Before this commit, it was possible to add some redundant constraints
(e.g. vertical, horizontal or midpoint) without failing the sketch,
because SolveBySubstitution() removed the redundant equations.
However, this could result in the solve failing later because
the system didn't converge, without any pointers as to the true
cause of the failure.
It was broken because of three bugs:
* Uninitialized variables in RunCommand;
* Trying to use (OEM-encoded) main() argc/argv arguments instead
of GetCommandLineW();
* Trying to pass relative paths directly into ssfopen.
Before this commit, if constraints with newly introduced params were
loaded from a file that linked other files, the upgrade code would
attempt to look up a non-existent entity.
All of our executables need resources; e.g. the vector font is
a resource and it is necessary for generation. Before this commit,
the GUI executable loaded the resources in a nice way, and everything
else did it in a very ad-hoc, fragile way.
After this commit, all executables are placed in <build>/bin and
follow the same algorithm:
* On Windows, resources are compiled and linked into every
executable.
* On Linux, resources are copied into <build>/res (which is
tried first) and <prefix>/share/solvespace (which is tried
second).
* On macOS, resources are copied into <build>/res (which is
tried first) and <build>/bin/solvespace.app/Contents/Resources
(which is tried second).
In practice this means that we can add as many executables as we want
without duplicating lots of code. In addition, on macOS, we can
place supplementary executables into the bundle, and they can use
resources from the bundle transparently.
There are two main reasons to desire an OpenGL 2 renderer:
1. Compatibility. The compatibility profile, ironically, does not
offer a lot of compatibility, and our OpenGL 1 renderer will not
run on Android, iOS, or WebGL.
2. Performance. The immediate mode does not scale, and in fact
becomes very slow with only a moderate amount of lines on screen,
and only a somewhat large amount of triangles.
This commit implements a basic OpenGL 2 renderer that uses only
features from the (OpenGL 3.2) core profile. It is not yet faster
than the OpenGL 1 renderer, primarily because it uses a lot of small
draw calls.
This commit uses OpenGL 2 on Linux and Mac OS X directly (i.e. links
to the GL symbols from version 2+); on Windows this is impossible
with the default drivers, so for now OpenGL 1 is still used there.
This is a common and convenient behavior; the basename is
pre-selected, so exporting multiple views requires just one keystroke
to put the cursor after the basename.
If a generated mesh is non-watertight, and this is not noticed for
some reason (e.g. the markers are dismissed), and the mesh is
exported for further processing, it could cause problems down
the line.
These points can be used for constraining the width of the text
(or to the width of the text).
The main parts of the commit are:
* TtfFont is restructured to be able to return the aspect ratio
for a given string.
* This aspect ratio is written to the savefile, such that even if
the font is missing, the sketch would still be solved correctly.
* The two additional points are constrained via perpendicularly
to the two main points (which form a v vector).
The compatibility features are as follows:
* When the font is missing in old files, 1:1 aspect ratio is used,
which works for the replacement symbol anyhow.
* When the two additional points are missing in old files, their
would-be positions are calculated and they are moved there,
avoiding 'jumping' of underconstrained sketches.
Sometimes, after a large change in a sketch, constraints that are
geometrically fine may still cause the rank test to fail. One way
this can happen is VectorsParallel() pivoting wrong due to the big
move, converging anyways but ending up singular. It would then
re-pivot correctly on the new solution when you re-solve, making
this a transient error. This is visible when dragging the arm in
the jansen-asm.slvs example.
After this commit, if the rank test fails, equations are regenerated
the Jacobian is rewritten, and the rank test is retried, which
prevents these transient errors from interfering with dragging.
The problem described above was invisible before c011444, as rank
test was only performed before solving.
Unfortunately there is no portable way to open an Unicode path with
std::fstream. On *nix, it is enough to call the const char*
constructor. On MSVC, it is enough to call a nonstandard
const wchar_t* constructor. However, on MinGW, there is no way at all
to construct an std::fstream with a wide path, not even using
undocumented APIs. (There used to be a const wchar_t* overload added
back in libstdc++ 4.7, but it got removed for a reason that I was not
able to find out.)
We're using gcov+lcov, since these tools appear to be the only
usable ones that use the SC/CC metric; and measuring just the line
coverage would be practically criminal negligence.
gcov only works with GCC and Clang, and MSVC's own coverage
measurement tools are not up to the task; so MSVC is out of luck.
This commit alters the build system substantially; it adds another
platform, `headless`, that provides stubs in place of all GUI
functions, and provides a library `solvespace_headless` alongside
the main executable. To cut down build times, only the few files
that have #if defined(HEADLESS) are built twice for the executable
and the library; the rest is grouped into a new `solvespace_cad`
library. It is not usable on its own and just serves for grouping.
This commit also gates the tests behind a -DENABLE_TESTS=ON CMake
option, ON by default (but suggested as OFF in the README so that
people don't ever have to install cairo to build the executable.)
The tests introduced in this commit are (so far) rudimentary,
although functional, and they serve as a stepping point towards
introducing coverage analysis.
Without -fno-exceptions, the branch coverage information is
practically useless, as every call becomes a branch.
The functionality of Expr is retained as-is, although SjLj error
handling is a maintenance nightmare. However, the entire parser
probably should be eventually replaced, so for now it is not
a great concern.
SurfaceRenderer is a new renderer implementing the Canvas interface
running entirely on the CPU; it projects strokes and triangles
in the exact same way as OpenGL would, and it can be used for
rendering into raster or vector 2d surfaces.
This has the following benefits:
* Less geometry to generate; we can do both in one pass;
* Less geometry to draw;
* Eliminate overdraw of outlines on top of emphasized edges;
* In future, being able to seamlessly stitch stippled lines.
The contour edges are now also drawn before emphasized edges;
this makes intersections of contour and emphasized edges look better
as the thinner emphasized edge doesn't clobber the depth buffer.
This has several desirable consequences:
* It is now possible to port SolveSpace to a later version of
OpenGL, such as OpenGLES 2, so that it runs on platforms that
only have that OpenGL version;
* The majority of geometry is now rendered without references to
the camera in C++ code, so a renderer can now submit it to
the video card once and re-rasterize with a different projection
matrix every time the projection is changed, avoiding expensive
reuploads;
* The DOGD (draw or get distance) interface is now
a straightforward Canvas implementation;
* There are no more direct references to SS.GW.(projection)
in sketch rendering code, which allows rendering to multiple
viewports;
* There are no more unnecessary framebuffer flips on CPU on Cocoa
and GTK;
* The platform-dependent GL code is now confined to rendergl1.cpp.
* The Microsoft and Apple headers required by it that are prone to
identifier conflicts are no longer included globally;
* The rendergl1.cpp implementation can now be omitted from
compilation to run SolveSpace headless or with a different
OpenGL version.
Note these implementation details of Canvas:
* GetCamera currently always returns a reference to the field
`Camera camera;`. This is so that a future renderer that caches
geometry in the video memory can define it as asserting, which
would provide assurance against code that could accidentally
put something projection-dependent in the cache;
* Line and triangle rendering is specified through a level of
indirection, hStroke and hFill. This is so that a future renderer
that batches geometry could cheaply group identical styles.
* DrawPixmap and DrawVectorText accept a (o,u,v) and not a matrix.
This is so that a future renderer into an output format that
uses 2d transforms (e.g. SVG) could easily derive those.
Some additional internal changes were required to enable this:
* Pixmap is now always passed as std::shared_ptr<{const ,}Pixmap>.
This is so that the renderer could cache uploaded textures
between API calls, which requires it to capture a (weak)
reference.
* The PlatformPathEqual function was properly extracted into
platform-specific code. This is so that the <windows.h> header
could be included only where needed (in platform/w32* as well
as rendergl1.cpp).
* The SBsp{2,3}::DebugDraw functions were removed. They can be
rewritten using the Canvas API if they are ever needed.
While no visual changes were originally intended, some minor fixes
happened anyway:
* The "emphasis" yellow line from top-left corner is now correctly
rendered much wider.
* The marquee rectangle is now pixel grid aligned.
* The hidden entities now do not clobber the depth buffer, removing
some minor artifacts.
* The workplane "tab" now scales with the font used to render
the workplane name.
* The workplane name font is now taken from the normals style.
* Workplane and constraint line stipple is insignificantly
different. This is so that it can reuse the existing stipple
codepaths; rendering of workplanes and constraints predates
those.
Some debug functionality was added:
* In graphics window, an fps counter that becomes red when
rendering under 60fps is drawn.
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.
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.
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.
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.
This helps to ensure that a base class that changes underneath us
would not leave any overridden functions hanging.
This already highlighted some questionable use of GTKMM's API,
which were also fixed in this commit.
This commit integrates the bitmap font in the resource system, so
that cross-compilation would be easier.
The font handling code was carefully written to do glyph parsing
lazily; in practice this means that after this commit, startup
is less than 25ms slower, most of it spent in inflate().
This should also result in faster rendering, since there is no
rampant plane switching anymore; instead, all characters that are
actually used are stashed into same one texture.
This commit integrates icons in the resource system so that they
can be loaded (or reloaded, without restarting) in @2x mode, which
will be added in a future commit. png2c is no longer necessary.
png2c used to perform the following transformation:
if(r + g + b < 11) r = g = b = 11;
This is now achieved by switching the icons to RGBA mode and adding
alpha channel with the following imagemagick invocation, which is
equivalent to the transformation above:
for i in *.png; do
convert -fuzz 4% -channel rgba -matte \
-fill "rgba(255,255,255,0)" -opaque black \
$i $i
done
The Debian package solvespace now includes /usr/share/solvespace;
this should be split out into solvespace-data later.
Currently, icons, fonts, etc are converted to C structures at compile
time and are hardcoded to the binary. This presents several problems:
* Cross-compilation is complicated. Right now, it is necessary
to be able to run executables for the target platform; this
happens to work with wine-binfmt installed, but is rather ugly.
* Icons can only have one resolution. On OS X, modern software is
expected to take advantage of high-DPI ("Retina") screens and
use so-called @2x assets when ran in high-DPI mode.
* Localization is complicated. Win32 and OS X provide built-in
support for loading the resource appropriate for the user's
locale.
* Embedding strings can only be done as raw strings, using C++'s
R"(...)" literals. This precludes embedding sizable strings,
e.g. JavaScript libraries as used in Three.js export, and makes
git history less useful. Not embedding the libraries means we
have to rely on external CDNs, which requires an Internet
connection and adds a glaring point of failure.
* Linux distribution guidelines are violated. All architecture-
independent data, especially large data such as fonts, is
expected to be in /usr/share, not in the binary.
* Customization is impossible without recompilation. Minor
modifications like adding a few missing vector font characters
or adjusting localization require a complete development
environment, which is unreasonable to expect from users of
a mechanical CAD.
As such, this commit adds a resource system that bundles (and
sometimes builds) resources with the executable. Where they go is
platform-dependent:
* on Win32: into resources of the executable, which allows us to
keep distributing one file;
* on OS X: into the app bundle;
* on other *nix: into /usr/share/solvespace/ or ../res/ (relative
to the executable path), the latter allowing us to run freshly
built executables without installation.
It also subsides the platform-specific resources that are in src/.
The resource system is not yet used for anything; this will be added
in later commits.
This is good practice and helps to catch bugs. Several changes
were made to accomodate the newly enabled warnings:
* -Wunused-function:
* in exposed/, static functions that were supposed to be inlined
were explicitly marked as inline;
* some actually unused functions were removed;
* -Wsign-compare: explicit conversions were added, and in
the future we should find a nicer way than aux* fields;
* -Wmissing-field-initializers: added initializers;
* -Wreorder: reordered properly;
* -Wunused-but-set-variable: remove variable.
-Wunused-parameter was turned off as enabling it would result in
massive amount of churn in UI code. Despite that, we should enable
it at some point as it has a fairly high SNR otherwise.
Before this commit, when exporting a vector file without the shaded
model shown, or similarly when using formats that we do not export
the mesh to, we still generate (and then discard) the mesh in paint
order. This is a waste of time.
The immediate reason for refactoring this was that the GTK port broke
after 52af7256 since config.h is not included anymore, but it was
a fragile piece of code I will shed no tears for.
While we're at it, get rid of the mutable std::string &file to be
consistent with our conventions.
config.h now includes the git hash and so, as long as it's included
in solvespace.h, any change of git HEAD will trigger a complete
recompilation, which makes bisecting especially annoying.
While we're at it, remove HAVE_STDINT_H from it, since we require
C++11 and all MSVC versions that include C++11 also include stdint.h.
Before this commit, the graphics window edit control always had
a width of 30 average character widths.
After this commit, the edit control has a width of 5 average
character widths (for numeric constraints) or 30 average character
widths (for comment constraints), or just enough to display
the entire value being edited, whichever is greater.
This makes the edit control overlap the sketch less in case of
editing numeric constraints (since in most cases, the numbers being
edited are short), and removes annoying scrolling in case of editing
long comments.
Before this commit, the position of the edit box was adjusted
by trial and error, as far as I can tell. This commit changes
the positioning machinery for edit controls as follows:
The coordinates passed to ShowTextEditControl/ShowGraphicsEditControl
now denote: X the left bound, and Y the baseline.
The font height passed to ShowGraphicsEditControl denotes
the absolute font height in pixels, i.e. ascent plus descent.
Platform-dependent code uses these coordinates, the font metrics
for the font appropriate for the platform, and the knowledge of
the decorations drawn around the text by the native edit control
to position the edit control in a way that overlays the text inside
the edit control with the rendered text.
On OS X, GNU Unifont (of height 16) has metrics identical to
Monaco (of height 15) and so as an exception, the edit control
is nudged slightly for a pixel-perfect fit.
Also, since the built-in vector font is proportional, this commit
also switches the edit control font to proportional when editing
constraints.
A new button is added, "Show/hide outline of solid model".
When the outline is hidden, it is rendered using the "solid edge"
style. When the outline is shown, it is rendered using the "outline"
style.
In SolveSpace's true WYSIWYG tradition, the 2d view export follows
the rendered view exactly.
Moreover, shell edges are not rendered anymore, since there is not
much need in them anymore and not drawing them lessens the overlap
between various kinds of lines, which already includes entities,
solid edges and outlines.
Before this change, the two buttons "Show/hide shaded model" (S) and
"Show/hide hidden lines" (H) resulted in drawing the following
elements in the following styles:
Button | Non-occluded | Non-occluded | Occluded | Occluded
state | solid edges | entities | solid edges | entities
--------+--------------+--------------+-------------+--------------
!S !H | | | solid-edge | entity style
--------+ | +-------------+--------------
S !H | | | invisible
--------+ solid-edge | entity style +-------------+--------------
!S H | | | |
--------+ | | solid-edge | entity style
S H | | | |
--------+--------------+--------------+-------------+--------------
After this change, they are drawn as follows:
Button | Non-occluded | Non-occluded | Occluded | Occluded
state | solid edges | entities | solid edges | entities
--------+--------------+--------------+-------------+--------------
!S !H | | | solid-edge | entity style
--------+ | +-------------+--------------
S !H | | | invisible
--------+ solid-edge | entity style +-------------+--------------
!S H | | | |
--------+ | | hidden-edge | stippled¹
S H | | | |
--------+--------------+--------------+-------------+--------------
¹ entity style, but the stipple parameters taken from hidden-edge
In SolveSpace's true WYSIWYG tradition, the 2d view export follows
the rendered view exactly.
Also, it is now possible to edit the stipple parameters of built-in
styles, so that by changing the hidden-edge style to non-stippled
it is possible to regain the old behavior.
Before this commit, trying to export image on *nix platforms yielded
a black rectangle, since since there is nowhere to render to
when we're not in a GUI toolkit draw callback.
On Windows, nothing changes: we do a repaint without the toolbar,
glReadPixels, export. On *nix, we create another offscreen rendering
context, render into it, then destroy it. As a bonus this avoids
some minor flickering that would happen if we reused the regular
rendering path.
Most of these were just converting char* into std::string back and
forth; some more used ReadUTF8, which was converted to use nicer
STL-style iterators over UTF-8 text.
The remaining ones are:
* arguments to Expr::From, which we'll change when refactoring
the expression lexer;
* arguments to varargs functions, which we'll change when adding
localization (that requires custom printf-style functions to
allow for changing argument order);
* arguments where only string literals are ever passed, which
are OK;
* in platform-specific code, which is OK.
After commit 2f734d9, inactive groups are no longer regenerated
for trivial changes, e.g. changing parameters, so it's possible to
switch to an earlier group and work on it without incurring
the computational (slowdown) and cognitive (annoyance by red
background) overhead of later groups failing to solve.
However, if a group--any group anywhere--was not solved OK,
the interface reacted accordingly, which diminished usefulness of
the change, especially given that, if we have groups A and B with
B depending on A, if B is broken by a change in A and we activate A
and fix it, B will not be regenerated.
After this commit, only active groups are considered when deciding
if generating the entire sketch would fail.
This font is less complete than our bitmap font, Unifont: Unifont
has essentially complete Unicode coverage and LibreCAD's font only
has Latin, Cyrillic and Japanese, but it can be extended rather
easily, so this should be fine for now.
These embedded fonts fatten glhelper.o quite a bit:
bitmapfont.table.h is about 8M in gzip-compressed bitmaps and
vectorfont.table.h is about 2M in raw vector data.
In spite of that it takes just around five seconds to build
glhelper.c on my laptop, so it should be fine.
The final executable grows from about 2M to about 8M, but this
is a small price to pay for fairly extensive i18n support.
The new font has somewhat different metrics, so the rendering
code has been fudged to make it look good.
Benefits:
* Much simpler code.
* Handles the entire TTF spec, not just a small subset that
only really worked well on Windows fonts.
* Handles all character sets as well as accented characters.
* Much faster parsing, since Freetype lazily loads and
caches glyphs.
* Support for basically every kind of font that was invented,
not just TTF.
Note that OpenType features, e.g. ligatures, are not yet supported.
This means that Arabic and Devanagari scripts, among others, will
not be rendered in their proper form.
RTL scripts are not supported either, neither in TTF nor in
the text window. Adding RTL support is comparatively easy, but
given that Arabic would not be legibly rendered anyway, this is not
done so far.
Commit 89eb208 has improved the overall situation with chord
tolerance, but it changed the display chord tolerance to use
an absolute value in millimeters as a stopgap measure.
This commit changes the display chord tolerance to be specified
in percents of entity bounding box instead of millimeters.
As a result, the linearized curves are both zoom level and sketch
scale independent.
In order to compute the bounding box, all entities are generated
twice. However, this shouldn't result in a noticeable slowdown,
since the bounding box calculation does not need the expensive
triangle mesh generation and the solver will converge immediately
on the second run.
Since the meaning of the preference has changed, a new name is
used (ChordTolerancePct instead of ChordTolerance), so that it
would be reset to the default value after updating SolveSpace.
The default value, 0.5%, was selected using trial and error by
judging whether cylinders of moderate dimensions were looking
aesthetically pleasing enough.
After this change, the only real function of the spacebar
shortcut is to reload imported groups, since manual regeneration
should not change anything anymore unless there is a bug.
Before this commit, a single chord tolerance was used for both
displaying and exporting geometry. Moreover, this chord tolerance
was specified in screen pixels, and as such depended on zoom level.
This was inconvenient: exporting geometry with a required level of
precision required awkward manipulations of viewport. Moreover,
since some operations, e.g. mesh watertightness checking, were done
on triangle meshes which are generated differently depending on
the zoom level, these operations could report wildly different
and quite confusing results depending on zoom level.
The chord tolerance for display and export pursue completely distinct
goals: display chord tolerance should be set high enough to achieve
both fast regeneration and legible rendering, whereas export chord
tolerance should be set to match the dimension tolerance of
the fabrication process.
This commit introduces two distinct chord tolerances: a display
and an export one. Both chord tolerances are absolute and expressed
in millimeters; this is inappropriate for display purposes but
will be fixed in the next commits.
After exporting, the geometry is redrawn with the chord tolerance
configured for the export and an overlay message is displayed;
pressing Esc clears the message and returns the display back to
normal.
When a solver error arises after a change to the sketch, it should
be easy to understand exactly why it happened. Before this change,
two functionally distinct modes of failure were lumped into one:
the same "redundant constraints" message was displayed when all
degrees of freedom were exhausted and the had a solution, but also
when it had not.
To understand why this is problematic, let's examine several ways
in which we can end up with linearly dependent equations in our
system:
0) create a triangle, then constrain two different pairs of edges
to be perpendicular
1) add two distinct distance constraints on the same segment
2) add two identical distance constraints on the same segment
3) create a triangle, then constrain edges to lengths a, b, and c
so that a+b=c
The case (0) is our baseline case: the constraints in it make
the system unsolvable yet they do not remove more degrees of freedom
than the amount we started with. So the displayed error is
"unsolvable constraints".
The constraints in case (1) remove one too many degrees of freedom,
but otherwise are quite like the case (0): the cause of failure that
is useful to the user is that the constraints are mutually
incompatible.
The constraints in cases (2) and (3) however are not like the others:
there is a set of parameters that satisfies all of the constraints,
but the constraints still remove one degree of freedom too many.
It makes sense to display a different error message for cases (2)
and (3) because in practice, cases like this are likely to arise from
adjustment of constraint values on sketches corresponding to systems
that have a small amount of degenerate solutions, and this is very
different from systems arising in cases like (0) where no adjustment
of constraint values will ever result in a successful solution.
So the error message displayed is "redundant constraints".
At last, this commit makes cases (0) and (1) display a message
with only a minor difference in wording. This is deliberate.
The reason is that the facts "the system is unsolvable" and
"the system is unsolvable and also has linearly dependent equations"
present no meaningful, actionable difference to the user, and placing
emphasis on it would only cause confusion.
However, they are still distinguished, because in case (0) we
list all relevant constraints (and thus we say they are "mutually
incompatible") but in case (1) we only list the ones that constrain
the sketch further than some valid solution (and we say they are
"unsatisfied").
The current messages accurately reflect what happens to the system
of equations that represents the sketch, but can be quite confusing
to users that only think in terms of the constraints.
We use "unsolvable" and not "impossible" because while most of
the cases that result in this error message will indeed stem from
mutually exclusive sets of constraints, it is still possible that
there is some solution that our solver is unable to find using
numeric methods.
Most people just want a single self-contained .html file, but more
advanced usage will involve embedding in a webpage, where the default
viewer would be copied and customized, and fed with bare mesh export.
It works. Mostly. Sort of. Only on Windows fonts. Sometimes it
randomly refuses to render glyphs (try `х`, that's not a latin ex).
I'm not really sure why, the logic seems right.
Why do we have a homegrown TTF parser anyway? It's kind of awful.
It breaks on any slightly unusual input. It plows through UTF-16BE
font names like a nuclear-powered steamroller. It outright ignores
composite glyphs (is that why it's broken this time?). The kerning
is seizure-inducing. It ignores any characters outside BMP by design.
Maybe we should just replace it with freetype.
This removes the arbitrary 64 byte restriction (which effectively
limits us to as little as 16 Unicode characters with CJK encodings),
makes classes smaller, and is easier to use.
As a consequence of making the length of all ex-NameStr fields
unbounded, all functions that returned a buffer derived from those
were changed to return std::string. Then, functions that are
contextually similar to the ones described above were changed
to return std::string. Then, functions that now happened to mostly
take an std::string argument converted to a C string were changed
to accept std::string.
This has produced a bit of churn, but is probably for the better.
Now it is possible to give non-ASCII names to groups
as well as see non-ASCII filenames of imported files.
In the future this makes localization possible.
This works for LTR languages, such as European and CJK,
but not RTL such as Arabic. Does Arabic even exist in
monospaced form? I have no idea.
After this commit, SolveSpace can robustly handle non-ASCII filenames
on every OS. Additionally, on Windows, filenames longer than 260
characeters can be used, and files on network shares can be opened
directly, without mounting them as a network drive.
After this commit, SolveSpace always uses UTF-8 strings internally
(represented as char* and std::string) on every OS, for everything:
UI labels, paths and user input. OS X always uses UTF-8; on Windows,
strings are converted at boundary; and on Linux/BSD/etc SolveSpace
refuses to start unless the current locale is UTF-8. This will
negatively affect literally no one.
This commit does not bring forth any user-visible improvement yet;
in order to correctly handle Unicode in filenames, it is still
necessary to change the fopen calls to _wfopen on Windows.
After this commit, SolveSpace deals with paths as follows:
* Paths are generally treated as opaque platform-specific strings.
This helps on Linux, because paths on Linux don't have any
specific encoding and it helps to avoid any operations on them.
* The UI in some places wants to get a basename. In this case,
the newly introduced PATH_SEP is used. This allows to treat
backslash as a regular character, which it is on Linux and OS X.
* The only place where any nontrivial operations on paths are
performed is the g->impFile/impFileRel logic.
Specifically, when saved, g->impFile always contains an absolute
path with separators of the current platform, and g->impFileRel
always contains a relative path with UNIX separators. This allows
to treat backslash as a regular character.
Old files will contain g->impFileRel with Windows separators;
these are detected by looking for a drive letter in g->impFile
and in that case mapping Windows separators to UNIX ones.
There is no need to treat UNIX separators (forward slashes) in
any special way on Windows because there is no way on Windows,
not even via UNC paths, to create or address a directory entry
with a forward slash in its name.
What do we gain from this? Several things.
* First, usage of PATH_MAX (the POSIX constant) is eliminated.
PATH_MAX is actually a lie; Linux and OS X (and probably other BSDs
too) do not have an actual path length limit. Linux claims 4096,
OS X claims 1024, but it is trivial to construct paths that are
longer.
* Second, while Windows does enforce a limit of MAX_PATH (the Win32
constant) for its ASCII functions, the Unicode variants, when
used with UNC paths, do not have this restriction.
The capability to use UNC paths is useful by itself, as it allows
to access files on network shares directly.
* Third, representing paths as std::string will make it easier to
interoperate with *W WinAPI functions later.
On Windows, freeze.{cpp,h} was factored into w32main.cpp.
The old implementation was too redundant and leaked registry
key handles.
On all platforms, Cnf* functions now use std::string.
This simplifies code everywhere, but will be particularly useful
when the Windows port switches to the *W WinAPI functions.
This is necessary to store non-POD classes in these containers.
Note that List and IdList do not use MemRealloc anymore;
this is necessarily (slightly) less efficient, but is the right
semantics, as you cannot just move non-POD types, e.g. std::string,
around in memory. All STL containers provide the same guarantees
and share the performance hit.
The slowdown is mostly mitigated by moving the contained objects,
so that no additional heap allocations or copies occur beyond
that of the object itself.
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.
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.
Some extra code is necessary to determine that the back faces
should not be drawn in red for transparent solids. It is expected
that the user will first ensure that the shell is watertight
and then set the opacity; back faces are still drawn if
the opacity is exactly 1.
The savefile format is changed backwards-compatibly by stashing
the alpha value in uppermost byte of 4-byte hex color value
in Surface and Triangle clauses. The existing files have 00
in the high byte, so RgbColor::FromPackedInt treats that
as "opaque".
