Skip to content
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

Travis: test builds on arm64, ppc64le, s390x (for stable-4.11) #3966

Merged
merged 2 commits into from
Apr 18, 2020

Conversation

fingolfin
Copy link
Member

@fingolfin fingolfin commented Apr 16, 2020

Besides just adding three jobs to the Travis build matrix, we have to
deal with some issues with the VMs for the three new platforms:

  • Switch to using Python 3 instead of Python 2 to workaround an issue where python is python2.7 but pip is pip3.6.
    Always call the right pip matching the python binary; and prepare for python being Python 3.
    For details, see https://travis-ci.community/t/8165

  • Work around the fact that the pip cache dir is owned by root.
    For details, see https://travis-ci.community/t/7822/6

  • Use travis_terminate liberally in .travis.yml to workaround another issue with the three new platforms; as a side benefit, we also get Travis to reliably and early on abort CI builds with failures.
    For details, see https://travis-ci.community/t/7659/.

  • Adjust a test for OPERS_CACHE_INFO to pass on the three new platforms.

This is a backport of PR #3744. I am hoping that we see the crash from issue #3919 so that we can then test PR #3965.

@fingolfin fingolfin added release notes: not needed PRs introducing changes that are wholly irrelevant to the release notes topic: ci Anything related to GitHub Actions, Codecov, AppVeyor, Coveralls, Travis, ... labels Apr 16, 2020
@fingolfin fingolfin force-pushed the mh/s390x-4.11 branch 2 times, most recently from 80a690d to 739d309 Compare April 16, 2020 23:44
@fingolfin
Copy link
Member Author

fingolfin commented Apr 17, 2020

Nice: the arm64 build failed exactly as predicted (cf this travis log), i.e. we can reproduce the reports by @jamesjer and Bill. I'll now add the fix to this PR to see if it really helps.

@coveralls
Copy link

coveralls commented Apr 17, 2020

Coverage Status

Coverage decreased (-0.0002%) to 84.56% when pulling 7bc1660 on fingolfin:mh/s390x-4.11 into 523a034 on gap-system:stable-4.11.

@fingolfin
Copy link
Member Author

And now the fix makes it pass. So I guess this PR is simultaneously backports PR #3744 and PR #3965

Besides just adding three jobs to the Travis build matrix, we have to
deal with some issues with the VMs for the three new platforms:

- Always call the right pip matching the `python` binary; and
  prepare for `python` being Python 3.
  For details, see <https://travis-ci.community/t/8165>

- Work around the fact that the pip cache dir is owned by root.
  For details, see <https://travis-ci.community/t/7822/6>

- Use travis_terminate liberally in .travis.yml to workaround another
  issue with the three new platforms; as a side benefit, we also get
  Travis to reliably and early on abort CI builds with failures.
  For details, see <https://travis-ci.community/t/7659/>.

- Adjust a test for OPERS_CACHE_INFO to pass on the three new platforms.
Inlining this function can break the garbage collector. This was the case on
Linux aarch64.

Thanks to Jerry James (Fedora) and Bill Allombert (Debian) for help tracking
this down.

Fixes gap-system#3919
@fingolfin fingolfin merged commit f0a8f49 into gap-system:stable-4.11 Apr 18, 2020
@fingolfin fingolfin deleted the mh/s390x-4.11 branch April 18, 2020 23:37
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
release notes: not needed PRs introducing changes that are wholly irrelevant to the release notes topic: ci Anything related to GitHub Actions, Codecov, AppVeyor, Coveralls, Travis, ...
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants