[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Random BTS musings

On Sun, Jul 29, 2001 at 01:49:44PM +1000, Anthony Towns wrote:

> Is it just me, or do bug reports seem to be tending more towards the
> wrong end of the:
> [...]
> scale? If we get more and more users, or more and more newbie users,

It is not just you.

> or more and more users who want an "information appliance" instead of
> a computer, I can only see this getting worse. Is there anything we
> can do to improve this? Maybe encouraging people who can only provide
> (5)-(8) to discuss it on a mailing list (or with members of their local
> LUG?) first is the way to go? (Having bugs filed as RC that don't at least
> meet (4) is quite annoying from my perspective, at any rate)
> Thoughts? Is this worth thinking about or trying to ameliorate?

I agree that having possible un-bugs as RC is frustrating.  One possible
solution would be to limit the maximum severity of unreproducible, poorly
reported bugs (perhaps to important or normal).  That wouldn't solve the
problem for maintainers having to sort through these bugs, though.

I don't think we should discourage bug reports, but try to make them easier to
manage.  The BTS really needs a redesign, to learn from the other bug reporting
systems out there.  Bugzilla, though I have no hands-on experience with it,
appears to do a much better job of classifying bugs.  Each bug has separate
categorization for status, resolution, priority and severity. See:


Debbugs supports some of this categorization with tagging, some with severity,
but it isn't nearly as effective.  For Debian, it would be useful to have at
least a bug status that is separate from the current "severity" (which is
directly tied to a notion of "priority").

Here are some ideas for status values, in roughly bug-lifetime order:

	The bug has been reported and filed, but not yet examined by anyone
(maintainer, QA, etc.).  QA and others could regularly scan a list of these
bugs, weed out obvious user error, correct package assignments, etc.

	The bug has been examined, but it is not clear that it is a real bug.
More information may be needed from the bug submitter.  Replaces the tags of
the same names.

	Just what it says.  The report is rejected (user error, etc.).

	The bug has been acknowledged as legitimate and correctly assigned,
preferably by a developer, and needs to be fixed.  QA and others could
regularly scan a list of verified bugs older than a certain time, to locate
real bugs which are not being addressed.

	A real bug which, for whatever reason, cannot or will not be fixed.
Replaces the tag of the same name.

	Someone has taken responsibility for fixing the bug, either the
maintainer or someone else with the necessary skills.  Bugs which are assigned
for a long time, but not fixed, should be reclassified as Verified, so that
someone else can handle them.  It would be nice if the BTS were to track who is
currently responsible for a bug, but just knowing that someone is working on it
would be useful in itself.

	The bug has been fixed, but the fix is not yet in Debian.  The current
"fixed" tag should be renamed to "NMU".

	The fix has been uploaded (to unstable).  If the fix is not successful, the
bug should return to "Assigned".

	The fix has been verified by someone else.  This may not be practical
without more infrastructure to make code review easy, but it would help us to
ensure that bugs are fixed correctly.

	The fix has been included in a stable release.  This may only be useful
if we start actively paying attention to which version a bug was first reported in.
Then, we could use a "Released" status to determine which bugs are present in the
current stable release, decide when to release again, etc.

A quick and confusing flow chart illustrating these status ideas can be found


I would like to create a system where bug status could be tracked without
having to read through the messages in the BTS.  Each time I scan through the
RC bug list, or skim debian-bugs-dist, I have to read the messages for the bug
to determine what is happening with it.  If we were to keep track of bug
status, more effort could be focused on the bugs which need the most help.

This is surely beyond the original topic, but I think that a lot of improvement
could be made here.  Is any development effort still going into debbugs, or
should a replacement be written?

 - mdz

Attachment: pgp0wtWcuVORG.pgp
Description: PGP signature

Reply to: