On 07/18/2013 03:41 PM, Karol Blazewicz wrote:

On Thu, Jul 18, 2013 at 11:24 PM, Alexander Rødseth <rodseth@gmail.com> wrote:

There are quite a few orphaned packages nobody is maintaining or updating even though they have updated PKGBUILDs posted in the comments. Users can always ask for a package to be orphaned.

Not that my opinion has much weight around here, but I think it's a bad idea to wanton orphan packages like that. If a package is actually still working and building, etc; it might cause more panic than good.. </$22.22 ( Adjusted for inflation) > -- John D Jones III Perl/Javascript/Systemd Zealot unixgeek1972@gmail.com http://www.zoelife4u.org/

Thank you for the comments, appreciate it! About causing panic when orphaning: the package will continue to be available and work even if it's orphaned, but now there is a chance that someone will pick up the thread where the previous maintainer left of and improve and update the PKGBUILD. If someone panics from this, perhaps it's still a good trade? The reason I'm asking is because I suspect that I'm more fond of cleaning up old cruft than the average TU/Dev, and I don't wish to orphan or delete packages in a way that is perceived as rash. In my opinion, being relatively quick to orphan, but hesitant to delete, should result in a better AUR repository for everyone, as long as the criterions for disowning packages is somewhat conservative. I would say a user being inactive for more than a year is quite sufficient. A comment from another TU or Dev would be especially helpful. -- Sincerely, Alexander Rødseth xyproto / TU

On Fri 19 Jul 2013 at 01:09, Maxime GAUDUIN wrote:

On Fri, Jul 19, 2013 at 12:24 AM, Alexander Rødseth <rodseth@gmail.com>wrote:

...

[...] About causing panic when orphaning: the package will continue to be available and work even if it's orphaned, but now there is a chance that someone will pick up the thread where the previous maintainer left of and improve and update the PKGBUILD. If someone panics from this, perhaps it's still a good trade?

The reason I'm asking is because I suspect that I'm more fond of cleaning up old cruft than the average TU/Dev, and I don't wish to orphan or delete packages in a way that is perceived as rash.

In my opinion, being relatively quick to orphan, but hesitant to delete, should result in a better AUR repository for everyone, as long as the criterions for disowning packages is somewhat conservative. I would say a user being inactive for more than a year is quite sufficient.

A comment from another TU or Dev would be especially helpful.

I support this too. Orphaning is reversible and will most (preferably always :P) of the time lead to an improvement of the PKGBUILD, I see no reason not to do it when the user is clearly inactive. As for deletions, I would also tend to delete old stuff more easily, but I understand and respect the wish to keep placeholders in the AUR.

I think we should orphan all packages of a maintainer that is clearly not active, especially if they have packages marked out of date for a long time or requests in the comments that have not been addressed. If in doubt, an email to them won't hurt, but in my experience this has always lead to disowning them. This will then allow other users to make improvements immediately rather than having to wait two weeks for the current maintainer to (not) reply, and then make a request here and wait to have that actioned. -- Jonathan Steel

On Sat 20 Jul 2013 at 07:49, Frederik "Freso" S. Olesen wrote:

Den 19-07-2013 22:45, Jonathan Steel skrev:

...

[...] If in doubt, an email to them won't hurt, but in my experience this has always lead to disowning them.

FWIW, I've sent e-mails to maintainers that replied back and got around to updating their packages. You likely don't hear about these, as the requests (seldomly, anyway) come to this mailing list. It's true that I have more often not gotten a reply, but it does happen that I have.

We are talking about users who have been inactive for a more than a year, especially ones that have ignored flags/comments for a long time; have users really come back after a year or so just because you emailed them? -- Jonathan Steel

On Sat, Jul 20, 2013 at 1:44 PM, Willem <willemw12@gmail.com> wrote:

What about the package users?

Will users who have voted for a package be emailed that the package is about to be deleted (automatically)? And be asked whether he wants to become the maintainer?

I'm confused. Are we talking about orphaning or deleting packages? Even if there's no comment left about the package being orphaned, you can periodically run e.g. aurphan to check. There's also https://bugs.archlinux.org/task/31851

Den 20-07-2013 09:39, Jonathan Steel skrev:

On Sat 20 Jul 2013 at 07:49, Frederik "Freso" S. Olesen wrote:

...

FWIW, I've sent e-mails to maintainers that replied back and got around to updating their packages. You likely don't hear about these, as the requests (seldomly, anyway) come to this mailing list. It's true that I have more often not gotten a reply, but it does happen that I have.

We are talking about users who have been inactive for a more than a year, especially ones that have ignored flags/comments for a long time; have users really come back after a year or so just because you emailed them?

I have had at least one response from someone who was no longer actively using Arch Linux and had been inactive for some time, and then went and disowned their package. Whether they'd been gone for more than a year by the time they were contacted, I cannot remember, and I doubt I'd be able to check that now if I went and looked up who it was. -- Frederik "Freso" S. Olesen <http://freso.dk/>

On Thu, Jul 18, 2013 at 6:25 PM, Geoffrey van Wyk <geoffrey.vanwyk@bernadine.biz> wrote:

On Thursday 18 July 2013 14:56:40 Anatol Pomozov wrote:

...

"if a package is marked out-of-date for more than 3 months then it disowns automatically"

Looks like a good policy.

As long as it fires off a warning email or two (at 1 month out-of-date, and again at 2), I don't see any problems with doing that.

On 07/18/2013 04:36 PM, Geoffrey van Wyk wrote:

On Thursday 18 July 2013 19:26:33 Daniel Micay wrote:

And a last warning 5 days before orphanage.

I like this... though I think 6 months would be better than 3 on the initial Orphaning. There should for sure be a final "DUDE!!! Fix your crap" Warning email sent to the maintainer, like above, perhaps a 7 day warning? -- John D Jones III Perl/Javascript/Systemd Zealot unixgeek1972@gmail.com http://www.zoelife4u.org/

On 19 July 2013 11:02, Phillip Smith <lists@fukawi2.nl> wrote:

A daily process to automatically disown abandoned packages would be great, but it needs to account for the last modified date too:

I realized just as I hit send that a package can never be flagged out of date longer than the last updated date. Derp.

On Fri, Jul 19, 2013 at 4:39 AM, Anatol Pomozov <anatol.pomozov@gmail.com> wrote:

Hi

On Thu, Jul 18, 2013 at 6:02 PM, Phillip Smith <lists@fukawi2.nl> wrote:

...

On 19 July 2013 10:28, John D Jones III <unixgeek1972@gmail.com> wrote:

...

I like this... though I think 6 months would be better than 3 on the initial Orphaning. There should for sure be a final "DUDE!!! Fix your crap" Warning email sent to the maintainer, like above, perhaps a 7 day warning?

A daily process to automatically disown abandoned packages would be great, but it needs to account for the last modified date too:

1. Find packages flagged out of date > 1 month and last modified > 1 month == Send reminder email. (Package out of date, please update). 2. Find packages flagged out of date > 2 month and last modified > 1 month == Send warning email. (Package still out of date, automatic disown in 1 month). 3. Find packages flagged out of date > 2.75 month and last modified > 1 month == Send warning email. (Last chance!). 4. Find packages flagged out of date > 3 month and last modified > 1 month == Disown, send notification email, add comment so others who are subscribed to comments is aware it has been disowned (You lost it).

Is there any style checker tool for PKGBUILD files? Something similar to lint? If yes then it worth checking for style violations as well, e.g. "PKGBUILD does not have package()" function.

makepkg?

So bot can send a note to owner and mark the package somehow e.g. "Needs improvements" If no improvement has been done during some period then the package is disowned.

Hi On Thu, Jul 18, 2013 at 8:02 PM, Agustin Ferrario <agustin.ferrario@hotmail.com.ar> wrote:

On 07/18/2013 11:39 PM, Anatol Pomozov wrote:

...

Hi

Is there any style checker tool for PKGBUILD files? Something similar to lint? If yes then it worth checking for style violations as well, e.g. "PKGBUILD does not have package()" function. So bot can send a note to owner and mark the package somehow e.g. "Needs improvements" If no improvement has been done during some period then the package is disowned.

Namcap...

Yeah, namcap looks like what I kept in mind. I think there should be a bot that nags about style violation + disowning if no improvements were made during long period of time (e.g. 1 year). This should improve AUR packages quality.

On Fri, Jul 19, 2013 at 1:26 AM, Daniel Micay <danielmicay@gmail.com> wrote:

As long as it fires off a warning email or two (at 1 month out-of-date, and again at 2), I don't see any problems with doing that.

This. We have a policy of "e-mail the maintainer and wait 2 weeks for response" so disowning a package w/o trying to contact the maintainer seems to contradict the way we've done things for awhile.

On 18 July 2013 23:24, Alexander Rødseth <rodseth@gmail.com> wrote:

Hi,

The two packages that are out of date has not been updated for three years.

My plan is to orphan all his packages if nobody thinks that's a horribly bad idea.

I'm also interested in comments about what should be done for similar situations in the future. I assume most users would be happy just to see the pacakges being updated instead of hoarded and would think it was fine if TUs just orphan them after a similar investigation of the situation.

+1 from me. If the user is inactive for long time and the packages apparently needs care (this is actually quite important, because you can have VCS packages not updated for a long time and they will still work), they should be orhpaned. On 18 July 2013 23:56, Anatol Pomozov <anatol.pomozov@gmail.com> wrote:

Hi IMHO Arch developers should be more proactive in disowning de-facto orphaned packages. Something like "if a package is marked out-of-date for more than 3 months then it disowns automatically". Similar rule can be applied to all packages of inactive users.

I'm strongly against the idea of automatic orphaning. Sometimes the package may be outdated simply because the new version doesn't work or has some serious issues. Lukas

On 19 July 2013 15:39, Anatol Pomozov <anatol.pomozov@gmail.com> wrote:

Hi

In this case the maintainer should unmark the package and clearly explain in comments why the package cannot be upgraded to the new version. Ideally maintainer should also work with upstream on resolving the issues. But silently leave the package in "out-of-date" state forever is not the best solution neither.

The problem is that the users don't read. It's happening all the time. I had a package once or twice that couldn't be updated but even though I stated it in the comments people were still marking it out of date. BTW, leaving the package as out of date is a good reminder that the maintainer should check if the problem was fixed every now and then. On 19 July 2013 15:45, Doug Newgard <scimmia22@outlook.com> wrote:

----------------------------------------

Do you consider clicking a link twice (once to unflag, again to reflag) every 6 months or so overly burdensome?

It's simple to forget about that and it's nonsense to do it just to keep your package. I don't think it's any better than the current approach with emailing the maintainer first. As I mentioned earlier, I don't have anything against mass orphaning if the maintainer is clearly inactive and his package has problems, but this automatic approach doesn't take that into account. Lukas

On 19 July 2013 20:40, Anatol Pomozov <anatol.pomozov@gmail.com> wrote:

Hi

And that is what email notifications for.

They are good when someone flags our package out of date. But after that it is easy to forget about it.

I don't think that *manual* mass orphaning is a good idea. There should be an automatic way to do this.

It doesn't happen that often. A viable solution for me would be sending an automated email to trusted users about possibly inactive accounts and it would be up to TUs to decide whether the packages should be orphaned (or even deleted).

But if you think that "package was not fixed for 6 months" is a bad indicator of user inactivity what would be a good indicator then?

The problem is that out of date doesn't mean broken. On 19 July 2013 20:54, cyberdupo56 <cyberdupo56@gmail.com> wrote:

It sounds like there are two separate problems here. We need to clarify what the out-of-date marker on the AUR means. Does it mean it is out of date with upstream, or out of date with the newest viable/working version? Perhaps there should be another field added, so we can mark packages as out-of-date-but-cannot-be-upgraded-at-this-time. That way, if a package is marked out-of-date but not cannot-be-upgraded with no activity for a period of time, it should be safe to orphan after sending a warning email.

I think a flag "broken" would be more appropriate. I guess it would be better than reporting problems in comments as it's done now, because everyone would immediately know if there's some problem with a package without having to read the comments and the PKGBUILD (which everyone should do anyway). Lukas

On Fri, Jul 19, 2013 at 2:16 PM, Lukas Jirkovsky <l.jirkovsky@gmail.com> wrote:

The problem is that the users don't read. It's happening all the time. I had a package once or twice that couldn't be updated but even though I stated it in the comments people were still marking it out of date. BTW, leaving the package as out of date is a good reminder that the maintainer should check if the problem was fixed every now and then.

It sounds like there are two separate problems here. We need to clarify what the out-of-date marker on the AUR means. Does it mean it is out of date with upstream, or out of date with the newest viable/working version? Perhaps there should be another field added, so we can mark packages as out-of-date-but-cannot-be-upgraded-at-this-time. That way, if a package is marked out-of-date but not cannot-be-upgraded with no activity for a period of time, it should be safe to orphan after sending a warning email. The second problem is that users don't read. Frankly, I don't have a magic solution for this, but could there be a way to limit how many times a user can mark out-of-date or a package can be marked out-of-date until they prove their reading abilities? Actually, a 12 or 24 hour limit between marking and unmarking out of date for everyone except the package owner and TUs on each package seems reasonable, now that I think about it, possibly limited to "hot" packages for which this is a problem.