The codegen only ever needed a protoc binary, but every entry point installed
grpcio-tools to get one. That drags in the grpcio C extension, which has no
Windows/ARM64 wheel and falls back to building from source there, so the ARM64
job died with "Failed building wheel for grpcio" -> "protoc not found".
tools/codegen_toolchain.py resolves the toolchain instead: protoc from $PROTOC,
PATH, a cache, grpc_tools when already installed, or a pinned checksum-verified
protoc release unpacked into .codegen-tools/; protobuf and pyyaml from the
calling interpreter or a cached virtualenv it re-execs into (distro Pythons
refuse `pip install` under PEP 668). All four build scripts and all three CI
jobs are now just `python tools/run_codegen.py`, with no pip lines around it.
tools/config_metadata_pb2.py was the one generated file checked into git. The
orca.* option extensions are now read out of the descriptor set, which already
carries config_metadata.proto via --include_imports, so nothing is generated
into the tree -- and the protobuf>=6.33.5,<7 CI pin goes away with it, since it
only existed to satisfy gencode's hard ValidateProtobufRuntimeVersion check.
Generated C++ verified byte-identical under upb and the pure-Python protobuf
runtime (what win/arm64 installs), and under both grpc_tools' and standalone
protoc.
Also fixed:
- build_release_macos.sh still passed -DPython3_EXECUTABLE=<codegen venv>,
pointing the bundled *embed* interpreter at the codegen environment -- the
same confusion fae4b124 fixed on the CMake side.
- Tab.cpp #includes TabLayout_generated.cpp but had no dependency on
codegen_config, so an incremental build after a .proto edit could compile it
while the file was being rewritten. libslic3r already had this guard.
- ConfigCodegen.cmake now probes with `codegen_toolchain.py --check` (which
never downloads or installs), prefers a host interpreter over the embed one,
and lets a fresh clone generate at configure time instead of erroring out.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The Windows build's "Install codegen tools" step ran `pip install grpcio-tools`,
whose grpcio C-extension has no Windows/ARM64 wheel for current Python and fails
to build from source on windows-11-arm, so protoc was unavailable and
run_codegen aborted at the compile step.
The codegen only needs a protoc binary + the protobuf runtime + pyyaml. Install
those directly on Windows instead: pip install protobuf (>=6.33.5, matching the
committed config_metadata_pb2.py runtime check) + pyyaml, and download a
standalone protoc (win64; runs under x64 emulation on arm64). run_codegen
prefers a standalone protoc on PATH, so grpcio-tools is no longer needed on
Windows. Linux/macOS steps keep grpcio-tools (wheels available there).
Verified locally: run_codegen with a protobuf-only venv (no grpcio-tools) +
standalone protoc 28.3 -> Lint passed (712 fields), Validation PASSED.
Resolve the 4 proto-refactored conflicts (Preset.cpp, Print.cpp,
PrintConfig.cpp, Tab.cpp) by keeping the branch's generated-include
versions; auto-merged non-conflict changes from main are preserved.
Reconcile main's new settings into the proto schema. main added 52 new
active init_fff_params settings not present in the proto; ported them all
(type/label/tooltip/default/enum/preset) into print/filament/printer.proto
so the config registers, profiles load, and the GUI does not crash on
missing defs. Added GuiType.plugin_picker to config_metadata.proto and
regenerated config_metadata_pb2.py.
Codegen: run_codegen validate PASSED (712 fields). UI layout (layout.yaml
placement) for the 52 new settings is a follow-up; they are registered and
serialized but not yet placed in GUI tabs.
Export wxWidgets include dirs and defines from libslic3r_gui on Linux so
tests outside src/ compile against its GUI headers. Relocate the bundled
Python runtime to Contents/Resources with a Contents/MacOS/python symlink
(codesign cannot seal the dotted python3.12 dirs under Contents/MacOS) and
replace the deprecated codesign --deep with explicit inside-out signing of
every Mach-O in the bundle.
Resolve five conflicts, all of which needed both sides rather than a pick:
- BackgroundSlicingProcess: ours was a pure tabs->spaces reformat of base, so
keep main's per-filament volume/nozzle map read-back (its only change here).
- GUI_App: main's #12506 else-if attached to an `if` this branch deleted;
re-expressed onto the same-agent early-return path (the agent factory caches
per id, so pointer equality is the same predicate).
- MainFrame: both sides relocated Sync Presets independently; keep main's
push_notification plus the branch's Plugins menu items.
- Tab: the "TODO: Orca: Support hybrid" blocks were unchanged base, not a branch
decision; take main's enabled Hybrid to match the already auto-merged siblings.
- test_config: union of both sides' cases (6 plugin + 9 multi-nozzle).
* ci: run unit tests on Windows and macOS
The Unit Tests CI job only ran on Linux, so platform-specific bugs
invisible to a Linux build could land undetected (e.g. the MSVC-only
NfpPlacer crash fixed in #14267). The full non-[NotWorking] suite already
builds and passes on every shipped arch, so this wires them into CI.
A reusable unit_tests.yml, called once per built arch, downloads that
arch's test artifact and runs ctest. Each build leg builds the test
executables and uploads them; a single publish_test_results job on Linux
aggregates the JUnit results into one check.
Coverage: Linux x86_64 + aarch64, Windows x64 + arm64, macOS arm64.
macOS x86_64 is deferred (cross-built on arm64, needs Rosetta).
- build_release_vs.bat: "tests" token enables BUILD_TESTS
- build_release_macos.sh: -T builds and runs tests; ORCA_TESTS_BUILD_ONLY
builds them without running (used by CI)
- scripts/run_unit_tests.sh: parameterized test dir and build config
Addresses #11273.
* ci: bump actions/checkout to v7 in reusable unit_tests workflow
Match the actions/checkout v6->v7 bump (#14517) that upstream applied to
the inline test job this reusable workflow replaces.
* feat: native Windows ARM64 build support
Builds on the merged DEPS_ARCH=arm64 plumbing (#13424) by adding the
dependency and source fixes needed for a green native ARM64 build on the
windows-11-arm runner. Validated end-to-end on Snapdragon X Elite hardware
(via a downstream fork using the same fixes); see OrcaSlicer/OrcaSlicer#8271
for the full writeup.
Dependencies:
- OpenEXR 2.5.5: ImfSimd.h hard-codes IMF_HAVE_SSE2 for any MSVC, pulling in
<emmintrin.h> (x86-only) -> C1189. Patch the header to require an x86 target
and force SSE cache vars off on ARM64.
- Boost.Context: use the winfib implementation on ARM64 (Windows Fiber API)
to avoid the armasm64 / CMake ASM_ARMASM linker-module bug, while keeping
the Boost::context target Boost.Asio needs.
- OpenCV: disable WITH_IPP on ARM64 (Intel IPP/IPP-ICV is x86/x64 only;
otherwise ~200 unresolved ippicv* externals at link).
- OpenSSL: use VC-WIN64-ARM on ARM64.
- FindGLEW: add an ARM64 arch branch.
Sources:
- clipper Int128.hpp: _mul128 is an x64-only intrinsic guarded by _WIN64
(true on ARM64); guard on _M_X64 and use the portable path.
- imgui imgui_widgets.cpp: fix va_start(vaList, &text) -> va_start(vaList, text)
(the &-form compiled on x64 but is invalid on ARM64).
- crash reporter: StackWalker.cpp gains an _M_ARM64 branch; BaseException.cpp
uses Cpsr instead of the x86-only EFlags on ARM64.
CI:
- New build_windows_arm64.yml on windows-11-arm: pins CMake 3.31.x, stages
ARM64 GMP/MPFR from MSYS2 clangarm64 (with llvm-dlltool import libs),
caches deps with a fixed-depth hashFiles key, builds and uploads the binary.
OCCT/STEP, SVG-to-3D and text emboss all build and work on ARM64 (no stubs
needed). Full feature parity with x64.
* fix(ci): use forward-slash DESTDIR to avoid CMake '\a' escape error
deps configure failed at GMP/GMP.cmake: "Invalid character escape '\a'"
because DESTDIR carried Windows backslashes (C:\a\...) and is re-parsed
when re-set with the /usr/local suffix. Pass DESTDIR (and the slicer's
DEPS prefix) with forward slashes via %CD:\=/%.
* fix(ci): don't export DESTDIR env var (CMake staged-install doubles paths)
Setting a DESTDIR *environment* variable made CMake treat it as the staged
install prefix and prepend it to every dependency's install path, so e.g.
FreeType installed to <DESTDIR>/a/.../OrcaSlicer_dep/usr/local and OCCT
then couldn't find its headers. Compute the forward-slash path into a
differently-named var (ORCA_DESTDIR) and pass it only via -DDESTDIR.
* ci(windows-arm64): fold ARM64 build into the standard Windows matrix
Replace the standalone build_windows_arm64.yml with a matrix entry on the
existing build_windows job, so x64 and ARM64 share one reusable workflow
chain (build_all -> build_check_cache -> build_deps -> build_orca), per
review feedback on #14059.
- build_all.yml: build_windows now matrices over {x64: windows-latest,
arm64: windows-11-arm} and threads `arch` through. Self-hosted runner
stays x64-only.
- build_check_cache.yml: cache key and dep-prefix path are now
architecture-specific on Windows (deps/build-arm64/OrcaSlicer_dep).
- build_release_vs.bat: accept an `arm64` argument (mirrors
build_release_vs2022.bat) -> uses `-A ARM64` and the build-arm64 tree.
The top-level CMake auto-derives CMAKE_PREFIX_PATH from the build dir,
so no explicit prefix is needed.
- build_deps.yml / build_orca.yml: gate the ARM64-only prep behind
`inputs.arch == 'arm64'` -- pin CMake 3.31.x, and stage MSYS2
clangarm64 GMP/MPFR import libs. NSIS installer/PDB/profile_validator
remain x64-only; ARM64 ships the portable zip. Artifact names get an
arch suffix to avoid collisions between the two Windows jobs.
https://claude.ai/code/session_0164c7ZhCLsYBmCiVN9pWDjK
* ci(temp): generate GMP/MPFR win-arm64 blobs to commit to repo
* feat(deps): add prebuilt GMP/MPFR win-arm64 blobs
The repo ships prebuilt GMP/MPFR import libs + DLLs for win-x64 and
win-x86; the Windows ARM64 build path copies from win-${DEPS_ARCH}
(CMakeLists.txt) but the win-arm64 blobs were missing, so the slicer
configure failed at "file COPY cannot find .../win-arm64/libgmp-10.dll".
Add win-arm64 libgmp-10.{dll,lib} and libmpfr-4.{dll,lib}, generated from
the MSYS2 clangarm64 gmp/mpfr packages with MSVC-compatible import libs via
llvm-dlltool. Headers are shared across arches and unchanged.
* simplify OpenEXR.cmake
* set default arch
* support msix
* ship installer
* try to fix webview2runtime issue
---------
Co-authored-by: Adam Behrman <adam.behrman@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Adam Behrman <abehrman@users.noreply.github.com>
Build the Linux AppImage for ARM64 (aarch64) alongside x86_64: the Linux CI
job now matrixes over both architectures, with arch-aware deps caching and
artifact/asset names (amd64 keeps its existing names). The aarch64 AppImage is
published to the nightly and release pages like the x86_64 one.
Run the unit-test suite on the aarch64 runner (faster GitHub arm runner); the
tests are built on that leg. Self-hosted keeps tests on the amd64 server.
* docs: add MSIX Store build design spec
* docs: update MSIX spec (PFN deep link, .drc, Associate tab) and add implementation plan
* ci: add MSIX logo asset generator and generated assets
* ci: fix MSIX asset rendering edge bleed (PixelOffsetMode) and make output order deterministic
* ci: add MSIX AppxManifest template
* ci: add MSIX packaging script
* ci: make build_msix.ps1 stage-only exit dot-source safe
* ci: build MSIX Store package in Windows job
* ci: run MSIX pack after existing Windows uploads and keep it out of release downloads
* feat: add MSIX packaged-context detection helpers
* fix: resolve MSIX package APIs dynamically to keep Win7 loadable
* feat: suppress self-update in MSIX Store build
* feat: suppress runtime file associations in MSIX Store build
* feat: keep version check in MSIX build, point update dialog at the Store
The update check is notification-only (OrcaSlicer never auto-downloads),
so the Store build keeps checking for new versions instead of skipping
the check. What changes when packaged is the new-version dialog: the
Download button is hidden, the info text asks the user to update from
the Microsoft Store, and the hyperlink / wxID_YES action opens the Store
product page instead of the GitHub release page.
* docs: align spec verification plan with Store-redirect updater behavior
* feat: default MSIX identity to the reserved Partner Center values
* feat: render MSIX logos full-bleed from the gradient-circle SVG
* feat: point update dialog Download button at the Store in MSIX builds
* feat: link Associate tab to Windows Default Apps settings in MSIX builds
* docs: align spec with review-driven logo, dialog and Associate-tab changes
* clearn up
- add a new mode to build in docker
- Improve AppImage
1. fix libbz2 soname differeence issue on appimage
2. Downgrade to 22.04 for better compatibility
3. improve appimage overall
Bump GitHub Actions action versions
Update workflow action versions: upgrade geekyeggo/delete-artifact from v5 to v6 and microsoft/setup-msbuild from v2 to v3 across .github/workflows/build_all.yml, build_deps.yml, and build_orca.yml. This brings workflows to newer action releases for compatibility and fixes.
* Reapply "Switch to self hosted mac runner (#12024)" with proper fix
This reverts commit 783f9926e3.
* use conditional logic for self-hosted runners
* improve readbility
# Description
<!--
> Please provide a summary of the changes made in this PR. Include details such as:
> * What issue does this PR address or fix?
> * What new features or enhancements does this PR introduce?
> * Are there any breaking changes or dependencies that need to be considered?
-->
# Screenshots/Recordings/Graphs
<!--
> Please attach relevant screenshots to showcase the UI changes.
> Please attach images that can help explain the changes.
-->
## Tests
<!--
> Please describe the tests that you have conducted to verify the changes made in this PR.
-->
