Package Details: python-conda 25.1.0-2

Git Clone URL: https://aur.archlinux.org/python-conda.git (read-only, click to copy)
Package Base: python-conda
Description: OS-agnostic, system-level binary package manager and ecosystem https://conda.io
Upstream URL: https://github.com/conda/conda
Licenses: BSD-3-Clause
Provides: python-conda, python-conda-env
Submitter: cyrevolt
Maintainer: flying-sheep (brianrobt)
Last Packager: brianrobt
Votes: 34
Popularity: 0.57
First Submitted: 2016-04-08 07:41 (UTC)
Last Updated: 2025-01-22 15:17 (UTC)

Latest Comments

1 2 3 4 5 6 .. 11 Next › Last »

bcb commented on 2025-01-22 16:31 (UTC)

Yes, the new version successfully installed.

It may be that GitHub don't guarantee patches will always get the same hash. Either of those downloaded patches would successfully apply to the code, its only an issue when using the hashsum for integrity checking. When I'll get a chance I'll see if its mentioned in any documentation, and otherwise report it (or post in the community discussions) to get some clarification.

brianrobt commented on 2025-01-22 16:20 (UTC)

Thanks for digging into that @bcb! Were you able to install the new version successfully?

As for the GitHub issue with shortening the commit hash in the patch, that is an interesting bug. I've never seen that before either, and we probably aren't the only ones who have or will experience it. If you want to report it to GitHub, you can use this link to open a ticket: https://support.github.com/contact/bug-report

bcb commented on 2025-01-22 16:15 (UTC)

Thanks!

I just had another look at this. The cause is GitHub changing the length of the commit ID it reports in the patch (line 11). The one you based the SHA256 sum on has

index 91f37e144d0..d3b655e47f4 100644

where the one I get most of the time has

index 91f37e144d..d3b655e47f 100644

with one character removed at the end of each of the commit IDs. There was no other differences. Why that happens I don't know, I've never seen it happen before when I've added patches like that to a PKGBUILD.

brianrobt commented on 2025-01-22 15:11 (UTC) (edited on 2025-01-22 15:12 (UTC) by brianrobt)

That's very strange behavior. I agree that it should be included locally, so I'll go ahead and make the change.

bcb commented on 2025-01-22 08:57 (UTC) (edited on 2025-01-22 09:03 (UTC) by bcb)

@brianrobt -- clean build gives the same result.

I can also replicate this outside of makepkg by downloading the patch myself:

$ cd /tmp
$ curl https://github.com/conda/conda/commit/62196c897df3d7aea7063d0c08d1bf6e6fd91600.patch > test.patch
$ sha256sum test.patch
b193fe4fce7e141ec6dd7d40440a884f00a0b2510f665465585c6f3a2b2bd71f  test.patch

EDIT: I just tried the above steps again and this time I got the hashsum in the PKGBUILD?!?! For the record, I did this manual test yesterday when I got the build failure.

Simpler version:

$ curl https://github.com/conda/conda/commit/62196c897df3d7aea7063d0c08d1bf6e6fd91600.patch | sha256sum

If I run that multiple times, sometimes I get the hashsum from the PKGBUILD, but the majority of the time I get b193fe4.... Is something a bit unstable with the generation of the patch by GitHub? Would it make sense to add the patch to this repository to avoid this?

zianzit commented on 2025-01-22 08:47 (UTC)

Can you try a clean install and see if that resolves your issue?

Same problem for me. File py-3.13-logging.patch has b193fe4fce7e141ec6dd7d40440a884f00a0b2510f665465585c6f3a2b2bd71f but in PKGBUILD dcd0edb6cc59c67629ddfa6e9fb38f53eff293df92d8a0222ede051c8e66b149 Clean install doesnt help.

brianrobt commented on 2025-01-21 14:52 (UTC)

@bcb, dcd0edb6cc59c67629ddfa6e9fb38f53eff293df92d8a0222ede051c8e66b149 should be the correct sha256sum. Can you try a clean install and see if that resolves your issue?

bcb commented on 2025-01-21 08:32 (UTC)

With the latest update I get a hash mismatch for the new patch.

$ sha256sum py-3.13-logging.patch
b193fe4fce7e141ec6dd7d40440a884f00a0b2510f665465585c6f3a2b2bd71f

PKGBUILD has dcd0edb6cc59c67629ddfa6e9fb38f53eff293df92d8a0222ede051c8e66b149

brianrobt commented on 2025-01-20 16:48 (UTC)

@Dayear, you're the first to report that as far I know. I did a little digging and it looks like conda switched from appdirs to platformdirs with Python 3.12. I'll add it as a dependency. Thanks for reporting this!

Dayear commented on 2025-01-20 08:58 (UTC)

I got this error output: conda ModuleNotFoundError: No module named 'platformdirs', then I installed the package python-platformdirs to solve this problems. Is this a common problem that needs to be fixed?

1 2 3 4 5 6 .. 11 Next › Last »