Closed harryzcy closed 6 months ago
@arigo @nitzmahone Can you help to review this? Thank you
The CI is passing on my fork. See https://github.com/harryzcy/cffi/actions/runs/8071125490
Can you help to approve the CI workflow? or is there anything blocking? Thanks
I have no idea how .github/workflows/ci.yaml
works, but if Matt cannot do the review for the time being, I would be happy to accept your PR with just a few words from you: it looks like it's only adding support to run on some extra CI system of github, right? Is there any side-effect to take into account for us?
it looks like it's only adding support to run on some extra CI system of github, right?
Yes
Is there any side-effect to take into account for us?
I'm not aware of any. I don't think it effects existing wheels for other os/architecture, since the build happens separately for each system.
Thanks
@arigo Thanks!
:-)
Sorry, been buried the past few weeks leading up to a vacation (first day back)...
I should've commented earlier: I'd been trying to avoid adding new support for wheel builds on arches where we don't have native builders. The full test suite is prohibitively slow on emulated builds (and Windows, for reasons I still haven't had time to investigate), so these wheels will be effectively untested beyond the very tiny slice of the test suite that we run on the Windows wheels and anything non-Intel or Apple Silicon. Caveat Emptor :wink:
PS, been working a bit with @samdoran on fixing up the test fixture isolation to work properly with pytest-xdist
- that should allow us to run the tests much faster in parallel worker processes.
Unrelated to this PR, macos-14
would provide Apple Silicon and it has no intel-mac as a standard GitHub runner. I don't see how other linux non-intel arches would be possible without emulation.
super-linter is preparing for arm64 support https://github.com/super-linter/super-linter/issues/5070. It would be very helpful to have a
musllinux_aarch64
wheelFix #41