Wswitch=error on GCC/Clang and its MSVC equivalent; the exact way it is handled varies slightly, but what they all have in common is that in a switch statement over an enumeration, any enumerand that is not explicitly (via case:) or implicitly (via default:) handled in the switch triggers an error. Moreover, we also change the switch statements in three ways: * Switch statements that ought to be extended every time a new enumerand is added (e.g. Entity::DrawOrGetDistance(), are changed to explicitly list every single enumerand, and not have a default: branch. Note that the assertions are kept because it is legal for a enumeration to have a value unlike any of its defined enumerands, and we can e.g. read garbage from a file, or an uninitialized variable. This requires some rearranging if a default: branch is undesired. * Switch statements that ought to only ever see a few select enumerands, are changed to always assert in the default: branch. * Switch statements that do something meaningful for a few enumerands, and ignore everything else, are changed to do nothing in a default: branch, under the assumption that changing them every time an enumerand is added or removed would just result in noise and catch no bugs. This commit also removes the {Request,Entity,Constraint}::UNKNOWN and Entity::DATUM_POINT enumerands, as those were just fancy names for zeroes. They mess up switch exhaustiveness checks and most of the time were not the best way to implement what they did anyway.
2016-05-25 06:55:50 +00:00
ssassert(false, "Unexpected constraint type");
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::Draw(DrawAs how, Canvas *canvas) {
DoLayout(how, canvas, NULL, NULL);
Now I can add a constraint (a length), and it's displayed on-screen, and I can drag the label. That's progress. Also implement a bunch of untested expression stuff, since I'll need that for the values of the dimensions, for example. [git-p4: depot-paths = "//depot/solvespace/": change = 1668]
2008-04-14 10:28:32 +00:00
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
Vector Constraint::GetLabelPos(const Camera &camera) {
ObjectPicker canvas = {};
canvas.camera = camera;
Now I can display the edit control in the graphics window, and edit the label associated with a constraint. And that even works, changes the length of the line. [git-p4: depot-paths = "//depot/solvespace/": change = 1678]
2008-04-21 10:12:04 +00:00
Vector p;
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLayout(DrawAs::DEFAULT, &canvas, &p, NULL);
Now I can display the edit control in the graphics window, and edit the label associated with a constraint. And that even works, changes the length of the line. [git-p4: depot-paths = "//depot/solvespace/": change = 1678]
2008-04-21 10:12:04 +00:00
return p;
}
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
void Constraint::GetReferencePoints(const Camera &camera, std::vector<Vector> *refs) {
ObjectPicker canvas = {};
canvas.camera = camera;
Many user interface additions. Now I report when the solver fails, and in the case of a singular Jacobian, report which constraints can be removed to fix it. Also a mechanism to hover and select entities and constraints from the text window. [git-p4: depot-paths = "//depot/solvespace/": change = 1746]
2008-05-26 09:56:50 +00:00
Abstract all (ex-OpenGL) drawing operations into a Canvas interface. This has several desirable consequences: * It is now possible to port SolveSpace to a later version of OpenGL, such as OpenGLES 2, so that it runs on platforms that only have that OpenGL version; * The majority of geometry is now rendered without references to the camera in C++ code, so a renderer can now submit it to the video card once and re-rasterize with a different projection matrix every time the projection is changed, avoiding expensive reuploads; * The DOGD (draw or get distance) interface is now a straightforward Canvas implementation; * There are no more direct references to SS.GW.(projection) in sketch rendering code, which allows rendering to multiple viewports; * There are no more unnecessary framebuffer flips on CPU on Cocoa and GTK; * The platform-dependent GL code is now confined to rendergl1.cpp. * The Microsoft and Apple headers required by it that are prone to identifier conflicts are no longer included globally; * The rendergl1.cpp implementation can now be omitted from compilation to run SolveSpace headless or with a different OpenGL version. Note these implementation details of Canvas: * GetCamera currently always returns a reference to the field `Camera camera;`. This is so that a future renderer that caches geometry in the video memory can define it as asserting, which would provide assurance against code that could accidentally put something projection-dependent in the cache; * Line and triangle rendering is specified through a level of indirection, hStroke and hFill. This is so that a future renderer that batches geometry could cheaply group identical styles. * DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix. This is so that a future renderer into an output format that uses 2d transforms (e.g. SVG) could easily derive those. Some additional internal changes were required to enable this: * Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>. This is so that the renderer could cache uploaded textures between API calls, which requires it to capture a (weak) reference. * The PlatformPathEqual function was properly extracted into platform-specific code. This is so that the <windows.h> header could be included only where needed (in platform/w32* as well as rendergl1.cpp). * The SBsp{2,3}::DebugDraw functions were removed. They can be rewritten using the Canvas API if they are ever needed. While no visual changes were originally intended, some minor fixes happened anyway: * The "emphasis" yellow line from top-left corner is now correctly rendered much wider. * The marquee rectangle is now pixel grid aligned. * The hidden entities now do not clobber the depth buffer, removing some minor artifacts. * The workplane "tab" now scales with the font used to render the workplane name. * The workplane name font is now taken from the normals style. * Workplane and constraint line stipple is insignificantly different. This is so that it can reuse the existing stipple codepaths; rendering of workplanes and constraints predates those. Some debug functionality was added: * In graphics window, an fps counter that becomes red when rendering under 60fps is drawn.
2016-05-31 00:55:13 +00:00
DoLayout(DrawAs::DEFAULT, &canvas, NULL, refs);
Add a mechanism to record the lines drawn when we display a constraint, so that we can export that too. This includes the lines for the vector font. A little ugly; it needs some kind of line style (color or width) to distinguish those lines from the real geometry in the exported file. [git-p4: depot-paths = "//depot/solvespace/": change = 2007]
2009-07-03 20:55:57 +00:00
}
Add const qualifiers to functions where trivially possible. This will allow us in future to accept `const T &` anywhere it's necessary to reduce the amount of copying. This commit is quite conservative: it does not attempt very hard to refactor code that performs incidental mutation. In particular dogd and caches are not marked with the `mutable` keyword. dogd will be eliminated later, opening up more opportunities to add const qualifiers. This commit also doesn't introduce any uses of the newly added const qualifers. This will be done later.
2016-05-21 05:18:00 +00:00
bool Constraint::IsStylable() const {
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
if(type == Type::COMMENT) return true;
Make normals and workplanes non-stylable. This does not seem to have any useful application, and it can result in odd misrenderings with some styles.
2016-03-25 08:45:49 +00:00
return false;
}
DXF: export color as indexed, not RGB. The version of AutoCAD we advertise is unable to read RGB color, so software attempting to be compatible with it ignores color attributes.
2016-05-07 10:54:44 +00:00
hStyle Constraint::GetStyle() const {
if(disp.style.v != 0) return disp.style;
return { Style::CONSTRAINT };
}
Add const qualifiers to functions where trivially possible. This will allow us in future to accept `const T &` anywhere it's necessary to reduce the amount of copying. This commit is quite conservative: it does not attempt very hard to refactor code that performs incidental mutation. In particular dogd and caches are not marked with the `mutable` keyword. dogd will be eliminated later, opening up more opportunities to add const qualifiers. This commit also doesn't introduce any uses of the newly added const qualifers. This will be done later.
2016-05-21 05:18:00 +00:00
bool Constraint::HasLabel() const {
Allow snapping constraint labels to grid. This is pretty much the only way to get a sketch with tidily placed dimensions.
2016-04-18 06:40:42 +00:00
switch(type) {
Convert all enumerations to use `enum class`. Specifically, take the old code that looks like this: class Foo { enum { X = 1, Y = 2 }; int kind; } ... foo.kind = Foo::X; ... and convert it to this: class Foo { enum class Kind : uint32_t { X = 1, Y = 2 }; Kind kind; } ... foo.kind = Foo::Kind::X; (In some cases the enumeration would not be in the class namespace, such as when it is generally useful.) The benefits are as follows: * The type of the field gives a clear indication of intent, both to humans and tools (such as binding generators). * The compiler is able to automatically warn when a switch is not exhaustive; but this is currently suppressed by the default: ssassert(false, ...) idiom. * Integers and plain enums are weakly type checked: they implicitly convert into each other. This can hide bugs where type conversion is performed but not intended. Enum classes are strongly type checked. * Plain enums pollute parent namespaces; enum classes do not. Almost every defined enum we have already has a kind of ad-hoc namespacing via `NAMESPACE_`, which is now explicit. * Plain enums do not have a well-defined ABI size, which is important for bindings. Enum classes can have it, if specified. We specify the base type for all enums as uint32_t, which is a safe choice and allows us to not change the numeric values of any variants. This commit introduces absolutely no functional change to the code, just renaming and change of types. It handles almost all cases, except GraphicsWindow::pending.operation, which needs minor functional change.
2016-05-20 08:31:20 +00:00
case Type::COMMENT:
case Type::PT_PT_DISTANCE:
case Type::PT_PLANE_DISTANCE:
case Type::PT_LINE_DISTANCE:
case Type::PT_FACE_DISTANCE:
case Type::PROJ_PT_DISTANCE:
case Type::LENGTH_RATIO:
case Type::LENGTH_DIFFERENCE:
case Type::DIAMETER:
case Type::ANGLE:
Allow snapping constraint labels to grid. This is pretty much the only way to get a sketch with tidily placed dimensions.
2016-04-18 06:40:42 +00:00
return true;
default:
return false;
}
}