python3-magic: update to 0.4.27.

[Rutpiv]

Testing the changes

• I tested the changes in this PR: briefly

Local build testing

• I built this PR locally for my native architecture (x86_64)
• I built this PR locally for these architectures using specific masterdirs:
  • x86_64-musl
  • i686
• I built this PR locally for these architectures (crossbuilds):
  • aarch64
  • aarch64-musl
  • armv7l
  • armv7l-musl
  • armv6l
  • armv6l-musl

[classabbyamp]

why don't they work now?

[classabbyamp]

adding

checkdepends="python3-pytest"
make_check_pre="env LC_ALL=en_US.UTF-8"

and backporting ahupp/python-magic@545a2a5 and ahupp/python-magic@4ffcd59 makes them work

[Rutpiv]

I reached the same conclusion after your first comment.
I identified those upstream commits afterwards, but hadn’t updated the template yet.
I’ll follow your suggestion and update the template accordingly.

[Rutpiv]

For context, the tests were already failing with the previous version as well when built against current libmagic. The changes you suggested make them pass again.

[ahesford]

In general, if you can't explain why the tests don't work and justify why they shouldn't be expected to work, they should not be skipped. Disabling tests just because they fail defeats the purpose of our CI.

[Rutpiv]

You’re right, and I agree with that guideline. Skipping tests just because they fail isn’t appropriate unless we can clearly justify why they shouldn’t be expected to pass.

In this case I made a wrong assumption based on the tag date (June 7, 2022) and thought there wouldn’t be upstream changes affecting the test expectations without a new release. That turned out to be incorrect: the project did receive relevant upstream fixes without bumping the version, and my initial approach was rushed.

After classabbyamp’s question I took a deeper look and it does seem the failures likely started around the mid-2023 libmagic/file updates, and there are upstream commits addressing exactly this.

Sorry for the trouble.