chore(ci): target py3.12 only (align to the PROTEA stack)#18
Merged
Conversation
Drop the 3.10/3.11 CI matrix legs and set requires-python to >=3.12. The whole PROTEA stack runs on Python 3.12; the parity oracle is bit-exact there. The 3.10 leg failed only on a pandas index-ordering quirk (same rows, different order, before any value comparison) caused by unpinned deps resolving differently on the older interpreter, not a metric divergence.
frapercan
added a commit
that referenced
this pull request
Jun 24, 2026
Promotes develop's sparse evaluation work to main so the branch PROTEA installs (`@main`) actually uses it. main had drifted: it only carried today's docs/CI commits (#17, #18) while the sparse perf landed on develop and was never promoted. ## Brings to main - **#16** `perf(eval)`: predictions stored as `scipy.sparse` CSR, never materialising the dense N x T matrix (11.9 -> 7.0 GB). - **#15** `perf(graph)`: sparse CSR GO DAG adjacency + O(1) `deque` topological sort (about 3x faster build, ~3x less RAM). ## Verified (py3.12, the only interpreter the stack uses) - ruff + mypy clean. - Parity suite `CAFAEVAL_PARITY_PHASE=B pytest tests/diff`: 7 passed, 6 xfailed, **0 hard failures** (the sparse path stays bit-/ULP-exact vs the dense fallback and the upstream oracle). ## Conflict resolution Only README.md conflicted (both branches edited the optimisation table): took develop's rows (they document the C1/C3 sparse kernels) and re-applied the em-dash purge from #17. pyproject auto-merged (`requires-python >=3.12` from #18 + the scipy/pyarrow sparse-era deps). Follow-up: bump PROTEA's `poetry.lock` cafaeval pin so the platform consumes this.
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 cafaeval CI green by aligning it to the Python the project actually uses.
Change
["3.10", "3.11", "3.12"]->["3.12"].requires-python>=3.9->>=3.12(the PROTEA stack is py3.12; consistent with the mypy target set in docs(readme): purge em/en-dashes from prose #17).Why (parity context)
The red check was
Tests + parity (py3.10), failing atactual.index.equals(expected.index)with 56 vs 56 rows, i.e. a row-order difference before any value comparison, caused by unpinned pandas/numpy resolving differently on py3.10. It is not a metric divergence: the parity suite is bit-/ULP-exact on py3.12 (verified locally, 0 hard failures), which is the only interpreter the stack runs on. No parity logic or pins were touched.