Skip to content

Cover SuiteSparse:GraphBLAS 10.4 and 10.5 in the CI psg pool - #629

Merged
eriknw merged 5 commits into
mainfrom
ci-latest-deps
Sep 18, 2026
Merged

eriknw merged 5 commits into
mainfrom
ci-latest-deps

Conversation

@eriknw

@eriknw eriknw commented Aug 26, 2026

Copy link
Copy Markdown
Member

Stacked on #628. Second rung below the pre-existing stack; sits between #628 and #587.

The pool stopped at psg 10.3.1.0, so CI installed nothing newer except on the draw that picks "latest", which does not pin and so never produces a reproducible combination. Three releases have landed since, and each carries a SuiteSparse:GraphBLAS this project has to work against: 10.4.0 moved a field in the matrix struct, and 10.5.0 dropped the post-build iso check that four tests used to rely on.

psg 10.5.0.0 is on PyPI and, since 2026-08-26, on conda-forge, so the conda-forge, wheel and source draws all cover it.

scripts/check_versions.sh tracks the newest version we test, so its psg floor moves with the pool.

Only the psg pools change here. The other pools are revised further up the stack (#631), so raising them belongs there rather than in a commit that would conflict with every rebase.

Note on placement: this sits BELOW rung 07 so that every rung's CI can draw SuiteSparse 10.4/10.5 against the adaptations in #628. Until #628 and this PR merge into main, #587's Files tab also shows these two commits (the stack manager would not allow retargeting #587's base); it self-heals once these merge.

eriknw added 2 commits August 26, 2026 12:29
…ibrary

10.5.0 stopped checking, after a build, whether every value happened to be
the same: "GrB_Matrix_build and GrB_Vector_build: no longer do a post-iso
check; they leave the matrix in non-iso format even if all the entries are
the same." Four tests read that check as if it were python-graphblas's own
contract and went red. Each one wanted an iso object rather than an
inference, so each now asks for one with ``ss.build_scalar``, which means
the same thing on every SuiteSparse version and produces the same storage
formats the tests were selecting for. ``from_coo`` is left alone: adding a
scan of the values to recover the old behavior would re-add the cost
upstream just removed.

The version gates had a second, quieter problem. ``graphblas.core.ss``
derived the C library version by parsing ``suitesparse_graphblas.__version__``,
which is the version of the Python wrapper. The two normally agree, but a
development build between releases reports the previous one: psg
10.4.1.0 against a 10.5.0 library here, so every gate saw 10.4.1 and any
10.5 branch would have been silently skipped. Read the library's own
GxB_IMPLEMENTATION_* constants instead, falling back to the old parse where
a build does not expose them.

test_openmp_enabled keeps its hard assertion and gains a message saying what
is wrong and that the fault is in the installed build. It caught a real one:
conda-forge's first graphblas 10.5.0 build on osx-arm64 shipped without
OpenMP. The build 1 rebuild (2026-08-26) restored OpenMP and the assertion
passes again.
The pool stopped at psg 10.3.1.0, so CI installed nothing newer except on
the draw that picks "latest", which does not pin and so never produces a
reproducible combination. Three releases have landed since, and each carries
a SuiteSparse:GraphBLAS this project has to work against: 10.4.0 moved a
field in the matrix struct, and 10.5.0 dropped the post-build iso check that
four tests used to rely on.

psg 10.5.0.0 is on PyPI and, since 2026-08-26, on conda-forge, so the
conda-forge, wheel and source draws all cover it.

`scripts/check_versions.sh` tracks the newest version we test, so its psg
floor moves with the pool.

Only the psg pools change here. The other pools are revised further up this
branch stack, so raising them belongs there rather than in a commit that
would conflict with every rebase.
Base automatically changed from ssgb-105-adapt to main September 11, 2026 16:46
Add scripts/audit_{pypi,conda}_versions.py, which check the pools and
tables in ci_pick_versions.py against what is actually published. The
--validate stress test proves the picker agrees with the constraints it
documents; it cannot notice when a constraint drifts from reality, since
that happens when a package releases, not when we commit. A new weekly
workflow runs both audits.

In the picker, replace the hand-written scipy/numba compatibility rules
with SCIPY_MAX_NUMPY and NUMBA_MAX_NUMPY ceiling tables, add
SCIPY_MIN_PYTHON, and gate the tables behind check_ceilings() and
check_python_pools() so a half-updated table fails the run instead of
producing a silently wrong pick. Bump the pools to numpy 2.5, scipy 1.18,
awkward 2.13, numba 0.67 and sparse 0.19, and offer sparse on py3.13/3.14
now that numba supports them.

Also bump pre-commit hooks, update check_versions.sh to match, and ignore
PLR0917, CPY and ISC004 (all newly stable in ruff 0.16).

Fixes found while reviewing the above:

- test_and_build.yml discarded the picker's exit status. `eval "$(cmd)"`
  reports eval's status, and that step's shell has no -e, so a picker
  failure left every *ver variable unset and quietly installed the latest
  of everything. Assign first, then fail the step.
- pyproject: bound the sparse py3.14 marker below 3.15. sparse depends on
  numba, so it has to drop out wherever numba does.
- _pick_scipy filtered ceilings and requires-python but not the one numpy
  floor scipy imposes, leaving the order dependence it exists to remove.
- audit_pypi: report a missing scipy release instead of raising KeyError,
  honor a `<=` numpy cap rather than skipping the check, and keep the
  notes section inside the network guard.
@eriknw

eriknw commented Sep 18, 2026

Copy link
Copy Markdown
Member Author

The CI failure is a segfault that should be fixed by #589 (once we get to that one). This PR is foundational and improves how dependencies are handled in CI. Merging.

@eriknw
eriknw merged commit bb06322 into main Sep 18, 2026
17 of 18 checks passed
@eriknw
eriknw deleted the ci-latest-deps branch September 18, 2026 02:47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant