Install openblas to compile recent Numpy versions - #15589

Thank you for your contribution to Astropy! 🌌 This checklist is meant to remind the package maintainers who will review this pull request of some common things to look for.

• Do the proposed changes actually accomplish desired goals?
• Do the proposed changes follow the Astropy coding guidelines?
• Are tests added/updated as required? If so, do they follow the Astropy testing guidelines?
• Are docs added/updated as required? If so, do they follow the Astropy documentation guidelines?
• Is rebase and/or squash necessary? If so, please provide the author with appropriate instructions. Also see instructions for rebase and squash.
• Did the CI pass? If no, are the failures related? If you need to run daily and weekly cron jobs as part of the PR, please apply the "Extra CI" label. Codestyle issues can be fixed by the bot.
• Is a change log needed? If yes, did the change log check pass? If no, add the "no-changelog-entry-needed" label. If this is a manual backport, use the "skip-changelog-checks" label unless special changelog handling is necessary.
• Is this a big PR that makes a "What's new?" entry worthwhile and if so, is (1) a "what's new" entry included in this PR and (2) the "whatsnew-needed" label applied?
• Is a milestone set? Milestone must be set but we cannot check for it on Actions; do not let the green checkmark fool you.
• At the time of adding the milestone, if the milestone set requires a backport to release branch(es), apply the appropriate "backport-X.Y.x" label(s) before merge.

👋 Thank you for your draft pull request! Do you know that you can use [ci skip] or [skip ci] in your commit messages to skip running continuous integration tests until you are ready?

It works, the ppc job passed, and I just pushed a fix for the s390x job.
But armv7 job fails while compiling Numpy (building ninja, which itself uses scibit-build and cmake...). Maybe we just skip this one for now ?

armv7 is the 32-bit replacement added by @WilliamJamieson . Maybe he has ideas? Otherwise, maybe @olebole can advise?

armv7 is the 32-bit replacement added by @WilliamJamieson.

This is correct, the original intent was to install numpy via the OS's package manager:

This means if we want to require numpy>=1.25 to build the astropy wheel, we are stuck until Ubuntu or Debian get around to publishing numpy==1.25 packages or we work out what additional packages we need to install via the OS package manager in order to build numpy from source.

@WilliamJamieson

This means if we want to require numpy>=1.25 to build the astropy wheel, we are stuck until Ubuntu or Debian get around to publishing numpy==1.25 packages or we work out what additional packages we need to install via the OS package manager in order to build numpy from source.

Unfortunately, this seems to be delayed somehow. I opened Debian#1052320 6 weeks ago, and didn't get a response; no idea why.
The additional packages required to build are listed in Debian's source package for version 1.24, however some featured may be patched out (or fixed), so it may be useful to check the patches for 1.24 as well.

Right, then I think we can build directly with the system packages in non-isolated mode. Let's see...

An alternate to having to use the OS package manager to install the various packages needed to build astropy might be to move the exotic arch builds to follow what numpy is doing in its CI as suggested by #15351 (comment). This will require a complete redo of this part of the job.

Seems it worked, but all 4 jobs now fail on the same 2 tests:

  FAILED astropy/coordinates/tests/test_representation.py::TestCartesianRepresentation::test_transform_non_contiguous_matrix - assert False
   +  where False = representation_equal(<CartesianRepresentation (x, y, z) in kpc\n    (1., 8., 12.)>, <CartesianRepresentation (x, y, z) in kpc\n    (1., 2., 3.)>)
   +    where <CartesianRepresentation (x, y, z) in kpc\n    (1., 8., 12.)> = <bound method CartesianRepresentation.transform of <CartesianRepresentation (x, y, z) in kpc\n    (1., 2., 3.)>>(array([[1., 0., 0.],\n       [0., 1., 0.],\n       [0., 0., 1.]]))
   +      where <bound method CartesianRepresentation.transform of <CartesianRepresentation (x, y, z) in kpc\n    (1., 2., 3.)>> = <CartesianRepresentation (x, y, z) in kpc\n    (1., 2., 3.)>.transform
  FAILED astropy/time/tests/test_basic.py::test_to_datetime_pytz - pytz.exceptions.UnknownTimeZoneError: 'US/Hawaii'

pytz.exceptions.UnknownTimeZoneError: 'US/Hawaii'

Wut? 🤯

Do we also have to install tzdata? https://stackoverflow.com/questions/21717411/timezone-information-missing-in-pytz

@pllim on new Debian system (not sure whether before of after Debian 12 "Bookworm" release), one also needs the tzdata-legacy package. This is no longer a mandatory package.

Finally all builds are green !
Though 5h42m for aarch64 seems a bit excessive ... do we really need to run such a slow thing ?