[aur-general] Duplicate Package Names
Too many recent packages have been named the same as the "official package", this could easily get out of hand, what can be done about it? Discussion Points: - better education and parenting? - better message on AUR page? - code to detect package name and disallow/delete package? - hit uploader with big stick? - faster updates of "offical" Repo packages?
On Sat, Sep 27, 2008 at 9:12 PM, Jud <jud@judfilm.net> wrote:
Too many recent packages have been named the same as the "official package", this could easily get out of hand, what can be done about it?
Discussion Points: - better education and parenting? Always important, but how to get the point across?
- better message on AUR page? - code to detect package name and disallow/delete package? - hit uploader with big stick? I think these might be good -- perhaps a message saying that it is discouraged, and a confirmation page. Bonus points if the package name is linked up and would point the user at the appropriate package page to flag as out of date.
- faster updates of "offical" Repo packages? I don't think there are many options down this road.
On Sun, Sep 28, 2008 at 11:12:10AM +1000, Jud wrote:
Too many recent packages have been named the same as the "official package", this could easily get out of hand, what can be done about it?
Discussion Points: - better education and parenting? - better message on AUR page?
A better message can be easily done. I think that would also cover the first point. If people don't read the AUR Guidelines there isn't much else we can do.
- code to detect package name and disallow/delete package?
I don't think our current systems are designed to do this without a lot of ugly hackery.
- hit uploader with big stick?
What? Suspend their accounts or something? I think people upload those packages out of ignorance, or they are trying to help the community. I'd rather not have to whack people because their good intentions are slightly misguided.
- faster updates of "offical" Repo packages?
Not much we can do about this in AUR. I wonder why users are not supposed to file bug reports for outdated packages. I think it would actually help people gauge how important certain packages are. It would: - be globally readable and writable. - allow users to post updated PKGBUILDs. - allow developers to provide comments on why packages are out of date - help ensure that out of date notices don't get lost in the maintainer's mountain of other email, or wherever they go.
Not much we can do about this in AUR. I wonder why users are not supposed to file bug reports for outdated packages. I think it would actually help people gauge how important certain packages are.
It would: - be globally readable and writable. - allow users to post updated PKGBUILDs. - allow developers to provide comments on why packages are out of date - help ensure that out of date notices don't get lost in the maintainer's mountain of other email, or wherever they go. I am in favor of this. Or maybe instead of having people report it in that manner, have archweb automatically file a bug report when a user marks it out of date.
On Thu, Oct 16, 2008 at 11:04 PM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
Not much we can do about this in AUR. I wonder why users are not supposed to file bug reports for outdated packages. I think it would actually help people gauge how important certain packages are.
It would: - be globally readable and writable. - allow users to post updated PKGBUILDs. - allow developers to provide comments on why packages are out of date - help ensure that out of date notices don't get lost in the maintainer's mountain of other email, or wherever they go. I am in favor of this. Or maybe instead of having people report it in that manner, have archweb automatically file a bug report when a user marks it out of date.
why? The maintainer gets automatically notified when a package is flagged out of date. If no maintainer is set a message is send to the arch-dev mailing list. IMHO it would be really annoying to clutter the bugtracker with such reports. Ronald
On Thu, Oct 16, 2008 at 11:52:45PM +0200, Ronald van Haren wrote:
On Thu, Oct 16, 2008 at 11:04 PM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
Not much we can do about this in AUR. I wonder why users are not supposed to file bug reports for outdated packages. I think it would actually help people gauge how important certain packages are.
It would: - be globally readable and writable. - allow users to post updated PKGBUILDs. - allow developers to provide comments on why packages are out of date - help ensure that out of date notices don't get lost in the maintainer's mountain of other email, or wherever they go. I am in favor of this. Or maybe instead of having people report it in that manner, have archweb automatically file a bug report when a user marks it out of date.
why? The maintainer gets automatically notified when a package is flagged out of date. If no maintainer is set a message is send to the arch-dev mailing list. IMHO it would be really annoying to clutter the bugtracker with such reports.
Putting them in the bug tracker could help cover the seemingly common occurrance of that message getting lost in a pile of other mail. Sure the bug reports can also be pushed aside or ignored, but at least they can be searched, discussed, and tracked in a more centralised and convenient fashion.
On Thu, Oct 16, 2008 at 5:46 PM, Loui <louipc.ist@gmail.com> wrote:
On Thu, Oct 16, 2008 at 11:52:45PM +0200, Ronald van Haren wrote:
On Thu, Oct 16, 2008 at 11:04 PM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
Not much we can do about this in AUR. I wonder why users are not supposed to file bug reports for outdated packages. I think it would actually help people gauge how important certain packages are.
It would: - be globally readable and writable. - allow users to post updated PKGBUILDs. - allow developers to provide comments on why packages are out of date - help ensure that out of date notices don't get lost in the maintainer's mountain of other email, or wherever they go. I am in favor of this. Or maybe instead of having people report it in that manner, have archweb automatically file a bug report when a user marks it out of date.
why? The maintainer gets automatically notified when a package is flagged out of date. If no maintainer is set a message is send to the arch-dev mailing list. IMHO it would be really annoying to clutter the bugtracker with such reports.
Putting them in the bug tracker could help cover the seemingly common occurrance of that message getting lost in a pile of other mail.
They rarely get lost. We also have a list of packages that we see every time we log into the dev backend.
Sure the bug reports can also be pushed aside or ignored, but at least they can be searched, discussed, and tracked in a more centralised and convenient fashion.
I don't think discussion is the issue here. The issue is manpower and desire and real life and all that fun stuff. If instead of people saying "OMG out of date" they would send us successful testing results from their side. Regardless, this is the aur list, and we shouldn't be talking about this. Suffice to say that if a user has a "faster update" to a package, it's a better idea to mail it along to the official maintainer instead of putting it in the AUR
Hello I bring this discussion up again, because to me the problem of official out-of-date packages is still not solved. I will take the example of clanlib. Here are the informations of the package : clanlib 0.8.0-1 Architecture: i686 Repository: Extra Description: A multi-platform game development library Upstream URL: http://www.clanlib.org/ Maintainer: None LastUpdated: 2008-09-13 Required By (0): The last version is 0.8.1 released on March 12 2008. Without anyone to contact, I filed a bug which was closed "Reason for closing: Not a bug". I understand the reason for closing, but I feel that I have no other mean to discuss this. There is no information about the reason of out-of-date so long. There is no developer to contact. It seems to me that the official packages miss an official discussion page, like there is in the AUR. This would be a place where maintainer could explain why a package is still out-of-date (The package needs another package which conflicts with a main package), get user feedback, be posted contributed corrected PKGBUILD, and whatsoever 2008/10/17 Aaron Griffin <aaronmgriffin@gmail.com>:
I don't think discussion is the issue here. The issue is manpower and desire and real life and all that fun stuff. If instead of people saying "OMG out of date" they would send us successful testing results from their side.
If it was possible to say to the community what goes wrong for a package, this would be a gain of time. I could pretty well make this new version (this works for precompiled version) but this would be in a particular situation, and I imagine that here there is a trick. So sending a new PKGBUILD (with probably only the version changed) would be helpless. 2008/10/17 Aaron Griffin <aaronmgriffin@gmail.com>:
Regardless, this is the aur list, and we shouldn't be talking about this. Suffice to say that if a user has a "faster update" to a package, it's a better idea to mail it along to the official maintainer instead of putting it in the AUR
There is no profile for developers, so it is hard to find the corresponding mail address. It is true that for this case I am very temped to post a duplicate package on the AUR. My package needs the new version and I don't want to wait for the sky to fall over my head. It is probably the case for the other duplicate packages. Cilyan Olowen
On Tue, Oct 21, 2008 at 9:13 PM, Cilyan Olowen <gaknar@gmail.com> wrote:
2008/10/17 Aaron Griffin <aaronmgriffin@gmail.com>:
Regardless, this is the aur list, and we shouldn't be talking about this. Suffice to say that if a user has a "faster update" to a package, it's a better idea to mail it along to the official maintainer instead of putting it in the AUR
There is no profile for developers, so it is hard to find the corresponding mail address.
see http://www.archlinux.org/developers/
It is true that for this case I am very temped to post a duplicate package on the AUR. My package needs the new version and I don't want to wait for the sky to fall over my head. It is probably the case for the other duplicate packages.
most of the time the miantainer is set in the PKGBUILD itstelf on the first line. For clanlib however this won't bring you any further as the maintainer mentioned there hasn't been active for some time now. Either way, you can always contact one of the other devs if a package is really lagging like this one and he can take care of it or forward it to the other devs. If it is just an easy rebuild I can take care of clanlib tomorrow, but please stop this discussion here. Ronald
If it is just an easy rebuild I can take care of clanlib tomorrow, but please stop this discussion here.
I don't care about ClanLib (IMVHO it is a dependency of no packages, it should be taken to community or unsupported) it was just an example. What I meant by my message was that all these informations are hard to gather and are an obstacle to communication to me. The real goal of my message was more "Are you sure there is nothing we can do to improve communication between developers and contributors?". This would be in order to save trees not making too much sticks...
Personally in the case of "out of date for a long time with no maintainer", I would bring the discussion to arch-general mailing list, which most devs read. You could easily post updated pkgbuilds there, and so forth.
On Thu, Oct 16, 2008 at 4:52 PM, Ronald van Haren <pressh@gmail.com> wrote:
On Thu, Oct 16, 2008 at 11:04 PM, Daenyth Blank <daenyth+arch@gmail.com> wrote:
Not much we can do about this in AUR. I wonder why users are not supposed to file bug reports for outdated packages. I think it would actually help people gauge how important certain packages are.
It would: - be globally readable and writable. - allow users to post updated PKGBUILDs. - allow developers to provide comments on why packages are out of date - help ensure that out of date notices don't get lost in the maintainer's mountain of other email, or wherever they go. I am in favor of this. Or maybe instead of having people report it in that manner, have archweb automatically file a bug report when a user marks it out of date.
why? The maintainer gets automatically notified when a package is flagged out of date. If no maintainer is set a message is send to the arch-dev mailing list. IMHO it would be really annoying to clutter the bugtracker with such reports.
Agreed.
Loui wrote:
On Sun, Sep 28, 2008 at 11:12:10AM +1000, Jud wrote:
Too many recent packages have been named the same as the "official package", this could easily get out of hand, what can be done about it?
Discussion Points: - better education and parenting? - better message on AUR page?
A better message can be easily done. I think that would also cover the first point. If people don't read the AUR Guidelines there isn't much else we can do.
- code to detect package name and disallow/delete package?
I don't think our current systems are designed to do this without a lot of ugly hackery.
- hit uploader with big stick?
What? Suspend their accounts or something? I think people upload those packages out of ignorance, or they are trying to help the community. I'd rather not have to whack people because their good intentions are slightly misguided.
- faster updates of "offical" Repo packages?
Not much we can do about this in AUR. I wonder why users are not supposed to file bug reports for outdated packages. I think it would actually help people gauge how important certain packages are.
It would: - be globally readable and writable. - allow users to post updated PKGBUILDs. - allow developers to provide comments on why packages are out of date - help ensure that out of date notices don't get lost in the maintainer's mountain of other email, or wherever they go.
I wholeheartedly agree... allowing users to submit bug reports (or something similar) for outdated packages would at least allow us to judge package popularity.
participants (7)
-
Aaron Griffin
-
Cilyan Olowen
-
Daenyth Blank
-
Ghost
-
Jud
-
Loui
-
Ronald van Haren