Re: Bugs in Bugzilla

[Maciej Stachowiak <mjs eazel com>]

Joe Shaw <joe helixcode com> writes:

> > Well, it makes it impossible to follow up with the user if the info in
> > the bug report is insufficient. Or for the user to be notified when
> > the bug is closed, so it can be verified.
> 
> I admit, this is a disadvantage. But what is more important: people
> submitting bugs when something crashes or not at all?

I wouldn't mind hearing about it if someone has a crash, even without
contact info, but I don't want that mixed up with the real bug
reports.

> Remember, bug-buddy comes up anytime an application crashes. Most of the
> users are happy to submit a bug report because it's quick and easy. Yeah,
> it results in a lot of duplicated and occasionally bad bugs, but it's more
> of a resource than a detriment.

Some people disagree. Personally I am not sure. I know that for
Nautilus we have not had a single useful bug report submitted by
bug-buddy. But there were only about 81 total so far, and I can't rule
out the possibility that someday there will be a useful one.

> > I think mandating a valid email address is acceptable. If we want to
> > be able to get feedback other than that, it could be filed as "problem
> > reports" which are only turned into real bug reports when someone
> > looks at them and distills relevant information. But I don't think
> > that is particularly critical.
> 
> Mandating a valid email address is good, but requiring someone to register
> with the bug tracker and to remember a password to use it isn't.

Unfortunately some kind of registration process is the only real way
to validate an email address. It may be true that the password is
overkill, though.

> Although, if we do have one for "bugs" and something (that bug-buddy
> sends to) for "problem reports", then this isn't an issue. The only
> issue then is that someone is going to have to wade through the
> "problem reports" and promote them to "bugs" if applicable. And if
> there's that much volume involved and lackluster maintainers or
> problem-report-people, then what do we gain, really?

Well, the lower-quality problem reports will by default not drown out
the bugs. A lame person looking at only the bug reports and ignoring
the problem reports is a win relative to a lame person not looking at
anything because of the sheer volume.


One interesting thing to note is that most large proprietary software
companies have two levels of bug tracking. There's an external system
where users submit trouble tickets to support, and only support, QA
and the developers can generate bug reports in the real internal bug
tracker seen by developers. When support turns a trouble ticket into a
bug report, they create a cross-reference. I don't think we should do
something like that for Gnome, but it's something worth noting.

 - Maciej