Skip to content

transformations: Add version 2026.8.8 - #3221

Merged
luhenry merged 2 commits into
mainfrom
transformations
Oct 22, 2026
Merged

transformations: Add version 2026.8.8#3221
luhenry merged 2 commits into
mainfrom
transformations

Conversation

@luhenry

@luhenry luhenry commented Oct 14, 2026 •
edited
Loading

Copy link
Copy Markdown
Member

Compiles the single-file _transformations C extension (limited API) against numpy. Upstream publishes no riscv64 wheel.

Mirrors upstream's wheel.yml.

Differs from upstream

  • Nothing beyond the riscv64 image.

Matrix: cp312-abi3 (built on cp312, retested on cp313/cp314) and cp314t, matching upstream's cp312-* cp314t-*; cp315t skipped, no such interpreter exists yet.

Testing

  • same as upstream

License: OK

Local riscv64 (QEMU) cp312-abi3 build tested on cp313: 45 passed.

@luhenry
transformations: Add version 2026.8.8
48a85d7 
Build Christoph Gohlke's transformations C extension for riscv64,
mirroring upstream's cibuildwheel wheel.yml: one cp312-abi3 wheel
(retested on cp313/cp314) plus a free-threaded cp314t wheel, tested
with the package's own doctests.
luhenry added a commit that referenced this pull request Oct 14, 2026
@luhenry
queue: transformations -> ci-running (PR #3221, 2026.8.8), hydride ->…
8839237 
… ci-running (PR #3225, 1.2.3), gtsam-develop -> parked (nightly, gotcha 581; port gtsam instead)
@luhenry Claude

luhenry commented Oct 14, 2026 •
edited
Loading

Copy link
Copy Markdown
Member Author

check_commit_messages (job 115844531951) failed on check_no_ai_attribution, flagging a "Generated by Claude Code" footer in the PR description. Not a real issue with the current PR: create_pull_request auto-injects that footer, and it was stripped via update_pull_request right after creation — the live PR body has never had it since. The failed run just evaluated the pull_request "opened" webhook's frozen body snapshot (captured before the strip), which a GitHub Actions re-run can't refresh since re-runs reuse the original event payload.

The edit itself fired a separate pull_request.edited run (workflow run 38561845542) carrying the corrected body; once its check_commit_messages job executes (it's still queued behind the current backlog), it should pass and reflect the PR's actual state. No fix to push here — watching for that run to clear.

This was referenced Oct 14, 2026
Merged
Merged
Merged
Merged
Merged
Merged
Merged
Merged
@luhenry
luhenry marked this pull request as draft October 15, 2026 12:14
@luhenry
transformations: Install numpy from our registry only
df883d0 
PyPI's numpy 2.5.4 has no riscv64 wheel, so pip compiled it from sdist
in the build env and again for each of the cp313 and cp314 abi3 retests,
about 18 minutes apiece, and the cp312 job hit its 60-minute timeout.
Our registry carries numpy 2.5.3 for cp312, cp313, cp314 and cp314t.
@luhenry
luhenry marked this pull request as ready for review October 22, 2026 00:19
@luhenry
luhenry merged commit 85c5ab1 into main Oct 22, 2026
7 checks passed
@luhenry
luhenry deleted the transformations branch October 22, 2026 01:29
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

See More

Successfully merging this pull request may close these issues.

1 participant

@luhenry