Package Details: python-museval 0.4.1-1

Git Clone URL: https://aur.archlinux.org/python-museval.git (read-only, click to copy)
Package Base: python-museval
Description: Source separation evaluation tools for Python
Upstream URL: https://github.com/sigsep/sigsep-mus-eval
Licenses: MIT
Submitter: Auerhuhn
Maintainer: Auerhuhn
Last Packager: Auerhuhn
Votes: 0
Popularity: 0.000000
First Submitted: 2022-12-02 09:10 (UTC)
Last Updated: 2024-11-11 18:47 (UTC)

Latest Comments

Auerhuhn commented on 2024-06-02 14:43 (UTC) (edited on 2024-07-05 14:28 (UTC) by Auerhuhn)

This package has outlived its usefulness. Disowning.

Update: Adopted it again because turns out it’s useful for other source separation packages, too.

Auerhuhn commented on 2024-03-06 13:16 (UTC)

Uploaded version 0.4.0-4, which should fix the issue.

Thanks @Henry-ZHR and @Potajito for the detailed reports and guidance.

Auerhuhn commented on 2024-03-06 12:59 (UTC)

@Henry-ZHR

So your git apply actually does nothing.

Thanks for the pointer! The git apply doesn’t do anything unless the src directory is outside of the AUR worktree.

Building in a clean chroot can cause src to reside outside of the worktree, which is how I test my PKGBUILDs. This explains why I never ran into the error.

I can reproduce the error on my machine now by adding --verbose to the git apply command and then running makepkg -oC from the AUR worktree directly.

Will provide a fix by running git init from inside the extracted source tarball before running git apply.

Henry-ZHR commented on 2024-03-06 10:11 (UTC)

https://stackoverflow.com/a/62967602

Since there is no .git in the src directory, the base directory will be the one of the aur package itself. So your git apply actually does nothing.

And I'd suggest using --verbose for something like git apply.

Auerhuhn commented on 2024-03-05 15:16 (UTC)

Hi @Potajito,

Thanks for including the test failure.

Can you check whether github-pr-93.patch really applied cleanly before you started the build? That patch comes from upstream and increases the atol value to 0.1. In your test failure output, the value still seems to be 0.01 (1e-02).

I also just double-checked and I can confirm that the test is working with atol=0.1 (1e-01).

Potajito commented on 2024-03-05 15:07 (UTC)

While updating from 0.4.0-2 to 0.4.0-3 I'm getting


E               assert False
E                +  where False = <function allclose at 0x731644b608b0>(array([9.75952, 9.06763, 6.47532, 3.52292, 9.26284, 9.36795]), array([9.69665, 9.00579, 6.41683, 3.47194, 9.20547, 9.31067]), atol=0.01, equal_nan=True)
E                +    where <function allclose at 0x731644b608b0> = np.allclose

tests/test_regression.py:90: AssertionError
FAILED tests/test_regression.py::test_track_scores[Music Delta - 80s Rock] - assert False
================ 1 failed, 36 passed, 5 warnings in 63.37s (0:01:03) =================

While running the tests. Any idea?