You are not logged in.

#1 2016-07-07 18:24:34

Poohblah
Member
Registered: 2014-06-11
Posts: 11

Is it appropriate to create an AUR package that patches a single bug?

Ok, sorry the subject is a bit non-descriptive, but there seems to be a character limit on subject length.

Anyway, the [extra] package cyrus-sasl-gssapi has a bug that affects me, several people in my organization, and at least a small number of people in other organizations (the bug affects a much larger number of people on other distros, but that's irrelevant here). Almost two years ago, I filed a bug with Arch Linux, since the issue has remained unpatched upstream for over since January 2011, with no sign of being fixed upstream, even though there are a handful of patches for this bug floating around in the wild (I know the maintainers of Fedora have patched this bug, for instance).

As you can see by reading the comments on the bug report I filed against Arch, another member of the community was kind enough to rebuild the Arch Linux package using one of these patches. However, the maintainer of the official Arch package has not yet incorporated this patch into the build, so now I would like to take this patch and apply it to the base PKGBUILD to create an AUR package so that I don't have to have an untracked, manually-built package installed on my system, and so that I can share this patched package with the Arch community, as I know at least some of my co-workers will appreciate it and hopefully some other people will as well.

However, I'm not sure if simply applying a two-line patch to an official package is sufficient to meet the guidelines set out in the AUR Rules of Submission.  In particular, I am concerned about this clause: "If any version of it exists, do not submit the package. If the official package is out-of-date, flag it as such. If the official package is broken or is lacking a feature, then please file a bug report."  However, I've already filed a bug report with no sign of resolution in almost 2 years, so I'm not sure if it's appropriate to move forward despite this clause. Furthermore, even if it is, I'm not sure how I might name such a package.  "cyrus-sasl-gssapi-patched" seems too generic, and "cyrus-sasl-gssapi-without-maxssf-bug" or something similar seems too verbose.

Can the AUR community provide any insight into this issue?

Offline

#2 2016-07-07 18:42:27

drcouzelis
Member
From: Connecticut, USA
Registered: 2009-11-09
Posts: 4,092
Website

Re: Is it appropriate to create an AUR package that patches a single bug?

Has the official package maintainer, Andreas Radke, been made aware of the situation? Is the communication documented anywhere?

Package link: https://www.archlinux.org/packages/extr … sl-gssapi/

Last edited by drcouzelis (2016-07-07 18:43:21)

Offline

#3 2016-07-07 19:58:00

Poohblah
Member
Registered: 2014-06-11
Posts: 11

Re: Is it appropriate to create an AUR package that patches a single bug?

The short answer is "no".

I assumed that the maintainer was notified when I filed the bug report, but I could easily be wrong, as I have no proof of this claim.

That said, it looks like the bug report was not properly linked against the package (i.e. this is the page you get when you follow the "Bug Reports" link on the package webpage). Would linking the package to the bug report trigger a notification to the maintainer? Or should I notify the maintainer more directly? If so, what's the process for this? I can't find anything on the Arch Wiki.

(By the way, I did my best to follow the Bug Reporting Guidelines when submitting the bug report.  It is possible that I missed a step, but I don't see any steps in there regarding notifying the maintainer directly in the event that a bug report is filed against a specific package.)

Offline

#4 2016-07-07 20:09:11

drcouzelis
Member
From: Connecticut, USA
Registered: 2009-11-09
Posts: 4,092
Website

Re: Is it appropriate to create an AUR package that patches a single bug?

Ok. I just sent an email to the package maintainer. smile

Offline

#5 2016-07-07 20:19:46

Poohblah
Member
Registered: 2014-06-11
Posts: 11

Re: Is it appropriate to create an AUR package that patches a single bug?

Thank you smile

How did you get the maintainer's e-mail address? All I could find were the e-mail addresses associated with the PGP key used to sign the package, but I have no idea which of those the maintainer prefers.

Last edited by Poohblah (2016-07-07 20:20:27)

Offline

#6 2016-07-07 20:25:33

Poohblah
Member
Registered: 2014-06-11
Posts: 11

Re: Is it appropriate to create an AUR package that patches a single bug?

Oh, by the way, I just realized that the maintainer is not actually Andreas Radke. Andreas is the packager, which AFAICT just means he signs the package. Jan de Groot is the maintainer, and yes, Jan de Groot was made aware of the bug report - Jan is assigned to the bug report I filed.

Offline

#7 2016-07-08 19:12:54

Poohblah
Member
Registered: 2014-06-11
Posts: 11

Re: Is it appropriate to create an AUR package that patches a single bug?

Any other feedback from any other members of the community? Looks like I'm back to square one here.

Offline

#8 2016-07-11 14:02:32

drcouzelis
Member
From: Connecticut, USA
Registered: 2009-11-09
Posts: 4,092
Website

Re: Is it appropriate to create an AUR package that patches a single bug?

As far as I can tell, in the past 24 hours the patch has been applied, the package has been updated on the official repos, and bug has been closed.

(I can't believe I'm this excited about a patch I didn't work on for a bug that doesn't affect me for a piece of software that... I don't even know what it does.)

If it is fixed, then to all involved, thank you! smile smile

Offline

#9 2016-07-11 16:53:50

Poohblah
Member
Registered: 2014-06-11
Posts: 11

Re: Is it appropriate to create an AUR package that patches a single bug?

Hmm, I don't see the updated package in the official repos. Looks like it's still on 2.1.26-7 to me. I see now, the new release of the package is in [testing].

Also, though the specific case is solved, I'm still curious about the general case. Is it appropriate to create an AUR package that patches a single bug in an official package (in the case that the maintainers refuse to apply the patch to the official package)? And if it is appropriate, what naming scheme should the package follow?

Last edited by Poohblah (2016-07-11 16:58:47)

Offline

#10 2016-07-11 17:54:56

drcouzelis
Member
From: Connecticut, USA
Registered: 2009-11-09
Posts: 4,092
Website

Re: Is it appropriate to create an AUR package that patches a single bug?

Good question. Here is the complete rule:

AUR Rules of Submission wrote:

Check the official package database for the package. If any version of it exists, do not submit the package. If the official package is out-of-date, flag it as such. If the official package is broken or is lacking a feature, then please file a bug report.

And here is my unofficial interpretation:

"Any version" means "if version 7.8 is in the repos and you want version 7.9, don't create a new PKGBUILD file and put it in the AUR". The bug-fixes version that you wanted to submit was not out-of-date and the package was not broken. It WAS lacking a feature but... Actually, I would think "lacking a feature" would be an excellent reason to use the AUR to patch a package. tongue

For this specific scenario, what you did was the best solution.

If the official package was not going to be updated, then you could have created an AUR package named "cyrus-sasl-gssapi-patched-maxssf" sounds fine to me (unofficiall). smile

EDIT: And as an example of a package in the AUR with a "patched" name, the package "rxvt-unicode-patched" has 221 votes.

Last edited by drcouzelis (2016-07-11 17:58:18)

Offline

Board footer

Powered by FluxBB