-
Notifications
You must be signed in to change notification settings - Fork 1.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
py-scipy: update to 1.11.4; fix build on old systems #21875
Conversation
Notifying maintainers: |
Apparently python38 subport fails. Let me see what can be done. In the worst case we may peg it to an earlier version. UPD. Well, yeah:
Support for Python 3.8 dropped here: scipy/scipy@890ffe9 |
I will hopefully fix this now, gimme few min. |
@barracuda156 I was working on this update locally and we should be switching to the meson-python backend. As I said before the issue is that the variants with other BLAS/LAPACK versions like MKL are mot working. Did you verify that this all works correctly in this PR? Also, large patchfiles as are included here are far from ideal and upstream likely has reasons do change things; so just reverting upstream commits is not something we should do hastily. |
@reneeotten Let’s do this PR together then? I make it a draft for now. If you could implement meson backend either for python 3.12 (where it is perhaps needed) or for some subset of systems (we do not need meson backend for all systems, as of now, it seems to be rather defunct, judging by As for patches you refer to, they are applied only for systems where |
P. S. I did not test exotic BLAS variants. By the way, |
TODO:
|
I was also working on this port locally and can confirm that Meson is needed for Python 3.12. I tried adding a py312 subport to the existing setup.py-based port, but it failed because
Does it have to? I mean, for this port. In my attempt to build with Meson I tried building the py311 subport as well, and it worked just fine (at least after replacing the
I use an older system where ATLAS is supported. If needed, I can help test the |
it should be fixed, as far as I know. |
Well, recent switch to meson build for So I suspect that |
I see. I was not aware of the problems related to Meson, I thought the NumPy issues only were consequences of the OpenBLAS issues after the switch to CMake. If the Meson solution is experimental, I agree that it is best to restrict it to only where it is needed. If it had been stable, I would think that dealing with one build system would be better than dealing with two (and give a cleaner Portfile). |
@reneeotten We could have merged this first and fix the build, and then you could move new systems and python 3.12 to meson in another PR. What do you think? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@barracuda156 if you could comment/address the few comments I left then I think we should merge this. @michaelld hasn't responded since the beginning of this PR and we could merge these changes to get the update on all currently supported versions.
Python 3.12 support is still much needed, but perhaps providing that separately for now with the meson
backend is the quickest way forward. Nevertheless, with a newer release of SciPy we will have to bite the bullet and switch everything to the new backend as they will not be providing support for the setuptools-based installs and we should/will not keep on patching this in....
2d29082
to
012a319
Compare
@reneeotten I have moved tarball_from as you advised and dropped revbump for py38 version. UPD. So this is why I revbumped py38- one: https://github.com/macports/macports-ports/actions/runs/7423947967?pr=21875 |
please add the Trac tickets to the body of the commit message and then I will merge. I will add vanilla support for PY312 (i.e., possibly without variants right now) and we will work on moving to the new backend completely for the next release. If possible, please add fixing |
@barracuda156 I just added the Trac tickets to the commit message but now, for some reason, it says "This branch cannot be rebased due to too many changes" - no idea how to fix that yet.... |
@reneeotten It worked now, right? CI running, and commit body has links. |
probably a temporary GitHub glitch... |
Wonder what fails now, we literally changed nothing with the code besides |
@reneeotten I guess, some recent changes to cython breaking
|
Yes, there is a Trac ticket about it... Haven't had time to look at that - perhaps tonight. Perhaps we should use the earlier version for now (i.e., py-cython-compat). |
the last commit should hopefully fix it - it uses |
@reneeotten Great, thank you! |
Fixes: https://trac.macports.org/ticket/68014
Closes: https://trac.macports.org/ticket/68379
Closes: https://trac.macports.org/ticket/68380
Description
Update, fix old systems.
Type(s)
Tested on
macOS 14.2.1
Xcode 15.1
macOS 10A190
Xcode 3.2
Verification
Have you
port lint --nitpick
?sudo port test
?sudo port -vst install
?