- GMP: name the parameters in the configure probe from
0001-GMP_GCC15.patch. Unnamed parameters are a hard error on GCC 10
and older, so configure failed there.
- TBB: turn off the automatic hwloc search so a system libhwloc-dev is
not picked up.
- scripts/linux.d/debian: run apt update before checking which
webkit2gtk package exists, and prefer webkit2gtk-4.1 since the build
requires it. Add automake, which MPFR's autoreconf needs.
The deps superbuild passes the Visual Studio generator and platform to
every sub-build but not the toolset, so build_win.bat -d -l without -x
compiled every dependency with cl even though the superbuild had been
configured with -T ClangCL; CMake replaces the forwarded
CMAKE_<LANG>_COMPILER with whatever the toolset ran. The recipes that
adapt to clang-cl then disagreed with what had been built, and
wxInspector told FindwxWidgets to look in lib/clang_x64_lib while the
cl-built wxWidgets had installed into lib/vc_x64_lib:
Could NOT find wxWidgets (missing: wxWidgets_LIBRARIES
wxWidgets_INCLUDE_DIRS core base aui propgrid)
Forward CMAKE_GENERATOR_TOOLSET as well, so the dependencies compile
with clang-cl under MSBuild the way they already do under Ninja. Four
of them need more than that:
- OpenSSL always builds with cl, and MSBuild runs its nmake steps in
the project's toolset environment, where ClangCL puts clang's include
directory first and cl trips over clang's stdint.h. The project gets
the default toolset.
- Boost.Container's dlmalloc needs -Wno-incompatible-pointer-types
under clang. boost_container links as C++, and the Visual Studio
generator writes only the link language's flags into the project, so
its C file never saw CMAKE_C_FLAGS. Under that generator the option
goes through the C++ flags as well, with the defaults kept.
- Draco's tools and NLopt's testopt compile sources their own static
library also contains. MSBuild lists libraries before objects and
lld-link resolves archive members as each input arrives, so the
library's copy is pulled in before the executable's own object and
the link fails on duplicate symbols; link.exe defers the search and
Ninja lists the objects first. Nothing uses those executables, so
they get /FORCE:MULTIPLE there.
The Ninja path is unchanged: the generated configure commands of all
29 dependencies are identical before and after. OCCT's arm64 override
to cl still applies under Ninja but not under the Visual Studio
generator, where the toolset wins; that combination never built and is
left for a follow-up.
VS 2026's ARM64 code generator needs about 27 GB for
_PyUnicode_ToNumeric, a switch with 1951 cases in
Objects/unicodetype_db.h; the same file takes under 1 GB on x64. The
16 GB CI runner has an 18.9 GB commit limit and only gets through when
Windows grows the pagefile on the temp disk in time, so cold arm64
dependency builds fail at random with C1002 "compiler is out of heap
space". build_release_vs.bat returns 0 on failure, so the job still
reports success and the incomplete dependencies are cached.
A property sheet compiles that one file with optimisation off on arm64;
the rest stays whole-program optimised and x64 is unchanged.
MSBuild reads it from PCbuild/msbuild.rsp, which is now written at
configure time and copied in, so a checkout path with spaces works too.
* ci: build the Windows x64 dependencies and slicer with clang-cl
Passes -l -x from the Windows jobs, using the clang-cl and Ninja options
added in #15373. Ninja runs each dependency's build step as a plain command,
so the jobs set up a VC environment for OpenSSL's nmake. arm64 stays on MSVC
for now.
The deps cache key gains the compiler, so windows-x64 becomes
windows-x64-clang and windows-arm64 becomes windows-arm64-msvc.
* deps: select OpenSSL's ARM64 target from DEPS_ARCH
CMAKE_GENERATOR_PLATFORM is only set by -A, which Ninja never receives, so
the Ninja build selected the x64 target on ARM64. DEPS_ARCH is derived from
CMAKE_SYSTEM_PROCESSOR and is already independent of the generator.
* ci: build the Windows ARM64 dependencies and slicer with clang-cl
Three dependencies need handling first. libpng and OpenCV each build an ARM
SIMD path that does not compile with clang-cl, so those paths are off; PNG
already had the same opt-out for Apple ARM. OCCT is built with cl, since
clang-cl cannot emit one of its large generated files and there is no option
to turn that off. All three are gated to Windows ARM64 with clang.
git config wrote core.autocrlf into the repository git init creates in the
extracted CPython source. Nothing outside deps/build reads that repository, so
the setting was already contained.
-c applies the override to the one invocation instead, so no repository config is
written at all. It cannot reuse PATCH_CMD, so the shared flags are spelled out here.
* fix(deps): build the dependencies from scratch with clang-cl
Six dependencies fail once the superbuild compiles them with clang-cl instead
of cl:
- OpenSSL never goes through CMake. Its VC-WIN64A makefile only works with cl,
and an unquoted clang-cl path with spaces produces no .obj files at all, so
the lib step dies with LNK1181. Pin the upstream toolchain.
- Boost.Container's bundled dlmalloc passes int* to the Interlocked API. cl
warns, clang rejects it.
- curl 7.75's configure probes rely on C laxness clang rejects. The results
flip and nonblock.c ends up in the AmigaOS IoctlSocket branch.
- OCCT installs RelWithDebInfo into bini/libi while find_package looks in lib.
It also prepends -Wl,-s to the shared linker flags for every Clang build,
which the MSVC-style linker gets as an argument it does not know. Both
patched hunks sit inside if (MSVC) in the OCCT sources.
- wxWidgets lands in lib/clang_x64_lib, so wxWidgetsConfig.cmake falls back to
the layout that exists instead of assuming vc_x64_lib. It tries the derived
path first, so a cl-built tree consumed by clang-cl keeps resolving the way
it does today. The patch step also resets the one file it touches, so it can
run again after an interrupted build or after the patch itself changed.
- wxInspector goes through FindwxWidgets, which only searches lib/vc*_lib
because _WX_TOOL is hardcoded to vc. It now gets the root and lib dir
derived the same way wxWidgetsConfig.cmake derives them.
Eigen is the seventh, and it breaks on the generator rather than the compiler.
Its test, lapack and blas/testing subdirectories all call
enable_language(Fortran), and they default to ON because the dependency
configures as its own top-level project. Whether that hurts depends on what
CMake finds: the Visual Studio generator supports no Fortran and finds nothing,
clang-cl sits next to the LLVM toolset's flang and works, while MSVC with Ninja
finds Strawberry Perl's MinGW gfortran, which this build already requires for
OpenSSL, and hands it the MSVC-style /machine:x64 that MinGW's ld reads as a
missing input file. The configure dies there and takes every dependency still
in flight with it. Only the headers are consumed here, so the three subprojects
are off.
* fix(deps): honor the superbuild's generator and compiler in sub-builds
orcaslicer_add_cmake_project pinned every dependency sub-build to the Visual
Studio generator whenever MSVC was true, which is also true for clang-cl. That
generator selects its compiler by toolset and ignores the CMAKE_C_COMPILER and
CMAKE_CXX_COMPILER this file already forwards, so the dependencies were built
with cl.exe no matter which generator or compiler the superbuild was given.
Key the three affected decisions on the generator instead: which generator the
sub-builds use, whether CMAKE_BUILD_TYPE is forwarded, and /m versus -j. A
Visual Studio superbuild is unchanged, so the default path and CI behave
exactly as they do today.
build_release_vs.bat now accepts -l to select clang-cl, alongside the existing
-x for Ninja, so the generator and the compiler can be chosen independently. On
the Visual Studio generator -l reaches the slicer only, through the ClangCL
toolset, because the dependency sub-builds have no toolset to inherit; a deps
build in that combination says so rather than quietly using MSVC.
* fix(deps): use upstream wxWidgets compiler layout fix
The compiler-prefix layout fix now comes from SoftFever/Orca-deps-wxWidgets#7, so remove the duplicated local patch and apply step.
* fix(deps): stop Assimp enabling ccache on the RC rule
ASSIMP_BUILD_USE_CCACHE defaults on and applies the launcher through the
global RULE_LAUNCH_COMPILE property, so it wraps the resource-compiler rule
as well. Under Ninja that rule goes through cmcldeps, which does not survive
being launched by ccache, and the build fails with clang-cl reporting /fo as
a missing file.
The superbuild already forwards CMAKE_<LANG>_COMPILER_LAUNCHER, which CMake
applies per language and so keeps clear of the RC rule.
---------
Co-authored-by: SoftFever <103989404+SoftFever@users.noreply.github.com>
Co-authored-by: raistlin7447 <kris.austin@gmail.com>
SLIC3R_CAD=OFF now builds without SolveSpace or OCCT's ModelingAlgorithms
module, leaving the dependency set identical to upstream's, and the Design
tab's mate connector preference no longer appears in builds without the tab.
When the deps prefix and the project disagree about the option, the configure
fails naming the cause, rather than failing at link time or at first launch
on Windows. The Windows packaging step stages exactly the toolkits libslic3r
links.
git apply inherits the caller's configuration, so under core.autocrlf=input it
rewrites the patched PCbuild/find_python.bat to LF. cmd.exe cannot resolve goto
labels in an LF batch file, so CPython's build fails with "The system cannot
find the batch label specified - begin_search" and then "Cannot locate
python.exe on PATH or as PYTHON variable".
git init already runs in the extracted source, so setting core.autocrlf on the
repository it creates is enough, without touching the shared PATCH_CMD.
Review request on PR #15238: "Move the SLVS to deps if the source code remains
unmodified. If any changes were made to the source code, move it to deps_src."
It is unmodified — all 20 files under src/libslic3r/slvs were byte-identical to
JacobStoren/SolveSpaceLib@4d87045, the extraction of solvespace.com's libslvs.
So deps/ it is, fetched by hash like every other dependency.
Only the CMakeLists is ours: upstream's builds a demo executable and installs
nothing, so deps/SLVS/CMakeLists.txt.in replaces it via PATCH_COMMAND — the same
shape deps/OpenCSG already uses. The public header keeps its spelling, so
SketchSolver.cpp still says `#include <slvs.h>` and needs no edit.
The CI deps cache is keyed on hashFiles('deps/**'), so it rebuilds itself.
Verified: dep_SLVS builds and installs, libslic3r links against SLVS::slvs, and
the CAD suite is unchanged at 2518 assertions in 194 test cases.
The static libavcodec.a/avutil.a compiled the auto-detected
videotoolbox/audiotoolbox objects, which reference VideoToolbox
framework symbols (_VTDecompressionSession*). The app link line
happened to satisfy them transitively, but the orca_stubgen module
link (CI-only) failed with undefined symbols. The player decodes in
software (swscale), so disable both HW-accel paths to keep the
static libs self-contained.
Co-Authored-By: Claude <noreply@anthropic.com>
The literal --enable-shared was always overridden by ${_link_cmd}
(--enable-static --disable-shared on Apple, --enable-shared elsewhere)
and FFmpeg configure processes these flags in order, last one wins.
Remove it and the stale comment documenting the workaround.
Co-Authored-By: Claude <noreply@anthropic.com>
Ian Chua