caiyu/solvespace
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
8af3a933cf
solvespace/src/bsp.cpp

624 lines
19 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]
2013-07-29 06:08:34 +08:00
//-----------------------------------------------------------------------------
// Binary space partitioning tree, used to represent a volume in 3-space
// bounded by a triangle mesh. These are used to compute Boolean operations
// on meshes. These aren't used for anything relating to an SShell of
// ratpoly surfaces.
//
// Copyright 2008-2013 Jonathan Westhues.
//-----------------------------------------------------------------------------
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
#include "solvespace.h"
Rewrite declarations of form f(void) as f(). In C++ there is no difference and newly added functions are all declared as f(), so this brings back consistency.
2016-05-05 13:54:05 +08:00
SBsp2 *SBsp2::Alloc() { return (SBsp2 *)AllocTemporary(sizeof(SBsp2)); }
SBsp3 *SBsp3::Alloc() { return (SBsp3 *)AllocTemporary(sizeof(SBsp3)); }
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
SBsp3 *SBsp3::FromMesh(SMesh *m) {
SBsp3 *bsp3 = NULL;
int i;
Replace all ZERO and memset with C++11 brace-initialization. This will allow us to use non-POD classes inside these objects in future and is otherwise functionally equivalent, as well as more concise. Note that there are some subtleties with handling of brace-initialization. Specifically: On aggregates (e.g. simple C-style structures) using an empty brace-initializer zero-initializes the aggregate, i.e. it makes all members zero. On non-aggregates an empty brace-initializer calls the default constructor. And if the constructor doesn't explicitly initialize the members (which the auto-generated constructor doesn't) then the members will be constructed but otherwise uninitialized. So, what is an aggregate class? To quote the C++ standard (C++03 8.5.1 §1): An aggregate is an array or a class (clause 9) with no user-declared constructors (12.1), no private or protected non-static data members (clause 11), no base classes (clause 10), and no virtual functions (10.3). In SolveSpace, we only have to handle the case of base classes; Constraint and Entity have those. Thus, they had to gain a default constructor that does nothing but initializes the members to zero.
2015-03-27 23:31:23 +08:00
SMesh mc = {};
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
for(i = 0; i < m->l.n; i++) {
mc.AddTriangle(&(m->l.elem[i]));
}
Go back to randomizing the faces before adding to the BSP, rather than sorting by area. I had hoped that would help with normal accuracy, but I don't think it helped much, and it sometimes became very slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1850]
2008-02-07 17:57:59 +08:00
srand(0); // Let's be deterministic, at least!
int n = mc.l.n;
while(n > 1) {
int k = rand() % n;
n--;
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 23:43:28 +08:00
swap(mc.l.elem[k], mc.l.elem[n]);
Go back to randomizing the faces before adding to the BSP, rather than sorting by area. I had hoped that would help with normal accuracy, but I don't think it helped much, and it sometimes became very slow. [git-p4: depot-paths = "//depot/solvespace/": change = 1850]
2008-02-07 17:57:59 +08:00
}
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
for(i = 0; i < mc.l.n; i++) {
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
bsp3 = InsertOrCreate(bsp3, &(mc.l.elem[i]), NULL);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
mc.Clear();
return bsp3;
}
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 13:18:00 +08:00
Vector SBsp3::IntersectionWith(Vector a, Vector b) const {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
double da = a.Dot(n) - d;
double db = b.Dot(n) - d;
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.
2016-05-19 06:51:36 +08:00
ssassert(da*db < 0, "Expected segment to intersect BSP node");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
double dab = (db - da);
return (a.ScaledBy(db/dab)).Plus(b.ScaledBy(-da/dab));
}
void SBsp3::InsertInPlane(bool pos2, STriangle *tr, SMesh *m) {
Vector tc = ((tr->a).Plus(tr->b).Plus(tr->c)).ScaledBy(1.0/3);
bool onFace = false;
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.
2013-08-27 03:36:00 +08:00
bool sameNormal = false;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
double maxNormalMag = -1;
Vector lln, trn = tr->Normal();
SBsp3 *ll = this;
while(ll) {
if((ll->tri).ContainsPoint(tc)) {
onFace = true;
// If the mesh contains almost-zero-area triangles, and we're
// just on the edge of one of those, then don't trust its normal.
lln = (ll->tri).Normal();
if(lln.Magnitude() > maxNormalMag) {
sameNormal = trn.Dot(lln) > 0;
maxNormalMag = lln.Magnitude();
}
}
ll = ll->more;
}
if(m->flipNormal && ((!pos2 && !onFace) ||
(onFace && !sameNormal && m->keepCoplanar)))
{
m->AddTriangle(tr->meta, tr->c, tr->b, tr->a);
Trim trailing whitespace.
2015-03-29 08:30:52 +08:00
} else if(!(m->flipNormal) && ((pos2 && !onFace) ||
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
(onFace && sameNormal && m->keepCoplanar)))
{
m->AddTriangle(tr->meta, tr->a, tr->b, tr->c);
} else {
m->atLeastOneDiscarded = true;
}
}
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 16:31:20 +08:00
void SBsp3::InsertHow(BspClass how, STriangle *tr, SMesh *instead) {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
switch(how) {
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 16:31:20 +08:00
case BspClass::POS:
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(instead && !pos) goto alt;
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
pos = InsertOrCreate(pos, tr, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
break;
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 16:31:20 +08:00
case BspClass::NEG:
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(instead && !neg) goto alt;
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
neg = InsertOrCreate(neg, tr, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
break;
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 16:31:20 +08:00
case BspClass::COPLANAR: {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(instead) goto alt;
SBsp3 *m = Alloc();
m->n = n;
m->d = d;
m->tri = *tr;
m->more = more;
more = m;
break;
}
}
return;
alt:
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 16:31:20 +08:00
if(how == BspClass::POS && !(instead->flipNormal)) {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
instead->AddTriangle(tr->meta, tr->a, tr->b, tr->c);
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 16:31:20 +08:00
} else if(how == BspClass::NEG && instead->flipNormal) {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
instead->AddTriangle(tr->meta, tr->c, tr->b, tr->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 16:31:20 +08:00
} else if(how == BspClass::COPLANAR) {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(edges) {
edges->InsertTriangle(tr, instead, this);
} else {
// I suppose this actually is allowed to happen, if the coplanar
// face is the leaf, and all of its neighbors are earlier in tree?
Annotate constants passed as boolean function arguments. This is to ensure that: * it is clear, when looking at the point of usage, what is the purpose of "true" or "false"; * when refactoring, a simple search will bring up any places that need to be changed. Also, argument names were synchronized between declaration and implementation. As an exception, these are not annotated: * Printf(/*halfLine=*/), to avoid pointless churn.
2016-05-25 20:08:19 +08:00
InsertInPlane(/*pos2=*/false, tr, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
} else {
instead->atLeastOneDiscarded = true;
}
}
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 16:31:20 +08:00
void SBsp3::InsertConvexHow(BspClass how, STriMeta meta, Vector *vertex, int n,
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
SMesh *instead)
{
switch(how) {
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 16:31:20 +08:00
case BspClass::POS:
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(pos) {
pos = pos->InsertConvex(meta, vertex, n, instead);
return;
}
break;
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 16:31:20 +08:00
case BspClass::NEG:
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(neg) {
neg = neg->InsertConvex(meta, vertex, n, instead);
return;
}
break;
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.
2016-05-19 06:51:36 +08:00
default: ssassert(false, "Unexpected BSP insert type");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
int i;
for(i = 0; i < n - 2; i++) {
STriangle tr = STriangle::From(meta,
vertex[0], vertex[i+1], vertex[i+2]);
InsertHow(how, &tr, instead);
}
}
SBsp3 *SBsp3::InsertConvex(STriMeta meta, Vector *vertex, int cnt,
SMesh *instead)
{
Vector e01 = (vertex[1]).Minus(vertex[0]);
Vector e12 = (vertex[2]).Minus(vertex[1]);
Vector out = e01.Cross(e12);
#define MAX_VERTICES 50
if(cnt+1 >= MAX_VERTICES) goto triangulate;
int i;
Vector on[2];
bool isPos[MAX_VERTICES];
bool isNeg[MAX_VERTICES];
bool isOn[MAX_VERTICES];
Fix "jump to label 'foo' crosses initialization of 'bar'" errors Newer C++ compilers do not allow goto-ing over an initialized variable declaration, as this violates C++98. We address this by (in most cases) separating the initialization from the declaration, or moving the goto to an equivalent location. Refer to the following discussions for more background: http://stackoverflow.com/questions/11306799/why-does-c-enforce-such-behavior-in-crosses-initialization http://stackoverflow.com/questions/12992108/crosses-initialization-of-variable-only-when-initialization-combined-with-decl
2013-08-27 03:08:16 +08:00
int posc, negc, onc; posc = negc = onc = 0;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
for(i = 0; i < cnt; i++) {
double dt = n.Dot(vertex[i]);
isPos[i] = isNeg[i] = isOn[i] = false;
if(fabs(dt - d) < LENGTH_EPS) {
isOn[i] = true;
if(onc < 2) {
on[onc] = vertex[i];
}
onc++;
} else if(dt > d) {
Trim trailing whitespace.
2015-03-29 08:30:52 +08:00
isPos[i] = true;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
posc++;
} else {
isNeg[i] = true;
negc++;
}
}
if(onc != 2 && onc != 1 && onc != 0) goto triangulate;
if(onc == 2) {
if(!instead) {
SEdge se = SEdge::From(on[0], on[1]);
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
edges = SBsp2::InsertOrCreateEdge(edges, &se, n, out);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
}
if(posc == 0) {
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 16:31:20 +08:00
InsertConvexHow(BspClass::NEG, meta, vertex, cnt, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
return this;
}
if(negc == 0) {
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 16:31:20 +08:00
InsertConvexHow(BspClass::POS, meta, vertex, cnt, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
return this;
}
Vector vpos[MAX_VERTICES];
Vector vneg[MAX_VERTICES];
Fix "jump to label 'foo' crosses initialization of 'bar'" errors Newer C++ compilers do not allow goto-ing over an initialized variable declaration, as this violates C++98. We address this by (in most cases) separating the initialization from the declaration, or moving the goto to an equivalent location. Refer to the following discussions for more background: http://stackoverflow.com/questions/11306799/why-does-c-enforce-such-behavior-in-crosses-initialization http://stackoverflow.com/questions/12992108/crosses-initialization-of-variable-only-when-initialization-combined-with-decl
2013-08-27 03:08:16 +08:00
int npos, nneg; npos = nneg = 0;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
Vector inter[2];
Fix "jump to label 'foo' crosses initialization of 'bar'" errors Newer C++ compilers do not allow goto-ing over an initialized variable declaration, as this violates C++98. We address this by (in most cases) separating the initialization from the declaration, or moving the goto to an equivalent location. Refer to the following discussions for more background: http://stackoverflow.com/questions/11306799/why-does-c-enforce-such-behavior-in-crosses-initialization http://stackoverflow.com/questions/12992108/crosses-initialization-of-variable-only-when-initialization-combined-with-decl
2013-08-27 03:08:16 +08:00
int inters; inters = 0;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
for(i = 0; i < cnt; i++) {
int ip = WRAP((i + 1), cnt);
if(isPos[i]) {
vpos[npos++] = vertex[i];
}
if(isNeg[i]) {
vneg[nneg++] = vertex[i];
}
if(isOn[i]) {
vneg[nneg++] = vertex[i];
vpos[npos++] = vertex[i];
}
if((isPos[i] && isNeg[ip]) || (isNeg[i] && isPos[ip])) {
Vector vi = IntersectionWith(vertex[i], vertex[ip]);
vpos[npos++] = vi;
vneg[nneg++] = vi;
Remove some of the oops() calls that tend to blow because of issues in the numerical code. And clean some other stuff, in particular simplifying whenever I have to take the maximum (or minimum) of three quantities. [git-p4: depot-paths = "//depot/solvespace/": change = 1855]
2008-02-10 21:34:32 +08:00
if(inters >= 2) goto triangulate; // XXX shouldn't happen but does
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
inter[inters++] = vi;
}
}
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.
2016-05-19 06:51:36 +08:00
if(npos > cnt + 1 || nneg > cnt + 1) ssassert(false, "Impossible");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(!instead) {
if(inters == 2) {
SEdge se = SEdge::From(inter[0], inter[1]);
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
edges = SBsp2::InsertOrCreateEdge(edges, &se, n, out);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else if(inters == 1 && onc == 1) {
SEdge se = SEdge::From(inter[0], on[0]);
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
edges = SBsp2::InsertOrCreateEdge(edges, &se, n, out);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else if(inters == 0 && onc == 2) {
// We already handled this on-plane existing edge
Remove some of the oops() calls that tend to blow because of issues in the numerical code. And clean some other stuff, in particular simplifying whenever I have to take the maximum (or minimum) of three quantities. [git-p4: depot-paths = "//depot/solvespace/": change = 1855]
2008-02-10 21:34:32 +08:00
} else {
goto triangulate;
}
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
Remove some of the oops() calls that tend to blow because of issues in the numerical code. And clean some other stuff, in particular simplifying whenever I have to take the maximum (or minimum) of three quantities. [git-p4: depot-paths = "//depot/solvespace/": change = 1855]
2008-02-10 21:34:32 +08:00
if(nneg < 3 || npos < 3) goto triangulate; // XXX
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
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 16:31:20 +08:00
InsertConvexHow(BspClass::NEG, meta, vneg, nneg, instead);
InsertConvexHow(BspClass::POS, meta, vpos, npos, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
return this;
triangulate:
// We don't handle the special case for this; do it as triangles
SBsp3 *r = this;
for(i = 0; i < cnt - 2; i++) {
STriangle tr = STriangle::From(meta,
vertex[0], vertex[i+1], vertex[i+2]);
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
r = InsertOrCreate(r, &tr, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
return r;
}
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
SBsp3 *SBsp3::InsertOrCreate(SBsp3 *where, STriangle *tr, SMesh *instead) {
if(where == NULL) {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(instead) {
if(instead->flipNormal) {
instead->atLeastOneDiscarded = true;
} else {
instead->AddTriangle(tr->meta, tr->a, tr->b, tr->c);
}
return NULL;
}
// Brand new node; so allocate for it, and fill us in.
SBsp3 *r = Alloc();
r->n = (tr->Normal()).WithMagnitude(1);
r->d = (tr->a).Dot(r->n);
r->tri = *tr;
return r;
}
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
where->Insert(tr, instead);
return where;
}
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
void SBsp3::Insert(STriangle *tr, SMesh *instead) {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
double dt[3] = { (tr->a).Dot(n), (tr->b).Dot(n), (tr->c).Dot(n) };
int inc = 0, posc = 0, negc = 0;
Replace all ZERO and memset with C++11 brace-initialization. This will allow us to use non-POD classes inside these objects in future and is otherwise functionally equivalent, as well as more concise. Note that there are some subtleties with handling of brace-initialization. Specifically: On aggregates (e.g. simple C-style structures) using an empty brace-initializer zero-initializes the aggregate, i.e. it makes all members zero. On non-aggregates an empty brace-initializer calls the default constructor. And if the constructor doesn't explicitly initialize the members (which the auto-generated constructor doesn't) then the members will be constructed but otherwise uninitialized. So, what is an aggregate class? To quote the C++ standard (C++03 8.5.1 §1): An aggregate is an array or a class (clause 9) with no user-declared constructors (12.1), no private or protected non-static data members (clause 11), no base classes (clause 10), and no virtual functions (10.3). In SolveSpace, we only have to handle the case of base classes; Constraint and Entity have those. Thus, they had to gain a default constructor that does nothing but initializes the members to zero.
2015-03-27 23:31:23 +08:00
bool isPos[3] = {}, isNeg[3] = {}, isOn[3] = {};
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
// Count vertices in the plane
for(int i = 0; i < 3; i++) {
if(fabs(dt[i] - d) < LENGTH_EPS) {
inc++;
isOn[i] = true;
} else if(dt[i] > d) {
posc++;
isPos[i] = true;
} else {
negc++;
isNeg[i] = true;
}
}
// All vertices in-plane
if(inc == 3) {
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 16:31:20 +08:00
InsertHow(BspClass::COPLANAR, tr, instead);
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
return;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
// No split required
if(posc == 0 || negc == 0) {
if(inc == 2) {
Vector a, b;
if (!isOn[0]) { a = tr->b; b = tr->c; }
else if(!isOn[1]) { a = tr->c; b = tr->a; }
else if(!isOn[2]) { a = tr->a; b = tr->b; }
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.
2016-05-19 06:51:36 +08:00
else ssassert(false, "Impossible");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(!instead) {
SEdge se = SEdge::From(a, b);
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
edges = SBsp2::InsertOrCreateEdge(edges, &se, n, tr->Normal());
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
}
if(posc > 0) {
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 16:31:20 +08:00
InsertHow(BspClass::POS, tr, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else {
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 16:31:20 +08:00
InsertHow(BspClass::NEG, tr, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
return;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
// The polygon must be split into two pieces, one above, one below.
Vector a, b, c;
if(posc == 1 && negc == 1 && inc == 1) {
bool bpos;
// Standardize so that a is on the plane
if (isOn[0]) { a = tr->a; b = tr->b; c = tr->c; bpos = isPos[1];
} else if(isOn[1]) { a = tr->b; b = tr->c; c = tr->a; bpos = isPos[2];
} else if(isOn[2]) { a = tr->c; b = tr->a; c = tr->b; bpos = isPos[0];
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.
2016-05-19 06:51:36 +08:00
} else ssassert(false, "Impossible");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
Vector bPc = IntersectionWith(b, c);
STriangle btri = STriangle::From(tr->meta, a, b, bPc);
STriangle ctri = STriangle::From(tr->meta, c, a, bPc);
if(bpos) {
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 16:31:20 +08:00
InsertHow(BspClass::POS, &btri, instead);
InsertHow(BspClass::NEG, &ctri, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else {
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 16:31:20 +08:00
InsertHow(BspClass::POS, &ctri, instead);
InsertHow(BspClass::NEG, &btri, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
if(!instead) {
SEdge se = SEdge::From(a, bPc);
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
edges = SBsp2::InsertOrCreateEdge(edges, &se, n, tr->Normal());
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
return;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
if(posc == 2 && negc == 1) {
// Standardize so that a is on one side, and b and c are on the other.
if (isNeg[0]) { a = tr->a; b = tr->b; c = tr->c;
} else if(isNeg[1]) { a = tr->b; b = tr->c; c = tr->a;
} else if(isNeg[2]) { a = tr->c; b = tr->a; c = tr->b;
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.
2016-05-19 06:51:36 +08:00
} else ssassert(false, "Impossible");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else if(posc == 1 && negc == 2) {
if (isPos[0]) { a = tr->a; b = tr->b; c = tr->c;
} else if(isPos[1]) { a = tr->b; b = tr->c; c = tr->a;
} else if(isPos[2]) { a = tr->c; b = tr->a; c = tr->b;
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.
2016-05-19 06:51:36 +08:00
} else ssassert(false, "Impossible");
} else ssassert(false, "Impossible");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
Vector aPb = IntersectionWith(a, b);
Vector cPa = IntersectionWith(c, a);
STriangle alone = STriangle::From(tr->meta, a, aPb, cPa);
Vector quad[4] = { aPb, b, c, cPa };
if(posc == 2 && negc == 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 16:31:20 +08:00
InsertConvexHow(BspClass::POS, tr->meta, quad, 4, instead);
InsertHow(BspClass::NEG, &alone, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else {
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 16:31:20 +08:00
InsertConvexHow(BspClass::NEG, tr->meta, quad, 4, instead);
InsertHow(BspClass::POS, &alone, instead);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
if(!instead) {
SEdge se = SEdge::From(aPb, cPa);
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
edges = SBsp2::InsertOrCreateEdge(edges, &se, n, alone.Normal());
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
return;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
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 13:18:00 +08:00
void SBsp3::GenerateInPaintOrder(SMesh *m) const {
Add hidden line and surface removal, and vector shaded surface export. So I calculate lighting for each triangle in the mesh, make a BSP, and then traverse it in-order and output those as SVG or EPS. And I test edges against the mesh, removing those portions of the edge that overlap a triangle in front of them (using the kd-tree to accelerate). [git-p4: depot-paths = "//depot/solvespace/": change = 1931]
2009-03-18 00:33:46 +08:00
// Doesn't matter which branch we take if the normal has zero z
// component, so don't need a separate case for that.
if(n.z < 0) {
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
if(pos) pos->GenerateInPaintOrder(m);
Add hidden line and surface removal, and vector shaded surface export. So I calculate lighting for each triangle in the mesh, make a BSP, and then traverse it in-order and output those as SVG or EPS. And I test edges against the mesh, removing those portions of the edge that overlap a triangle in front of them (using the kd-tree to accelerate). [git-p4: depot-paths = "//depot/solvespace/": change = 1931]
2009-03-18 00:33:46 +08:00
} else {
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
if(neg) neg->GenerateInPaintOrder(m);
Add hidden line and surface removal, and vector shaded surface export. So I calculate lighting for each triangle in the mesh, make a BSP, and then traverse it in-order and output those as SVG or EPS. And I test edges against the mesh, removing those portions of the edge that overlap a triangle in front of them (using the kd-tree to accelerate). [git-p4: depot-paths = "//depot/solvespace/": change = 1931]
2009-03-18 00:33:46 +08:00
}
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 13:18:00 +08:00
const SBsp3 *flip = this;
Add hidden line and surface removal, and vector shaded surface export. So I calculate lighting for each triangle in the mesh, make a BSP, and then traverse it in-order and output those as SVG or EPS. And I test edges against the mesh, removing those portions of the edge that overlap a triangle in front of them (using the kd-tree to accelerate). [git-p4: depot-paths = "//depot/solvespace/": change = 1931]
2009-03-18 00:33:46 +08:00
while(flip) {
m->AddTriangle(&(flip->tri));
flip = flip->more;
}
if(n.z < 0) {
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
if(neg) neg->GenerateInPaintOrder(m);
Add hidden line and surface removal, and vector shaded surface export. So I calculate lighting for each triangle in the mesh, make a BSP, and then traverse it in-order and output those as SVG or EPS. And I test edges against the mesh, removing those portions of the edge that overlap a triangle in front of them (using the kd-tree to accelerate). [git-p4: depot-paths = "//depot/solvespace/": change = 1931]
2009-03-18 00:33:46 +08:00
} else {
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
if(pos) pos->GenerateInPaintOrder(m);
Add hidden line and surface removal, and vector shaded surface export. So I calculate lighting for each triangle in the mesh, make a BSP, and then traverse it in-order and output those as SVG or EPS. And I test edges against the mesh, removing those portions of the edge that overlap a triangle in front of them (using the kd-tree to accelerate). [git-p4: depot-paths = "//depot/solvespace/": change = 1931]
2009-03-18 00:33:46 +08:00
}
}
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
/////////////////////////////////
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 13:18:00 +08:00
Vector SBsp2::IntersectionWith(Vector a, Vector b) const {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
double da = a.Dot(no) - d;
double db = b.Dot(no) - d;
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.
2016-05-19 06:51:36 +08:00
ssassert(da*db < 0, "Expected segment to intersect BSP node");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
double dab = (db - da);
return (a.ScaledBy(db/dab)).Plus(b.ScaledBy(-da/dab));
}
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
SBsp2 *SBsp2::InsertOrCreateEdge(SBsp2 *where, SEdge *nedge, Vector nnp, Vector out) {
if(where == NULL) {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
// Brand new node; so allocate for it, and fill us in.
SBsp2 *r = Alloc();
r->np = nnp;
r->no = ((r->np).Cross((nedge->b).Minus(nedge->a))).WithMagnitude(1);
if(out.Dot(r->no) < 0) {
r->no = (r->no).ScaledBy(-1);
}
r->d = (nedge->a).Dot(r->no);
r->edge = *nedge;
return r;
}
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
where->InsertEdge(nedge, nnp, out);
return where;
}
void SBsp2::InsertEdge(SEdge *nedge, Vector nnp, Vector out) {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
double dt[2] = { (nedge->a).Dot(no), (nedge->b).Dot(no) };
Replace all ZERO and memset with C++11 brace-initialization. This will allow us to use non-POD classes inside these objects in future and is otherwise functionally equivalent, as well as more concise. Note that there are some subtleties with handling of brace-initialization. Specifically: On aggregates (e.g. simple C-style structures) using an empty brace-initializer zero-initializes the aggregate, i.e. it makes all members zero. On non-aggregates an empty brace-initializer calls the default constructor. And if the constructor doesn't explicitly initialize the members (which the auto-generated constructor doesn't) then the members will be constructed but otherwise uninitialized. So, what is an aggregate class? To quote the C++ standard (C++03 8.5.1 §1): An aggregate is an array or a class (clause 9) with no user-declared constructors (12.1), no private or protected non-static data members (clause 11), no base classes (clause 10), and no virtual functions (10.3). In SolveSpace, we only have to handle the case of base classes; Constraint and Entity have those. Thus, they had to gain a default constructor that does nothing but initializes the members to zero.
2015-03-27 23:31:23 +08:00
bool isPos[2] = {}, isNeg[2] = {}, isOn[2] = {};
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
for(int i = 0; i < 2; i++) {
if(fabs(dt[i] - d) < LENGTH_EPS) {
isOn[i] = true;
} else if(dt[i] > d) {
isPos[i] = true;
} else {
isNeg[i] = true;
}
}
if((isPos[0] && isPos[1])||(isPos[0] && isOn[1])||(isOn[0] && isPos[1])) {
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
pos = InsertOrCreateEdge(pos, nedge, nnp, out);
return;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
if((isNeg[0] && isNeg[1])||(isNeg[0] && isOn[1])||(isOn[0] && isNeg[1])) {
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
neg = InsertOrCreateEdge(neg, nedge, nnp, out);
return;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
if(isOn[0] && isOn[1]) {
SBsp2 *m = Alloc();
m->np = nnp;
m->no = ((m->np).Cross((nedge->b).Minus(nedge->a))).WithMagnitude(1);
if(out.Dot(m->no) < 0) {
m->no = (m->no).ScaledBy(-1);
}
m->d = (nedge->a).Dot(m->no);
m->edge = *nedge;
m->more = more;
more = m;
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
return;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
if((isPos[0] && isNeg[1]) || (isNeg[0] && isPos[1])) {
Vector aPb = IntersectionWith(nedge->a, nedge->b);
SEdge ea = SEdge::From(nedge->a, aPb);
SEdge eb = SEdge::From(aPb, nedge->b);
if(isPos[0]) {
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
pos = InsertOrCreateEdge(pos, &ea, nnp, out);
neg = InsertOrCreateEdge(neg, &eb, nnp, out);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else {
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
neg = InsertOrCreateEdge(neg, &ea, nnp, out);
pos = InsertOrCreateEdge(pos, &eb, nnp, out);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
Get rid of if(!this). According to the C++ standard, "this" is never NULL, so checks of the form "if(!this)" can be legally optimized out. This breaks SolveSpace on GCC 6, and probably on other compilers and configurations.
2016-03-25 22:16:58 +08:00
return;
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
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.
2016-05-19 06:51:36 +08:00
ssassert(false, "Impossible");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
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 16:31:20 +08:00
void SBsp2::InsertTriangleHow(BspClass how, STriangle *tr, SMesh *m, SBsp3 *bsp3) {
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
switch(how) {
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 16:31:20 +08:00
case BspClass::POS:
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(pos) {
pos->InsertTriangle(tr, m, bsp3);
} else {
Annotate constants passed as boolean function arguments. This is to ensure that: * it is clear, when looking at the point of usage, what is the purpose of "true" or "false"; * when refactoring, a simple search will bring up any places that need to be changed. Also, argument names were synchronized between declaration and implementation. As an exception, these are not annotated: * Printf(/*halfLine=*/), to avoid pointless churn.
2016-05-25 20:08:19 +08:00
bsp3->InsertInPlane(/*pos2=*/true, tr, m);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
break;
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 16:31:20 +08:00
case BspClass::NEG:
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
if(neg) {
neg->InsertTriangle(tr, m, bsp3);
} else {
Annotate constants passed as boolean function arguments. This is to ensure that: * it is clear, when looking at the point of usage, what is the purpose of "true" or "false"; * when refactoring, a simple search will bring up any places that need to be changed. Also, argument names were synchronized between declaration and implementation. As an exception, these are not annotated: * Printf(/*halfLine=*/), to avoid pointless churn.
2016-05-25 20:08:19 +08:00
bsp3->InsertInPlane(/*pos2=*/false, tr, m);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
break;
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.
2016-05-19 06:51:36 +08:00
default: ssassert(false, "Unexpected BSP insert type");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
}
void SBsp2::InsertTriangle(STriangle *tr, SMesh *m, SBsp3 *bsp3) {
double dt[3] = { (tr->a).Dot(no), (tr->b).Dot(no), (tr->c).Dot(no) };
Replace all ZERO and memset with C++11 brace-initialization. This will allow us to use non-POD classes inside these objects in future and is otherwise functionally equivalent, as well as more concise. Note that there are some subtleties with handling of brace-initialization. Specifically: On aggregates (e.g. simple C-style structures) using an empty brace-initializer zero-initializes the aggregate, i.e. it makes all members zero. On non-aggregates an empty brace-initializer calls the default constructor. And if the constructor doesn't explicitly initialize the members (which the auto-generated constructor doesn't) then the members will be constructed but otherwise uninitialized. So, what is an aggregate class? To quote the C++ standard (C++03 8.5.1 §1): An aggregate is an array or a class (clause 9) with no user-declared constructors (12.1), no private or protected non-static data members (clause 11), no base classes (clause 10), and no virtual functions (10.3). In SolveSpace, we only have to handle the case of base classes; Constraint and Entity have those. Thus, they had to gain a default constructor that does nothing but initializes the members to zero.
2015-03-27 23:31:23 +08:00
bool isPos[3] = {}, isNeg[3] = {}, isOn[3] = {};
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
int inc = 0, posc = 0, negc = 0;
for(int i = 0; i < 3; i++) {
if(fabs(dt[i] - d) < LENGTH_EPS) {
isOn[i] = true;
inc++;
} else if(dt[i] > d) {
isPos[i] = true;
posc++;
} else {
isNeg[i] = true;
negc++;
}
}
if(inc == 3) {
// All vertices on-line; so it's a degenerate triangle, to ignore.
return;
}
// No split required
if(posc == 0 || negc == 0) {
if(posc > 0) {
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 16:31:20 +08:00
InsertTriangleHow(BspClass::POS, tr, m, bsp3);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else {
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 16:31:20 +08:00
InsertTriangleHow(BspClass::NEG, tr, m, bsp3);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
return;
}
// The polygon must be split into two pieces, one above, one below.
Vector a, b, c;
if(posc == 1 && negc == 1 && inc == 1) {
bool bpos;
// Standardize so that a is on the plane
if (isOn[0]) { a = tr->a; b = tr->b; c = tr->c; bpos = isPos[1];
} else if(isOn[1]) { a = tr->b; b = tr->c; c = tr->a; bpos = isPos[2];
} else if(isOn[2]) { a = tr->c; b = tr->a; c = tr->b; bpos = isPos[0];
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.
2016-05-19 06:51:36 +08:00
} else ssassert(false, "Impossible");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
Vector bPc = IntersectionWith(b, c);
STriangle btri = STriangle::From(tr->meta, a, b, bPc);
STriangle ctri = STriangle::From(tr->meta, c, a, bPc);
if(bpos) {
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 16:31:20 +08:00
InsertTriangleHow(BspClass::POS, &btri, m, bsp3);
InsertTriangleHow(BspClass::NEG, &ctri, m, bsp3);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else {
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 16:31:20 +08:00
InsertTriangleHow(BspClass::POS, &ctri, m, bsp3);
InsertTriangleHow(BspClass::NEG, &btri, m, bsp3);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
return;
}
if(posc == 2 && negc == 1) {
// Standardize so that a is on one side, and b and c are on the other.
if (isNeg[0]) { a = tr->a; b = tr->b; c = tr->c;
} else if(isNeg[1]) { a = tr->b; b = tr->c; c = tr->a;
} else if(isNeg[2]) { a = tr->c; b = tr->a; c = tr->b;
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.
2016-05-19 06:51:36 +08:00
} else ssassert(false, "Impossible");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else if(posc == 1 && negc == 2) {
if (isPos[0]) { a = tr->a; b = tr->b; c = tr->c;
} else if(isPos[1]) { a = tr->b; b = tr->c; c = tr->a;
} else if(isPos[2]) { a = tr->c; b = tr->a; c = tr->b;
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.
2016-05-19 06:51:36 +08:00
} else ssassert(false, "Impossible");
} else ssassert(false, "Impossible");
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
Vector aPb = IntersectionWith(a, b);
Vector cPa = IntersectionWith(c, a);
STriangle alone = STriangle::From(tr->meta, a, aPb, cPa);
STriangle quad1 = STriangle::From(tr->meta, aPb, b, c );
STriangle quad2 = STriangle::From(tr->meta, aPb, c, cPa);
if(posc == 2 && negc == 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 16:31:20 +08:00
InsertTriangleHow(BspClass::POS, &quad1, m, bsp3);
InsertTriangleHow(BspClass::POS, &quad2, m, bsp3);
InsertTriangleHow(BspClass::NEG, &alone, m, bsp3);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
} else {
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 16:31:20 +08:00
InsertTriangleHow(BspClass::NEG, &quad1, m, bsp3);
InsertTriangleHow(BspClass::NEG, &quad2, m, bsp3);
InsertTriangleHow(BspClass::POS, &alone, m, bsp3);
Split the BSP stuff off the mesh stuff, and the text screens themselves of the text window utility functions. [git-p4: depot-paths = "//depot/solvespace/": change = 1806]
2008-06-23 16:37:12 +08:00
}
return;
}
Reference in New Issue Copy Permalink