-
Notifications
You must be signed in to change notification settings - Fork 2.4k
py-scipy: add v1.14.0 #44967
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: add v1.14.0 #44967
Conversation
I know we had this problem before, but I don't remember how we solved it. |
NumPy supports it already; the
The libraries and headers are in a system location; they'll be on the search path even when you're using compilers other than Apple Clang. Versions are directly tied to macOS version: on macOS >=13.3 the new Accelerate is always available, and it provides a LAPACK 3.9 API. In macOS 15.0 that will be upgraded to LAPACK 3.11. numpy/numpy#24053 may be instructive to look at. The main problem for Spack may be symbol naming; the symbol suffixes to access the updated library at |
See #42213 (comment). The problem was a too old version of |
Okay, hoping the last commit fixes this. Spack is using a newer version of setuptools to build pythran (as instructed). However, we have no constraints on the version of setuptools present at run-time. I don't know why an older version was chosen, but the latest commit should force a newer version to be used at run-time as well. I didn't realize the version of setuptools present at run-time mattered for pythran but it seems that that's the case. |
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.
LGTM, @rgommers do you have any comments?
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.
LGTM, seems ready to merge.
I didn't realize the version of setuptools present at run-time mattered for pythran but it seems that that's the case.
It seems odd indeed. It may not be inside pythran
itself, but rather than setuptools
is somehow used in however Spack has implemented entrypoint support for Python.
https://github.com/scipy/scipy/releases/tag/v1.14.0
@rgommers I would be interested in getting Accelerate working in the future. Spack only supports a single BLAS/LAPACK library in the DAG at a time. If I understand correctly, SciPy supports Accelerate now, but NumPy still doesn't? Are there plans for NumPy to support Accelerate? I don't even know where to start with adding an Accelerate package to Spack, not sure if you know where libraries live, how to check what versions are present, how those versions map to BLAS/LAPACK API versions, etc.