This is unfortunate, but OpenMP is not available on some toolchains
(most MinGW builds) and, more importantly, breaks the distribution
model we use on Windows, which is a single self-contained executable.
Leave the OpenMP disabled by default everywhere so that we don't have
any second-class platforms where SolveSpace is slower than on first-
class ones.
Fixes#631.
Historically SolveSpace used its own heap on Windows since it gave
better control and debugging options, but a lot of development these
days happens on Linux, where that heap was a stub around malloc/free,
and also Windows debugging tools got a lot better.
In terms of immediate benefit, this commit fixes heap corruption
on Windows introduced in commits b4e1ce44 and 47e82798, caused
by allocating with HEAP_NO_SERIALIZE in parallel from OpenMP threads.
Without HEAP_NO_SERIALIZE there's no performance benefit to keeping
our own heap, either.
The vl() function is also removed because for development there are
better tools now, and the only place where it was permanently called
from became a no-op, since temporary heap always validates after
FreeAllTemporary() recreates it.
Both GCC and Clang use x87 instructions by default on 32-bit x86
targets, but the loss of precision resulting from that has yielded
crashing tests (see referenced issue), and it can be safely assumed
there are very few CPUs that both don't support SSE2 and are expected to
run SolveSpace
This commit also removes a now-redundant check for 32-bit Windows
against TARGET, which doesn't seem to be actually set by CMake at all
Ref: #565
This has lead to linker problems if the environment does have standard
versions of zlib/png (like from MSYS2), see discussion in referenced
issue
Closes#559
It turns out that using cmake_policy(VERSION) is effectively same as
using cmake_minimum_required(VERSION) because it is a hard error
if the CMake running the script is too old. As such, replace 22525e65
with two changes:
* Set the minimum required version to 3.7.2 (Debian stretch),
which also configures policies.
* If CMake is at least as new as 3.11.0, then set the policies
according to that version.
This should actually have the intended behavior.
By setting WINVER=0x0501 (Windows XP) in CMakeLists.txt and adding a few
missing defines in guiwin.cpp and configuring OPENGL=1 in CMake
Solvespace (3.0~25b6eba1) compiles and works perfectly on Windows XP.
Tested with MinGW GCC-6.3.0-1
After this commit, if the target system does have modern OpenGL
drivers installed, ANGLE is configured to use them, bypassing most
translation (shaders still have to be translated from ESSL to GLSL).
If there are no OpenGL drivers, such as if the graphics drivers were
installed via Windows Update, DirectX translation is still used. This
results in a very noticeable startup delay and minor performance
degradation.
In addition it is no longer necessary to build with -DOPENGL=1 to be
able to run the binary in wine; everything works out of the box.
Before, wine's incomplete HLSL translator would crash.
This change required renaming the variable `texture` in shaders,
since it shadows the Core GLSL function with the same name, and ANGLE
translates texture2D() calls to texture() calls.
We plan to use flatbuffers in the future for the next generation of
the .slvs file format, so flatbuffers are built unconditionally; and
the Q3DO exporter itself is tiny.
It's a deprecated platform that has weird OpenGL-related bugs and
is incompatible with using EGL anyway. It was clear we're going
to drop it, the only question was when. Answer: now.
This commit fixes four issues:
* Instead of WRITE,APPEND, resource.rc was generated using
WRITE,WRITE, which erased #include <windows.h> and prevented
any symbolic definitions (like RT_MANIFEST) from working.
This silently included them using a string type instead,
which did nothing.
* WINVER is bumped to Win7, since that's what we target now.
* Index of RT_MANIFEST is changed to 2, since that's what
it has to be when ISOLATION_AWARE_ENABLED is defined.
* Platform is not restricted to X86 in manifest, since there
is no point in doing so.
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.
The only thing we need it anymore is the package version and platform
configuration, so only include it for that. As a result, less files
are rebuilt when the git commit changes and config.h is regenerated.
This commit performs two main changes:
* Alters the shaders to use only strictly conformant GLSL 2.0.
* Alters the Windows UI to use ANGLE via GL ES 2.0 and EGL 1.4.
This commit also drops official support for Windows XP, since ANGLE
requires a non-XP toolset to build. It is still possible to build
SolveSpace for Windows XP using:
cmake -T v120_xp -DOPENGL=1
We do not support clobbering the source tree with build products,
nor recommend this in any documentation, and this already leads to
weird bugs with the resource system. Show an explicit error in this
case to avoid confusion.
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.
This commit makes common external packages always be included through
find_package to eliminate differences in variables set, wraps
find_package for vendored libraries on Windows to factor out common
code, and removes miscellaneous useless code elsewhere in dependency
handling.
This also fixes a problem where pkg-config would pick up `build`
libraries instead of `host` when cross-compiling.
It was never really needed, since both Linux and OS X, where
GlOffscreen is used, guarantee that the API we need is present,
on all OS versions we're interested in.
Also, reorganize GlOffscreen consistently with the rest of our
codebase, and don't use RAII for OpenGL resource management because
of its requirement for an active context.
наб
whitequark
nabijaczleweli
Koen Schmeets
Ruevski
Sergiusz Bazanski
Elvira Khabirova