Drops the GraalPy 3.11 / GraalPy 24.2 builds (gp311_242) as agreed in
pypa/cibuildwheel#2741: a ~6-month overlap after the GraalPy 25 (gp312)
release. GraalPy 3.12 (gp312_250) stays.
This also removes the GraalPy 24-only workarounds that were explicitly
marked "Remove when GraalPy 24.x is dropped":
- the Visual Studio compiler-discovery (vswhere) and build-isolation
(graalpython#491) workarounds in the Windows backend
- the issue-491 Windows skip/filter hacks in test_pep518 and
test_dependency_versions (added in the original GraalPy PR #1538 and
unneeded since GraalPy 25 per #2597)
The uv-doesn't-support-graalpy skip (#2754), the `graalpy` enable group,
and the generated constraints pins apply to all GraalPy and are kept.
Assisted-by: ClaudeCode:claude-opus-4.8
* chore: use ConfigurationError when package_dir is outside cwd
Replace a generic Exception with errors. ConfigurationError when package_dir is not inside the working directory. This makes the error type consistent with the project's error handling and allows the CLI to treat this as a fatal error with the expected exit behavior.
* test: merge linux package_dir test into linux_build_steps_test
Fold the standalone linux_build_test.py into the existing
linux_build_steps_test.py, which already covers the linux platform
build orchestration. Reuses the file's existing import boilerplate and
module-qualified build() call style.
Assisted-by: ClaudeCode:claude-opus-4.8
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---------
Co-authored-by: Henry Schreiner <henryfs@princeton.edu>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Replace the explicit if/else used to resolve pinned container images with pinned_images.get(config_value, config_value) in the manylinux and musllinux loops inside _compute_build_options. This keeps behavior identical while reducing duplication and simplifying the code in cibuildwheel/options.py.
* ci: improve Azure test reliability
Azure has intermittent failures, often the macOS framework-Python
`posix_spawn: Undefined error: 0` spawn race under load, plus transient
download flakiness. Address both:
- Retry integration tests via pytest-rerunfailures (already required):
add `--reruns=2 --reruns-delay=5` to the serial and non-serial runs so a
single transient flake self-heals instead of failing the whole job.
- Cap Azure at `--num-processes 3` to reduce concurrent process spawning
(the macOS runner was using 4 xdist workers), lowering the spawn-race
probability and peak resource pressure.
- Cache downloads across runs: relocate CIBW_CACHE_PATH to a stable path
and add a Cache@2 task (interpreter downloads + per-worker test pip
caches live under it), cutting the network-download flake surface.
- Bump the Azure host interpreter to 3.13.
Assisted-by: ClaudeCode:claude-opus-4.8
* ci: stabilize iOS tests, bound rerun cost
An iOS run on Azure (build 9043) hung for the full 40-min pytest timeout
on the second build config of test_ios_platforms, then reran twice
(reruns=2), blowing past the 180-min job cap with no useful signal. The
hang was a stuck simulator/xcodebuild on the second config, after the
first config had run.
- Shut down running simulators before each parametrized config via a
`clean_ios_simulators` fixture, so a simulator left booted/wedged by the
previous config isn't reused.
- Drop iOS reruns from 2 to 1. One retry still covers the documented
"fails the first time" simulator-boot flake, while halving the
worst-case retry cost (3x40=120min -> 2x40=80min) on a hang.
Assisted-by: ClaudeCode:claude-opus-4.8
* ci: don't limit num processes
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* ci: rerun OCI unit tests that pull from Docker Hub
test_local_image et al. occasionally fail on a transient Docker Hub
anonymous-pull blip ("unauthorized: authentication required") when
pulling debian:trixie-slim. The integration runs already self-heal via
pytest-rerunfailures, but the unit run had no reruns, so a single flake
failed the whole job. Mark the three network-dependent OCI tests
(test_local_image, test_enter_error, test_multiarch_image) flaky so they
retry instead of blanket-rerunning the suite.
Assisted-by: ClaudeCode:claude-opus-4.8
* Drop iOS clean step
Co-authored-by: Henry Schreiner <HenrySchreinerIII@gmail.com>
* ci: scope integration reruns to test_0_basic::test
Drop the blanket --reruns from the serial and non-serial integration
runs and instead mark test_0_basic.py::test as flaky, so only the known
flaky basic test is retried.
Assisted-by: ClaudeCode:claude-opus-4.8
---------
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
`OCIContainer.__enter__` was hardened in #2879 to always clean up on
failure, but `__exit__` was left unguarded. If the container/bash had
already died, `bash_stdin.write("exit 0")` raises `BrokenPipeError`; if
bash refuses to exit, `process.wait(timeout=30)` raises `TimeoutExpired`.
Either propagated out of `__exit__` before `_remove_container()` ran,
leaking the container (and, on timeout, the `start` process too).
Wrap the teardown so a broken pipe or timeout instead forces the process
down (kill + wait) and always falls through to container removal, while
still respecting CIBW_DEBUG_KEEP_CONTAINER. Pipe closes are now also
guarded so a flush-on-close against a dead pipe can't mask cleanup.
Add non-docker unit tests for the clean-exit, already-dead-bash, and
shutdown-timeout paths.
Assisted-by: ClaudeCode:claude-opus-4.8
* fix: minor 4.0 issues in android validation, docs, and docstring
- android: raise a clear FatalError when ANDROID_API_LEVEL is not an
integer, instead of an uncaught ValueError deep in localized_vars
- docs: note that delvewheel is the default Windows repair-wheel-command
since 4.0 and how to skip it for platform-tagged wheels with no
extension module; update the now-outdated FAQ wording
- venv: fix _parse_pip_constraint_for_virtualenv docstring to say
marker-bearing constraints are evaluated against the host's default
environment (not skipped) when marker_env is None
Assisted-by: ClaudeCode:claude-opus-4.8
* fix: use int() with original error for ANDROID_API_LEVEL validation
Handle ANDROID_API_LEVEL validation via try/except around int() so the
original exception message is surfaced, and so non-ASCII Unicode digits
(which isdigit() accepts but int() may reject) are handled correctly.
Assisted-by: ClaudeCode:claude-opus-4.8
* refactor: address reivew by moving error catch
Assisted-by: ClaudeCode:claude-opus-4.8
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
---------
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: detect musl libc when downloading python-build-standalone
`platform.libc_ver()` only ever reports glibc and returns `("", "")` on
musl systems, so `"musl" in platform.libc_ver()` was never true and musl
hosts (e.g. Alpine) always selected the `gnu` python-build-standalone
asset, which won't run.
Detect musl by shelling out to `ldd --version`, which prints "musl libc"
on musl-based systems. This probes the actual system loader rather than
the host Python's linkage, so it stays correct even when cibuildwheel
runs under a relocatable/statically-linked interpreter. Falls back to
glibc when `ldd` is unavailable.
Assisted-by: ClaudeCode:claude-opus-4.8
* test: add type annotations to satisfy mypy and ruff
Assisted-by: ClaudeCode:claude-opus-4.8
The linux32 sanity check added in #2880 omitted `*platform_args`, unlike
the surrounding `uname -m` probes and the `create` command. On the i386
fallback path `platform_args` is reassigned to AMD64, so without
forwarding it the check could run against a different image variant than
the one actually built (e.g. on a multi-platform/containerd backend),
making the check meaningless or pulling the wrong variant.
Assisted-by: ClaudeCode:claude-opus-4.8
* Try to fix the problem with py trying to execute the shebang
* run whenever the branch name starts with gitlab
* Modernise GitLab CI config
- Windows: remove stale servercore:1809 image and choco install
(runners are now Windows 2022 VMs with Python 3.13 pre-installed)
- macOS: update deprecated macos-14-xcode-15 to macos-15-xcode-16
- Linux: bump Python image from 3.12 to 3.13
- All jobs: also run on gitlab* branches with CIBW_ENABLE=all
Generated with the help of AI (Amp/Claude Opus 4.7)
Amp-Thread-ID: https://ampcode.com/threads/T-019e6496-f89a-74e7-9f6b-655574290e68
Co-authored-by: Amp <amp@ampcode.com>
* ci: exclude graalpy on GitLab Windows runners
GraalPy is JVM-based and slow to start; on the small (2-core) GitLab
SaaS Windows runners, virtualenv's interpreter query for graalpy.exe
times out, failing test_0_basic::test[pip]. Restrict the Windows job's
CIBW_ENABLE to all groups except graalpy (other platforms keep "all").
CIBW_ENABLE rather than CIBW_SKIP is used so the test's expected-wheel
set (derived from CIBW_ENABLE) stays consistent with what is built.
Assisted-by: ClaudeCode:claude-opus-4.8
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
---------
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
Co-authored-by: Amp <amp@ampcode.com>
Co-authored-by: Henry Schreiner <henryfs@princeton.edu>
This was the case when running `nox -s tests -- --run-podman -v -s 'unit_test/oci_container_test.py::test_multiarch_image[podman-OCIPlatform.i386]'` on macOS arm64.
* fix: ensure clean-up on container start failure and warn on removal failure
Properly release resources on container start failure.
If we fail to remove the created container, warn when not running in CI.
* Potential fix for pull request finding
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
* chore: add unit tests for `OCIContainer._get_platform_args`
While this was already tested indirectly through `test_local_image`, this makes the behaviour check of `OCIContainer._get_platform_args` clear.
* fix: re-add `check=False` removed at some point in local testing
* feat: add SHA256 verification for direct downloads
Store SHA256 hashes when running update scripts and verify them
when downloading files at build time. This improves security by
detecting unexpected changes to downloaded artifacts.
Platforms covered: macOS (CPython, PyPy, GraalPy), iOS, Android,
virtualenv, and python-build-standalone. Windows (nuget) and
Linux (Docker) are excluded.
SHA256 sources per platform:
- macOS/iOS/Android CPython (python.org): sha256_sum from API
- GraalPy: .sha256 sidecar assets from GitHub releases
- python-build-standalone: SHA256SUMS file in release
- PyPy, BeeWare iOS, Maven (Chaquopy): stream-download and compute
Changes:
- cibuildwheel/util/file.py: add sha256 param to download()
- cibuildwheel/platforms/{macos,ios,android}.py: add sha256 to
PythonConfiguration and pass to download()
- cibuildwheel/venv.py: read sha256 from toml and pass to download()
- cibuildwheel/util/python_build_standalone.py: add sha256 to
PythonBuildStandaloneAsset and pass to download()
- cibuildwheel/resources/build-platforms.toml: add sha256 fields
- cibuildwheel/resources/virtualenv.toml: add sha256 field
- cibuildwheel/resources/python-build-standalone-releases.json: add sha256
- bin/update_pythons.py: compute/store sha256 per source strategy
- bin/update_virtualenv.py: compute sha256 by streaming download
- bin/update_python_build_standalone.py: parse SHA256SUMS file
Closes#908
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Assisted-by: copilot-cli:claude-sonnet-4.6
* fix: populate sha256 in resource files and fix Windows PythonConfiguration
- Add sha256 field to Windows PythonConfiguration (PyPy/GraalPy have
direct download URLs on Windows too)
- Pass sha256 to install_pypy() and install_graalpy() in windows.py
- Fix update_pythons.py: handle empty sha256 from CPython API (older
versions) by streaming download to compute it; fix condition to
check 'not sha256' rather than 'not in dict'
- Fix update_virtualenv.py: compute sha256 even when version unchanged
but sha256 is empty (first-time population)
- Fix update_python_build_standalone.py: resolve file path relative to
the script itself (not the installed package) so writes go to source
checkout, not the uv cache
- Populate actual sha256 values by running all three update scripts
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Assisted-by: copilot-cli:claude-sonnet-4.6
* fix: also include pyodide
Assisted-by: CopilotCLI:gpt-5.3-codex
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: PR review comments for cache verification and docs wording
Co-authored-by: henryiii <4616906+henryiii@users.noreply.github.com>
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: require sha256 for download configs
Require sha256 for URL-backed Python and virtualenv download configs. Update the GraalPy updater to refresh macOS x86_64 entries by selecting the latest release that still has a matching asset, and fill the two missing GraalPy checksums in build-platforms.toml.
Assisted-by: CopilotCLI:gpt-5.4
* ci: remove unit test for bin item
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* refactor: combine sha256 unit tests into test_sha256.py
Merge pyodide_test.py and python_build_standalone_test.py into a
single unit_test/test_sha256.py since both test sha256-related
behaviour.
Assisted-by: opencode:glm-5
---------
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: henryiii <4616906+henryiii@users.noreply.github.com>
Since delvewheel is now the default repair-wheel-command on Windows,
update all places where auditwheel/delocate were referenced but
delvewheel was missing:
- diagram.html: add delvewheel repair block for Windows (was grouped
with ios/pyodide as optional dot)
- generate_schema.py: add Windows delvewheel default to the platform
loop so schema.json includes the default
- schema.json: regenerated with Windows delvewheel default
- options.md: add Windows to defaults list, remove outdated tip about
delvewheel being early-stage/optional, update examples
- contributing.md: add delvewheel to the tool list
Assisted-by: OpenCode:glm-5.1
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* feat: add CIBUILDWHEEL_BUILD_IDENTIFIER environment variable
Set `CIBUILDWHEEL_BUILD_IDENTIFIER` to the current build identifier
(e.g. `cp311-manylinux_x86_64`) in the environment for all per-build
steps: `before_build`, the build itself, `repair_command`,
`before_test`, and `test_command`.
This allows scripts and commands to inspect which build is currently
running, which is useful for e.g. writing per-build output files:
CIBW_TEST_COMMAND='pytest --junit-xml=results-$CIBUILDWHEEL_BUILD_IDENTIFIER.xml'
The variable is set after the user's environment overrides are applied
and is only available for per-build steps (not `before_all`, where no
single identifier applies). All six platforms are covered: linux, macOS,
Windows, pyodide, Android, and iOS.
Closes#944, closes#2750
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Assisted-by: OpenCode:claude-sonnet-4.6
* fix: address review comment moving this for Android
Assisted-by: OpenCode:glm-5.1
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
---------
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Use auditwheel on Android
* Add auditwheel command to defaults
* pkgconfig fixes
* Pre-import ctypes before monkey-patching in _cross_venv
* Set PKG_CONFIG and PKG_CONFIG_RELOCATE_PATHS variables
* Initial attempt at using mzakharo/android-gfortran
* Switch to using termux/ndk-toolchain-clang-with-flang
* Move PKG_CONFIG variables from build_env to android_env
* Simplify flang installation
* Add cross build files for NumPy
* Add ldpaths entry for libomp
* Add `--rm` to docker command line
* Make Rust and Fortran shims consistent
* Update documentation
* Default to API level 24 on all Python versions
* Clarify comments
* Cleanups
* Fix tests:
* Set up Android env after installing pkgconf
* Add tests for successfully using an older API level
* Previous commit's auditwheel failure is fixed in the auditwheel PR
* Update how-it-works diagram
* Add tests for repair errors
* Add more repair tests
* Add test for Meson and Fortran
* Add test for cross build files
* Improve test_api_level error message
* Add xbuild-files option
* use pypa/auditwheel@main
* Remove dependencies which are no longer needed
* Fix README
* Update to auditwheel 6.7.0
* Fix compatibility with pkgconf 2.5.1.post2
* Documentation clarifications
Co-authored-by: Joe Rickerby <joerick@mac.com>
* Fortran shim improvements
* Add Jinja variables to new_meson_project
* Update run_example_ci_configs for changed new_meson_project signature
* Add missing dependency to run_example_ci_configs
---------
Co-authored-by: mayeut <mayeut@users.noreply.github.com>
Co-authored-by: Joe Rickerby <joerick@mac.com>
* Update dependencies
* Revert pip marker flattening for GraalPy 3.11 on Windows
* Revert pip marker flattening for GraalPy 3.12 on Windows
---------
Co-authored-by: cibuildwheel-bot[bot] <83877280+cibuildwheel-bot[bot]@users.noreply.github.com>
Co-authored-by: mayeut <mayeut@users.noreply.github.com>
* feat: add CPython 3.15 support for iOS
Assisted-by: OpenCode:Kimi-K2.6
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: copy in iOS support files
Assisted-by: Copilot:claude-sonnet-4.6
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: skip directories when copying iOS support files
When copying multiarch-specific support files (e.g. _cross_arm64_iphoneos.py),
the code was trying to copy all items in the directory including __pycache__
directories. The shutil.copy() function only works with files, not directories,
which caused an IsADirectoryError.
This fix adds a check to only copy files, skipping any directories like
__pycache__ that may have been created by Python imports.
Assisted-by: Copilot:claude-haiku-4.5
* fix: - in dir fine for now
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: apply review suggestions
Assisted-by: Copilot:claude-sonnet-4.6
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: only do this on 3.15+
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* Ensure target Python directory exists before build
Add assertion to check if target Python directory exists.
* fix(ios): use stdlib dir directly instead of copying to platform-config
Remove the _inject_support_files workaround for python.org 3.15+
distributions. Instead of copying sysconfig files from the stdlib into
a synthetic platform-config/ directory, pass the stdlib directory
directly to make_cross_venv.py.
make_cross_venv.py is updated to derive the multiarch tag from the
_sysconfigdata_ filename rather than assuming it comes from the
directory name.
Co-authored-by: Russell Keith-Magee <russell@keith-magee.com>
Assisted-by: OpenCode:Kimi-K2.6
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* Minor format cleanups on pre-commit skipped files.
* Clarify path names in make_cross_venv script.
---------
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
Co-authored-by: Russell Keith-Magee <russell@keith-magee.com>
* chore: Add agent and copilot setup files
Add AGENTS.md with project-specific developer notes for AI agents.
Add GitHub Actions workflow to validate Copilot setup steps.
Add changelog-entry skill for automated changelog generation.
Assisted-by: OpenCode:kimi
* Update .gitignore to include CLAUDE.md
Add CLAUDE.md to .gitignore for symlink instructions
Assert statements are skipped under python -O, making them unsuitable
for runtime validation. Replace assert with an explicit check that
raises PythonBuildStandaloneError if the target directory already exists.
Assisted-by: OpenCode:glm-5
* chore: minor fixups across errors, oci_container, and options
- Strengthen error message wording: 'is expected to' → 'must' and
'is expected to place one' → 'must place exactly one' in
FailedWheelRepairError classes.
- Remove shell=True from subprocess.run in oci_container debug_info,
passing command as a list instead of a string.
- Update pinned image error message from 'cibuildwheel 3.x' to
'cibuildwheel 4.x' and add comment about next warning candidate.
Assisted-by: OpenCode:glm-5
* Update cibuildwheel/options.py
* feat: add CPython 3.15 support for iOS and Android
Add cp315 identifiers to build-platforms.toml for both iOS (arm64_iphoneos,
x86_64_iphonesimulator, arm64_iphonesimulator) and Android (arm64_v8a, x86_64).
Update bin/update_pythons.py to route iOS 3.15+ to python.org instead of
BeeWare GitHub releases. Update README.md platform table to show Android and
iOS as supported for 3.15. Add cp315-cp315 to expected_wheels test defaults
for android and ios platforms.
Assisted-by: OpenCode:Kimi-K2.6
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: adjust expectations
All changes made and linting passes. Here's a summary of what was done:
1. test/utils.py:390-394 — Added "cp315-cp315": 24 to the Android API level mapping, fixing the KeyError: 'cp315-cp315' in test_android.py::test_expected_wheels.
2. cibuildwheel/resources/build-platforms.toml:247 — Removed the cp315-ios_x86_64_iphonesimulator entry since CPython 3.15 dropped Intel iOS simulator support.
3. test/utils.py:397-405 — Updated the iOS expected wheels logic to return an empty platform_tags list for x86_64 when the Python version is 3.15+, so no x86_64_iphonesimulator wheels are expected for cp315+ on Intel Macs.
The XCframework slice name (ios-arm64_x86_64-simulator) stays the same in ios.py — I verified the 3.15 tarball still uses that name. The target_python.exists() iOS ARM64 failure on the feature branch is a separate issue (the 3.15 python.org distribution has a different structure than the beeware distributions used for 3.13/3.14) — left alone per your request.
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
Assisted-by: OpenCode:glm-5.1
* revert: x86 is still supposed to exist
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* docs: fix readme for Pyodide
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* docs: mention eol
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: drop assert
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* feat: add CPython 3.15 support for Android only
Cherry-picked all changes from henryiii/feat/cp315-ios-android, then
reverted iOS-specific parts (iOS 3.15 build configs, ios.py assert
removal, update_pythons.py iOS URL logic, iOS test expectations, and
README iOS column for 3.15). Android cp315 entries, test expectations,
and README Android column remain.
Assisted-by: OpenCode:glm-5
---------
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
test: add Python 3.15 to CI test matrix and classifiers
Update the Max Python CI job from 3.14 to 3.15 and add the
Programming Language :: Python :: 3.15 classifier to pyproject.toml.
Assisted-by: OpenCode:GLM-5
* WIP - initial punt at audit command
* Add `abi3audit` as a dependency
* Add helper functions to check stable ABI wheels
* Run `abi3audit` for macOS and Windows wheels
* Copy out of container for repairing?
* Add some notes that `cibuildwheel` runs `abi3audit`
* Add basic unit tests
* Add a basic C extension with `Py_LIMITED_API`
* Add a test project that violates Stable ABI
* Fix linux test
* Skip abi3 wheel tests for Pyodide
* Patch the correct subprocess module
* wrap cleanup of abi3audit dir
* Write the docs for the new options
* Move to above testing in docs
* Implement audit-requires and audit-command
* Some cleanups after self-review
* Add default value
* fix type errors
* the key is `audit-command`, not `audit`
* Add a variety of tests for audit requires options
* Add `test_audit_requires` similar to `test_test_requires`
* Add some configurability-related audit tests
* Fix parsing error with options docs leaving out commands
* Better way to extract version (maybe helps Pyodide?)
* Fix a case of unbound `use_uv`
* Standardise: rename to `abi3_wheel`
* Fix audit command run message
* Simplify custom audit command a bit
* Remove unnecessary skip for Pyodide
* Pyodide should have no default audit command
* More accurate skip messages for Pyodide skips
* Wheels are audited after they are repaired
* Regenerate constraints to include `abi3audit`
* Fix typos
* Some attempts for Windows fixes
* Check `pyvenv.cfg` instead of directory existence
* Add validation for lack of wheel placeholders
* Try yet another Windows `uv` fix
* Regenerate diagram and re-trigger Azure CI
* Add missing `import sys` for abi3 C extension tests
* Remove audit-command at the global level
* Clarify `abi3audit` pinning a little bit
* Regen constraints
* Discard changes to cibuildwheel/resources/constraints-pyodide312.txt
* Discard changes to cibuildwheel/resources/constraints-pyodide313.txt
* try opt-in uv again
* fix issue on windows on Python 3.13 related to nested venvs
On win / python 3.13, virtualenv creates a venv where the 'home'
points back to the venv that sys.executable was running in, rather
than the root install. that seemingly leads to problems with package
resolution, where pip.exe couldn't find the pip python package.
this appears to fix it!
* Update constraints
* chore: revert python-discovery bump
Assisted-by: OpenCode:glm-5.1
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* fix: restore workaround for graalpy
Assisted-by: OpenCode:glm-5.1
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
---------
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
Co-authored-by: Agriya Khetarpal <74401230+agriyakhetarpal@users.noreply.github.com>
Co-authored-by: Henry Schreiner <henryfs@princeton.edu>
* Add a default repair wheel command for Windows
* Install `delvewheel` into Windows build tools
* Add delvewheel to constraints file
* Regenerate constraints
* Docs
* Suggest how to disable Windows wheel repair
* Add tests
* Run `delvewheel` with verbose mode as default
* Add note about telling delvewheel where to look
Suggested-by: Nils <nilsnolde@proton.me>
* Partially revert 81374bb8fc43acefed76cb13d761a9e6cf6afa58
* Add `pip` and `uv` cases for build installations
* Ignore `python-native.dll` for GraalPy
* Ignore MSVC DLLs, try Windows amd64 graalpy happy
* Fix last remaining Windows test failure (famous last words?)
* Update constraints
* Drop workaround unneeded with delvewheel v1.12.1
* Update docs/options.md
Co-authored-by: Aohan Dang <adang1345@gmail.com>
* Drop `test_delvewheel_default_on_windows`
* Partially revert "Update constraints"
This partially reverts commit 2dc4c0f4951858959e213463d90431a5ca03b96f.
* Add a test case when repair command is `""`
* Add back test that checks delvewheel run/disable
* Add a C project with a missing DLL
* Add tests for `missing_dll_project`
* Partially undo virtualenv/python-discovery update
* Fix test
* Discard changes to cibuildwheel/resources/constraints-pyodide312.txt
* Discard changes to cibuildwheel/resources/constraints-pyodide314.txt
* Fix bad merge of constraints
* Discard changes to cibuildwheel/resources/constraints-pyodide313.txt
* Manually add more GraalPy pip markers
* Dependency parsing was too naïve
* Maybe a better way to invoke the compiler on Windows
* Can we get away without a DLL?
---------
Co-authored-by: Aohan Dang <adang1345@gmail.com>
* Update dependencies
* fix: revert pip upgrade on graalpy for now
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* fix: older pip for graalpy on Windows
Assisted-by: OpenCode:glm-5.1
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
---------
Signed-off-by: Henry Schreiner <henryfs@princeton.edu>
Co-authored-by: cibuildwheel-bot[bot] <83877280+cibuildwheel-bot[bot]@users.noreply.github.com>
Co-authored-by: Henry Schreiner <henryfs@princeton.edu>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Henry Schreiner