You are not logged in.
Pages: 1
Hello,
I'm sorry if this is a stupid question with an obvious answer, but… where do I send patches to PKGBUILDs in the repos to?
I have a patch to update Gnucash to the latest version and improve its build process, and I use the latest version it since a few weeks without issues. I don't mind keeping my own PKGBUILD but I just realized that it's still flagged out of date even though the update is simple, so the maintainer probably has no time to update it and I figured I could help them.
Except I haven't found any recommended way to submit patches... so what do I do now?
Last edited by 3beb6e7c46a615a (2023-04-12 15:42:32)
Offline
You can email the patch to the respective maintainer.
However, regarding GNUCash it is not just a simple version bump.
I attempted to update it myself and it compiles with just a version bump, but several unit tests fail.
So there's more to do to make it work properly, apparently.
Inofficial first vice president of the Rust Evangelism Strike Force
Offline
You can email the patch to the respective maintainer.
Uh, mail… not exactly my favourite. I'll try to send a patch around…
However, regarding GNUCash it is not just a simple version bump.
I attempted to update it myself and it compiles with just a version bump, but several unit tests fail.
So there's more to do to make it work properly, apparently.
It was for me. This is my patch: https://gist.github.com/swsnr/eb4aec4e9 … 521239ab42
When building in a clean chroot ("aur build -cSR --force -d abs") ends with "98% tests passed, 2 tests failed out of 126", and according to the comment in the PKGBUILD two test failures are apparently expected. Note that "check" ignored test failures anyway by means of "|| :".
Offline
When building in a clean chroot ("aur build -cSR --force -d abs") ends with "98% tests passed, 2 tests failed out of 126
That's exactly what I was referring to.
Even though the packaging does not fail, which it should imho, I would not want a package in the stable repos that does not achieve 100% of test coverage.
Inofficial first vice president of the Rust Evangelism Strike Force
Offline
lunaryorn wrote:When building in a clean chroot ("aur build -cSR --force -d abs") ends with "98% tests passed, 2 tests failed out of 126
That's exactly what I was referring to.
Even though the packaging does not fail, which it should imho, I would not want a package in the stable repos that does not achieve 100% of test coverage.
But this is not related to the version bump; these test failures already exist in the current version... see the comment at https://github.com/archlinux/svntogit-c … GBUILD#L38
Offline
Fair enough. I didn't check a rebuild of the current release.
Still I think it's bad practice to ignore failed tests.
Inofficial first vice president of the Rust Evangelism Strike Force
Offline
Are the failing tests even (possibly) a downstream problem?
Don't forget that unit tests predominantly test the test: If there's a BS assumption in the test, the failing test just reveals that there was a BS assumption in the test.
Online
Look, I'm not here to debate the merits of testing up-, down- or crossstream… I've got a patch for a PKGBUILD, which doesn't regress, and I just wanted to know where I can send it to. I'm kinda disappointed to learn that I'm still supposed to "git send-email" (shudder) such things around in Arch , but it is what it is ¯\_(ツ)_/¯ I've mailed the patch to both Gnucash maintainers, and hope that they consider it, so as far as I'm concerned this topic is done.
I've marked the OP as solved.
Offline
I was more addressing Schard's "it's bad practice to ignore failed tests" itr.
As for the topic, nb. that there's https://archlinux.org/packages/communit … 4/gnucash/ from 2023-04-08 (after the community version got flagged on 2023-03-26 - when 4.14 and 5.0 were released) which itself was flagged on 2023-04-10
So since the maintainers are aware of the flag and the upgrade is trivial (by your own account), there might simply be an ongoing delibration whether to move to 4.14 or 5.0.
Online
Pages: 1