-
-
Notifications
You must be signed in to change notification settings - Fork 25
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
update scikit to 0.20.0 #100
Conversation
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( |
meh, ok for some reason https://pypi.io/packages/source/s/scikit-image/scikit-image-0.20.0.tar.gz doesn't point to the source dist... |
@conda-forge-admin, please rerender |
…nda-forge-pinning 2023.03.16.14.46.37
Thanks for the update @k-dominik! I don't know why the conda-forge bot went on strike. 🤔 I expect the tests are failing because the dependencies in the recipe need to be updated. Here's the current deps: |
ref: https://github.com/scikit-image/scikit-image/blob/907bfb71d2c8e4d58c2dffc6056760c1e00d070b/pyproject.toml#L70-L79 also removed Python 3.7 special pins - Python 3.7 is not longer supported by skimage as far as I can tell.
Thank you for the quick feedback @jni, I updated the dependencies as suggested, but there seems to be something off with pypi - the tests fail with:
okay, after looking at this, I realized that the artifact on pypi has |
not sure the required version of Is it possible that all the cross-compilation issues in CI are related to that? |
I guess @jarrodmillman would know. This minimum version requirement comes from scikit-image/scikit-image#6757. |
This is a known change. Meson, our new build system, follows the pypa rules for cleaning up sdist names, which setuptools (ironically) never did. |
This PR to the meson-python-feedstock suggests that those issues maybe should be fixed, but I see that in the recipe you didn't constrain meson-python, so I don't think that's where they come from. Hopefully @stefanv and @jarrodmillman can have a look at the build soon... |
I think Meson needs the cross files to cross-compile on arm64? See, e.g., the SciPy patch and the Meson docs. |
@k-dominik I am going to try to get the correct CPU version of ninja in place; if pushing to your PR in conda-forge is considered bad practice, I do apologize, and feel free to remove those patches. |
|
@stefanv rerendering does 2 things:
In order to not step on each other's toes, I'm going to make small attempts in Sorry for the noise here. |
I thought scipy had already trailblazed meson + scientific python? is that not true? I thought we were following the build procedure of an other major package? Can we use their recipe as a model? |
I haven't really kept up with how you all managed to build OSX on CIs with meson python. Maybe that can be extended to work on conda-forge? If you point me to it I can likely help with that. I noticed this bug report opened by isuruf (conda-forge core member) mesonbuild/meson-python#321 |
That's a decent set of passes. Now, most of the remaining ones are of the form:
Still no aarch64 / arm64 builds. |
I don't think there's much more I can do to help here. I'll follow along, just in case some knowledge of the build system is helpful. @hmaarrfk We don't cross-compile for macosx, we just build natively: https://github.com/scikit-image/scikit-image/blob/main/.github/workflows/wheel_tests_and_release.yml The one tricky bit with that build was a bug in meson-python where they look at |
Not that I could find at https://github.com/conda-forge/scipy-feedstock or https://github.com/conda-forge/scipy-feedstock/pulls |
OK, I found the work that SciPy's done that can (maybe?) get us our macosx cross-compiled binaries! |
4461633
to
ebab4db
Compare
This reverts commit 2ef040a.
See builds in aarch seems to run and pass tests on linux. |
@isuruf I apologize for tagging you personally, but I think we are quite close on this build here, and wondered whether you had any insight on how to get the arm64 cross-compilation going? |
#101 seems to be going in the correct direction |
closed in favor of #101 |
Checklist
0
(if the version changed)conda-smithy
(Use the phrase@conda-forge-admin, please rerender
in a comment in this PR for automated rerendering)