Keep the CI matplotlib pick compatible with the numpy pick - #621
Merged
Conversation
eriknw
force-pushed
the
40-ci-matplotlib-numpy-pick
branch
from
August 5, 2026 17:44
8516745 to
8a67fd9
Compare
eriknw
force-pushed
the
40-ci-matplotlib-numpy-pick
branch
from
August 5, 2026 18:03
8a67fd9 to
b773d37
Compare
eriknw
force-pushed
the
40-ci-matplotlib-numpy-pick
branch
from
August 5, 2026 18:05
b773d37 to
b7ac2a9
Compare
The overnight matrix produced the stack's first two post-restack failures, both the same one line: the notebooks task on macos-latest and windows-latest picked numpy 1.24 while installing matplotlib unpinned, and matplotlib 3.11.0's runtime check requires numpy >=1.25. Its conda-forge metadata does not carry that floor, so the solver pairs them happily and the notebooks die at import. Same class and same fix shape as the networkx and awkward pick constraints: the version picker now emits mplver, a bare ceiling of <3.11 whenever the numpy pick sits under 1.25 and empty otherwise, the notebooks install line consumes it, and validate() checks the pairing. The --validate stress test passes 160000 combos with zero failures, and both emission cases are confirmed by running the picker.
eriknw
force-pushed
the
40-ci-matplotlib-numpy-pick
branch
from
August 6, 2026 07:57
b7ac2a9 to
cd0be5f
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Stacked on #620. The overnight matrix produced the post-restack wave's first two failures, both identical:
build_and_test (macos-latest, notebooks)on33-udf-ret-dtypeandbuild_and_test (windows-latest, notebooks)on27-check-mask-movefailed atimport matplotlibwithThe random dependency picker chose numpy 1.24 while the notebooks task installs matplotlib unpinned; matplotlib 3.11.0's runtime check requires numpy >=1.25 but its conda-forge metadata does not carry that floor, so the solver pairs them and the job dies at import. Not a code bug in either PR; any rung can draw this combination in the notebooks lane.
Same class and fix shape as the existing networkx and awkward pick constraints (
3a065ba9,8c741ce9):ci_pick_versions.pynow emitsmplver(a bare<3.11ceiling when the numpy pick sits under 1.25, empty otherwise), the notebooks install line consumes it, andvalidate()checks the pairing.--validatepasses 160,000 combos with zero failures; both emission cases confirmed by running the picker.Gates: full pinned suite 1114 passed / 145 skipped; pre-commit all hooks pass.