Re: The way to the next dpkg release
On Wed, Jan 18, 2006 at 05:12:58PM -0800, Russ Allbery wrote:
> One of the difficulties that I see with dpkg at the moment is that it has
> a very large bug list. Many of those bugs are very old, the use of tags
> is sometimes different from bug to bug, and the database could probably
> use either a good set of usertags or some bug title massaging (or possibly
> both).
>
> I've been looking a little for a package to do bug triage on when I have
> the time, and this always looked like an excellent target if I had some
> guidelines as to what the maintainers wanted. Now seems like a great
> opportunity to come up with some guidelines so that some of the rest of us
> can pitch in and start getting the bug list down to something more
> reasonable.
>
> So, any general thoughts on a bug policy? Use of confirmed / patch / etc?
> Categories of bugs that we should create with usertags?
One thing you probably could safely do it converting the current subject
based tags to real usertags. That is mostly a assignment to certain
programs and a few others (like assert).
I think the use of unreproducible/moreinfo/patch/wontfix should be obvious.
IMHO patch should be used rather defensively i.e. only if there is really a
patch in there that is almost ready for inclusion.
What do the others think about the use of confirmed? It could be used
to have the bugs categorised in three categories (confirmed ->
unclassified -> unreproducible) instead of two (unclassified ->
unreproducible).
In case we want to use this tag I would propose
that to tag a bug confirmed you need to add an explanation
how to reproduce it. Especially for the old bugs in dpkg this sometimes
is a big problem (as I know from my own triaging efforts on dpkg-dev)
and so it might be worthwile to seperate those bugs that are
definetly reproducible.
Gruesse,
--
Frank Lichtenheld <djpig@debian.org>
www: http://www.djpig.de/
Reply to: