bug reporting.. bugzilla

Martin Pool mbp at samba.org
Mon Dec 9 04:43:00 EST 2002

On  5 Dec 2002, Sriram Ramkrishna <sramkris at ichips.intel.com> wrote:
> What we do in the GNOME project is to find volunteers to run triage
> and catalog the bugs.  If you have a "bugmaster" position who could
> coordinate something like this.  

The situation is rather different to GNOME, as jw notes: the code is
not growing very quickly, and there are far fewer developers available
to work on it.

> > There used to be another bug reporting system but it was being ignored
> > so Martin turned it off.

Yes, I did.  I think the system was broken by some kind of
infrastructure migration, and since nobody seemed to use it I put up
this page rather than fixing it:


One problem was that it used tridge's Jitterbug system, which is a
nice program but a bit harder to learn than Bugzilla, or at least less
familiar to most people.  Also, because it runs mostly over email, it
quickly fills up with spam.

But the main thing that discouraged me from maintaining it was just
that most of the entries were not valid bugs.  We had large numbers of 

 - misunderstandings of how to use rsync (operator error)

 - massively incomplete reports (e.g. just "it fails", without any
   error message.)

 - architectural limitations (e.g. upfront scan)

 - other junk entries

and in addition many of them were redundantly reported.  I think
probably >90% of entries were like this.

You can see this to a lesser extent in the FAQ-O-Matic:


Too many people fail to realize that filing a useful bug is actuallly
a lot of work and requires that the reporter actually put a bit of
thought into the problem. </rant>

So the database was full of things that were not really bugs, which
made it pretty useless either for people who wanted to find out about
a bug they might be experiencing, or developers wanting to know how
many bugs there are.

I'm sure GNOME has had this too, but if I understand correctly they
reduced their junk bug count in the first place by throwing out the
whole database, and then by putting a lot of work into triage and

I think a better way go forward would be for volunteers to help
maintain an FAQ.  This might be a good way to address common problems,
whether they result from misunderstandings or from program errors.  It
could be in FAQ-O-Matic or something else.  

Since new bug reports are relatively rare, but problems and
misunderstandings seem to occur repeatedly I think this would be the
most useful way to get all the information in one place.

So who's interested in working on that? 


More information about the rsync mailing list