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 CI failing due to missing features: fribidi harfbuzz raqm #198

Closed
hugovk opened this issue May 10, 2021 · 4 comments
Closed

Travis CI failing due to missing features: fribidi harfbuzz raqm #198

hugovk opened this issue May 10, 2021 · 4 comments

Comments

@hugovk
Copy link
Member

hugovk commented May 10, 2021

Travis CI builds for this repo have had several months of downtime. Since we got new Travis CI credits and re-enabled this repo (python-pillow/Pillow#5028 (comment)), the build is failing.

For example:

Tests run, but then it hangs and times out after features mismatch:

...
SKIPPED [1] Tests/test_shell_injection.py:34: cjpeg not available
SKIPPED [1] Tests/test_shell_injection.py:39: Netpbm not available
SKIPPED [1] Tests/test_shell_injection.py:45: Netpbm not available
XFAIL Tests/test_decompression_bomb.py::TestDecompressionBomb::test_exception_ico
  different exception
XFAIL Tests/test_image_resample.py::TestCoreResampleAlphaCorrect::test_levels_rgba
  Current implementation isn't precise enough
XFAIL Tests/test_image_resample.py::TestCoreResampleAlphaCorrect::test_levels_la
  Current implementation isn't precise enough
XPASS Tests/test_file_palm.py::test_p_mode Palm P image is wrong
= 2532 passed, 230 skipped, 3 xfailed, 1 xpassed, 17 warnings in 94.88s (0:01:34) =
Features should be: 'fribidi harfbuzz raqm transp_webp webp_anim webp_mux xcb'; but are 'transp_webp webp_anim webp_mux xcb'
No output has been received in the last 10m0s, this potentially indicates a stalled build or something wrong with the build itself.
Check the details on how to adjust your build configuration on: https://docs.travis-ci.com/user/common-build-problems/#build-times-out-because-no-output-was-received
The build has been terminated

The missing features are fribidi harfbuzz raqm.

From run_tests in config.sh:

pillow-wheels/config.sh

Lines 171 to 176 in 3b8d915

local features=$(python3 -c 'from PIL.features import *; print(" ".join(sorted(get_supported_features())))')
if [ "$features" != "$EXP_FEATURES" ]; then
echo "Features should be: '$EXP_FEATURES'; but are '$features'"
ret=1
fi
return $ret

Looks like run_tests is called by multibuild.

  • Why is there a feature mismatch causing this failure?
  • Why is it hanging instead of failing immediately?
@nulano
Copy link
Contributor

nulano commented May 10, 2021

The missing features are fribidi harfbuzz raqm.
...
Why is there a feature mismatch causing this failure?

From the 8.2.0 release thread (python-pillow/Pillow#5287 (comment)):

Oops, I don't think anyone actually tested the [Raqm/FriBiDi] PR on Travis. Disabling the check on Travis should be fine as the build is done with the --enable-raqm setup flag which ensures Raqm support was built, even if the tests fail to find FriBiDi after installation.

I just haven't got around to submitting a PR implementing this yet. I also don't know if I could test it since the switch (last I checked I was still on .org, and the amount of trial credits on .com is quite limited).

@hugovk
Copy link
Member Author

hugovk commented May 10, 2021

If you're still on org, best to stay there until they actually shut it down, as it uses no credits. I think the capacity problems have settled, at least the few repos I've seen still there.

It's said "Please be aware travis-ci.org will be shutting down in several weeks, with all accounts migrating to travis-ci.com. Please stay tuned here for more information." literally for months, it was meant to shut down at the end of 2020!

Or if you know what the patch is, we can test it here, or on my com repo, I have credits too.

@nulano
Copy link
Contributor

nulano commented May 11, 2021

Feature mismatch fixed in #199. Note that the PR doesn't fix the hanging build if the features don't match, just the list itself.

It seems the capacity (or rather load) is indeed better now, none of the test builds for the PR had to wait in a queue.

@hugovk
Copy link
Member Author

hugovk commented Aug 12, 2021

I think we can close this now, I don't recall any hangs and if it's rare we can restart the build.

@hugovk hugovk closed this as completed Aug 12, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants