Repository navigation
ci: install gating jobs from a constraints file exported from uv.lock - #194
Merged
Merged
Conversation
CI resolved dependencies fresh on every run, so an upstream release could turn main red overnight (numpy 2.5 broke median imputation this way), and uv.lock had silently drifted from pyproject.toml since July. - Refresh uv.lock (bumping the yanked build==1.5.1 pin) and export it as constraints/ci.txt (uv export --no-hashes --all-extras --no-emit-project). - quality-fast, truthbench, nightly-online-large, large-data-status, coverage-badge and performance-large install with -c constraints/ci.txt. - quality-fast type-checks after the tests, downgrading to numpy<2.5 only for mypy (NumPy 2.5 stubs need Python 3.12; mypy targets 3.10). - test-matrix stays unpinned as the early warning for upstream breakage. - New lockfile job fails when uv.lock or constraints/ci.txt is stale; `make constraints` refreshes both.
Contributor
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Advanced Run ID: Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
…w contract The contract test asserted the exact pre-constraints install line in performance-large.yml, so every test lane failed once the workflow began installing with -c constraints/ci.txt.
This was referenced Sep 14, 2026
kevincostner17
added a commit
that referenced
this pull request
Sep 14, 2026
- perf-regression.yml installs with -c constraints/ci.txt like the other gating jobs (left out of #194 to avoid conflicting with #193). - MissForest convergence converts with to_numpy(dtype="float64", na_value=np.nan): the same pandas < 2 masked-array conversion that #192 fixed in the streaming state. Adds a nullable Int16 MissForest test. - contributor-roadmap.md no longer advertises #33 and #31, which are resolved; it points at the good-first-issue label instead.
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.
Summary
The gating CI jobs resolved dependencies fresh on every run. That's how numpy 2.5 turned median imputation into an
OverflowErrorwithout any freshdata change (#187). Meanwhileuv.lockhad drifted: it was last updated 2026-07-11, anduv lock --checkfailed against today'spyproject.toml.This PR pins the jobs that gate merges and releases to the lockfile. The test matrix stays unpinned on purpose, so upstream breaks still show up early and in one place.
Changes
uv.lock: refreshed withuv lock, andbuildbumped off the yanked1.5.1(now1.6.1;1.4.4on Python 3.9).constraints/ci.txt(new):uv export --frozen --no-hashes --all-extras --no-emit-project.--no-emit-projectdrops the-e .line, which pip rejects in a constraints file. Pins carry Python-version markers, so one file serves every interpreter.ci.yml:quality-fast,truthbench,nightly-online-large,large-data-statusandcoverage-badgeinstall with-c constraints/ci.txt.quality-fastnow runs lint and tests on the locked numpy (2.5.1 on 3.12), then downgrades tonumpy<2.5just formypy. mypy targets 3.10, and NumPy 2.5's stubs require 3.12.test-matrixis unchanged and commented as intentionally floating.lockfilejob:uv lock --check, re-export, andgit diff --exit-code constraints/ci.txt. It fails with a hint to runmake constraints.performance-large.yml: installs with constraints.Makefile: newconstraintstarget, which runsuv lockand then the same export as the CI job.perf-regression.ymlisn't touched here, to avoid conflicting with the A/B gate PR. It can pick up-c constraints/ci.txtin a follow-up.Verification
uv lock --check: passes.make constraintsregeneratesconstraints/ci.txtbyte-for-byte (git diff --exit-codeis clean).pip install --dry-run -c constraints/ci.txt -e ".[dev,ml]", resolves 68 packages, including: