Repository navigation
Conversation
…ts.matrix Converts build-pot, build-pyvrp, build-regopy, build-scikit-learn, build-snowflake-connector-python, build-statsforecast, build-transformations and build-zuvloop from hardcoded python/libc matrices plus a layered include: override to the data-driven needs.setup.outputs.matrix, same as #4914/#4915. Each per-python or per-libc override (test_skip, test_requires additions, before_all, before_test, pytest args/dist, build selectors) is preserved as an inline GitHub Actions ternary at its point of use instead of a matrix field, since the generated matrix entries do not carry those fields. Several packages' hardcoded python lists diverged from what docs/packages/<pkg>.yaml actually declares; python-exclude is added to setup: where needed to keep excluding tags the old list always skipped beyond the default cp39.
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Same mechanical fix as #4914/#4915, applied to the 8 workflows in this batch that also layer a per-python or per-libc
include:override on top of the hardcoded matrix. Each hardcodedpython:/libc:array plusinclude:block is replaced withmatrix: { include: ${{ fromJSON(needs.setup.outputs.matrix) }} },matrix.libcis renamed tomatrix.platform, and every overridden field is converted to an inline ternary at its point of use (the generated entries carry no such fields).Packages and how each override was preserved
pot:extra_deselect(cp314t only) ->matrix.python == 'cp314t' && '--deselect ...' || ''inCIBW_TEST_COMMAND.pyvrp:pytest-args(cp311 only) -> same ternary pattern inCIBW_TEST_COMMAND.regopy:before_allkeyed onlibc->matrix.platform == 'manylinux' && 'dnf ...' || 'apk ...'inCIBW_BEFORE_ALL.scikit-learn: two fields keyed onpython(pandas,loky_broken) ->contains(fromJSON('[...]'), matrix.python) && ... || ''inCIBW_TEST_REQUIRES/CIBW_TEST_ENVIRONMENT. Flagging this one for a careful look given its downstream impact.snowflake-connector-python:pytest_dist(cp314t vs. rest) -> ternary inCIBW_TEST_COMMAND.statsforecast:test_skip(cp314/cp314t) ->contains(...) && '*' || ''inCIBW_TEST_SKIP.transformations: non-standard shape -pythonwas only a grouping label for abuildfield spanning multiple literalCIBW_BUILDselectors (onecp312leg builds+retests cp312/cp313/cp314 abi3, onecp314tleg is free-threaded only). Preserved exactly for those two keys; any other interpreter the generated matrix picks up (currentlycp315t) falls back to a plain${python}-${platform}_riscv64single-selector build.zuvloop:hypothesispin (cp314t only) -> ternary inCIBW_TEST_REQUIRES.Mismatches/exclusions found while verifying against
uv run ci_scripts/pending_versions.py --python-matrix <pkg>pot,regopy,snowflake-connector-python,statsforecast: the hardcoded lists always skippedcp310/cp311beyond the defaultcp39; addedpython-exclude: 'cp39 cp310 cp311'tosetup:to keep that scope.scikit-learn: hardcoded list always skippedcp311; addedpython-exclude: 'cp39 cp311'.regopy,statsforecast: both hardcodedcp314tinto their matrix even thoughdocs/packages/<pkg>.yamldeclares no3.14tfiles entry for either. The generated matrix correctly drops that phantom leg - letting the structural fix resolve it rather than reintroducing it.scikit-learn: the fix newly picks upcp311(now excluded above) and, for 1.9.1,cp315/cp315t, which the stale hardcoded list never saw.pyvrp,zuvloop: neither had alibc/platform axis before (manylinux only, hardcoded), butdocs/packages/<pkg>.yamlalready declares pendingmusllinuxfile entries for every version, so the generated matrix now also attempts musllinux legs. These may not actually build (pyvrp's numpy/matplotlib/contourpy/kiwisolver/pillow pins and zuvloop's native extension may lack riscv64 musllinux wheels/toolchain support) - left for CI to show; eachpublish:job'sartifact-patternstill filters to*-manylinux_riscv64, so a musllinux leg failing or succeeding changes nothing about what gets published.transformations: the fix newly picks upcp315t(see above), previously invisible to the stale["cp312", "cp314t"]list.No other files touched.