Sunday, May 18. 2008rocking hard and swimming upstreamTrackbacks
Trackback specific URI for this entry
No Trackbacks
Comments
Display comments as
(Linear | Threaded)
Would a non-expert user know whether a bug originates upstream or is caused by a customization? I guess not. In that case, providing a link to the upstream bug tracker would only complicate the bug reporting process for the user. Also, upstream developers probably don't like to be bothered with bugs introduced by customizations made by a distro.
In my opinion, the proper thing to do is encourage users to report bugs in the distro's bugtracker and let the package maintainers forward the bug reports upstream if they think it's an upstream bug. To make this efficient, it would be good to have support in the distro's bugtracker to automatically enter a bug in the upstream bugtracker and to monitor it. In theory there are an unlimited number of bug tracking systems, but in practice if you would support Debian's bugtracker, Bugzilla, SourceForge and Trac, you've got over 95% covered, I think.
Please read what I read
I especially excluded bugs where it is not clear for the user if it is an up- or downstream problem. My point where feature requests that clearly belong upstream but are not reported there just because the user doesn't understand the concept. The problem with package maintainer reporting bugs upstream is that they are constantly overworked so do not dedicate a lot of time to this. Concerning automated bug reporting: Bad idea - people will come and scream if this is just done slightly wrong.
Ah, you were talking specifically about wishlist bugs. In that case, it is often clear whether it is a distro issue (if it concerns packaging) or not. So I agree, for wishlist items this could work.
The way I see automated bug forwarding is that the package maintainer creates an account on the upstream bugtracker (one-time action), links that account to his/her distro bugtracker account and with a single button click on the distro's bugtracker can forward a bug upstream. So it's still a manually initiated action, but it takes very little effort. I think doing it this way is actually less error prone than having the packager copy-paste lots of fields. I think part of the reason why packager maintainers don't always forward bugs upstream is that it often takes a lot of effort. If it could be done in a couple of seconds, even an overworked package maintainer would find it worthwhile to do it.
Hmmm yes you have got a good point there.
So one of the main issues is actually: It is too much work to forward a bug upstream right now. And yes your proposed way would likely work. Will think about that a little longer and pitch it to the right people. Thanks for the input. |
Amarok LinksCalendarQuicksearchCategoriesBlog Administration |