Tests failing on Void Linux i686

oynqr commented

2023-12-01 07:25:59 +00:00

When packaging img2pdf for Void Linux, I noticed that tests were failing for the i686 architecture. This is a GitHub Actions run showing the problem: https://github.com/void-linux/void-packages/actions/runs/6913057883/job/18809496476. Relevant log segments are attached.

I currently opted to disable testing on non-64-bit architectures, but I would like to achieve some clarity as to why these tests are failing.

When packaging img2pdf for Void Linux, I noticed that tests were failing for the i686 architecture. This is a GitHub Actions run showing the problem: [https://github.com/void-linux/void-packages/actions/runs/6913057883/job/18809496476](https://github.com/void-linux/void-packages/actions/runs/6913057883/job/18809496476). Relevant log segments are attached. I currently opted to disable testing on non-64-bit architectures, but I would like to achieve some clarity as to why these tests are failing.

img2pdf_i686_actions_run.log.gz

13 KiB

josch commented

2023-12-01 13:04:15 +00:00

Owner

Thank you for packaging img2pdf for another distro!

I'm not familiar with github actions. What button do i have to click to see the log?

Note, that it is unfortunately currently normal for a few tests to fail. For example right now one reason is this:

https://github.com/ImageMagick/ImageMagick6/issues/285

I'm also maintaining img2pdf in Debian and have some tests disabled there:

https://sources.debian.org/src/img2pdf/0.5.1-1/debian/patches/imagemagick-issue-285/

Thank you for packaging img2pdf for another distro! I'm not familiar with github actions. What button do i have to click to see the log? Note, that it is unfortunately currently normal for a few tests to fail. For example right now one reason is this: https://github.com/ImageMagick/ImageMagick6/issues/285 I'm also maintaining img2pdf in Debian and have some tests disabled there: https://sources.debian.org/src/img2pdf/0.5.1-1/debian/patches/imagemagick-issue-285/

oynqr commented

2023-12-01 18:02:21 +00:00

Author

You sadly need to log in to see logs, but I attached the relevant sections in the file.

I have patches applied and tests passing already on x86_64, this is strictly affecting i686.

You sadly need to log in to see logs, but I attached the relevant sections in the [file](https://gitlab.mister-muffin.de/attachments/d08e4fc2-493e-4325-9604-b55e67e82b9b). I have patches applied and tests passing already on x86_64, this is strictly affecting i686.

josch commented

2023-12-02 06:09:53 +00:00

Owner

Okay, I have never seen that type of error. There are 70 tests failing and they all seem to fail because the testsuite expects an image with 8 bits of depth but imagemagick apparently reports it as 16 bits. In Debian we also run the tests on 32bit x86 and I have not seen those errors there. Maybe this is related to how you build imagemagick for i386 in void linux?

I reminds me of #119 and #120. Maybe imagemagick changed behaviour and now also reports a different meaning for the depth attribute?

Okay, I have never seen that type of error. There are 70 tests failing and they all seem to fail because the testsuite expects an image with 8 bits of depth but imagemagick apparently reports it as 16 bits. In Debian we also run the tests on 32bit x86 and I have not seen those errors there. Maybe this is related to how you build imagemagick for i386 in void linux? I reminds me of #119 and #120. Maybe imagemagick changed behaviour and now also reports a different meaning for the depth attribute?

Tests failing on Void Linux i686 #185