Proposal: KDE CVS Commit Policy
Maks Orlovich
kde-policies@mail.kde.org
Sun, 30 Mar 2003 15:27:34 -0500
> 9. Lack of maintenance of the stable release series. Often useful, smaller
> new features are not backported, while dangerous, untested "code
> cleanups" are immediately backported. Massively introduced regressions
> are not being cleaned up in the stable branch and end up in a release
> (I can't remember a single KDE release where kio_http's proxy support
> wasn't flawed in one or another way).
A comment on this: I think a related issue is that we don't prioritize which
bugs we fix first all that well. Some bugs are far more important than others
for the users. George Staikos has done an amazing (and IMHO much
under-appreciated) job nagging developers for fixes or at least workarounds
for the really important things for 3.1.1, but it's not a one-man job to any
extent - and it'd perhaps make sense to agree to some sort of convention on
how to mark those bugs. Perhaps just going the critical/grave route and
relaxing the traditional definitions...
I also think we should be more careful with WORKSFORME closing of bugs -- at
least some real & serious bugs were closed like that distorting the
perception of state of things (having a developer think that an existing, and
serious bug is fixed is pretty dangerous). At the very least, if people from
2 or 3 different systems/distros report the same issue, it's probably on our
ends and not packager ends.