Release 1.4.2: reproducible image builds - #28
Merged
Merged
Conversation
Rebuilding a released tag previously resolved whatever dependency stack was current: the base image was an unpinned python:3.11-slim and pyproject declared biopython, numpy, scipy and matplotlib without bounds. Backfilling the 0.3.0-1.4.1 images made that concrete - each one pairs its tagged source with an August 2026 stack rather than the original. The image build now pins the base by digest and installs under docker/constraints.txt, which records the exact resolved versions. Both the test and runtime stages use it, so what the tests validate is what ships. pyproject gains upper bounds only (biopython<2, numpy<3, scipy<2, matplotlib<4). Exact pins would be wrong here: alphajudge is installed alongside other scientific packages, and == constraints in a library's metadata cause resolver conflicts for consumers. Lower bounds are omitted because older versions are untested; the caps guard against the next breaking major, and reproducibility is handled by the constraints file. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Makes future release images reproducible, and cuts the first release that will exercise the
release:trigger merged in #27.Why
Rebuilding a released tag previously resolved whatever dependency stack was current:
python:3.11-slim(floating)sha256:90744cff…biopython,numpy,scipy,matplotlib— unboundeddocker/constraints.txtlock, exact versionsBackfilling the 0.3.0–1.4.1 images made this concrete: every one pairs its tagged source with an August 2026 dependency stack rather than the original. That is unavoidable retroactively, but it should not keep happening.
What changed
docker/constraints.txtrecords the exact resolved versions (numpy 2.4.6, scipy 1.17.1, biopython 1.88, matplotlib 3.11.1 and transitives), applied in both the test and runtime stages so what the tests validate is what ships.pyproject.tomlgains upper bounds only:biopython<2,numpy<3,scipy<2,matplotlib<4.On not exact-pinning pyproject
Deliberate.
alphajudgeis installed alongside other scientific packages, so==constraints in a library's metadata cause resolver conflicts for consumers. Reproducibility belongs in the image, which is what the constraints file provides. Lower bounds are omitted because older versions are untested — assertingnumpy>=1.24would be a guess. The caps guard against the next breaking major.Validation
The
dockerjob builds the test stage on this PR, runningpytestinside the image against the pinned stack.🤖 Generated with Claude Code