Reporting bugs upstream

[Mike A. Harris]

Pete Zaitcev wrote:

>>When I find a bug in fedora and it's obviously a bug upstream I report
>>it there. But do we have to report them in fedora bugzilla too and
>>reference the upstream bugreport? So if an other users files a bugreport
>>for the same bug and doesn't check upstream maintainer of the module
>>know it's already filed upstream. Or do we have to notify the maintainer
>>in other way?
> 
> 
> It depends on the package, but usually the Fedora maintainer taps into an
> announcement list for the upstream project, so he sees those.
> 
> This does not preclude users from filing into our bugzilla. I used to
> do this in preference to upstream. The logics behind it was that we may
> be shipping a derivative, while the upstream only wants to hear about
> bugs reproduced on the pristine upstream. Maintainer filters and reports
> upstream such bugs. It is the way kernel operated, because it used to
> be noticeably divergent from its upstream. But for most packages it's
> not needed, so your way is better.
> 
> BTW, I like dups, because users often confuse same symptoms with
> same bug. I have dozens of people reporting "my kernel said
> ``ep0 timeout''" for dozens of wildly different problems, all ending
> in the same bug which can never be resolved. I would much prefer users
> filing dups than polluting unrelated bugs. But in X, the situation
> is the opposite: 100 dups for the same thing, and Mike Harris wants
> his users to search for dups and avoid filing new ones (if I understand
> him right).

Not exactly...   For X, we experience both problems.  We get many people
reporting the exact same issue without them even bothering to check for
duplicates, in which there is no question at all that it is the exact
same issue.  That is kindof irritating, in particular when there is a
very large amount of work to do and deadlines pressing on us, and there
are a million new bug reports to wade through, and find the master dupe
for them, and close them all, etc.

However, we also get cases where someone reports "My Radeon crashes with
new X update", which then some other person tacks on "yeah, my nvidia
blah blah does that too, both with the original FCx release and with
the update", then someone else "I have the same problem too!  But with
Trident!", then another nvidia user with a totally unrelated problem
to any of the other people.  They all have their X server crashing for
various different unrelated reasons, but since the symptom sounds
similar enough, they all pile on to one big report.  Then when you ask
the original reporter for their log/config or to try something specific
to their problem - every one of them does it and reports back the
results.  Even if your advice was obviously specific to the radeon for
example, an nvidia user will try it on their card, and perhaps complain
that the option gave an error or something.

Very irritating.  However, all of the userbase does not have the same
technical skill level, or troubleshooting abilities, etc. so it is more
or less a fact of life that we have to deal with.  It's not as bad as
it was at one point though, or at least it seems a bit better now. ;o)

The biggest problem we face as far as X is concerned though, is that at
least the X server is like a part of the kernel, in the sense that it
has a bunch of hardware drivers with all the idiosyncracies of hardware
at play.   A large number of driver related problems get reported in
bugzilla, and our team does try to handle as many issues as we can,
given the resource constraints we are under, however the number of bugs
in the X server, drivers, xkb, etc. etc. vastly outstaggers the number
of manhours our entire team probably 100:1 or worse.  Many people insist
on filing bugs in our bugzilla rather than X.Org bugzilla for issues
that are not really distribution specific to Fedora Core however, and
so they pile up over time due to a combination of lack of hardware,
lack of manpower, and in many cases lack of reproduceability.

While some people are now reporting things directly to X.Org, the
majority of driver issues still hammer into our bugzilla.  The sad
thing is that many of the issues could likely be fixed in a day,
a week, or a month if reported to X.Org, simply because there are
so much more many developers and hardware available upstream, etc.

Ah well, such is the toil of bugzilla. ;)



-- 
Mike A. Harris  *  Open Source Advocate  *  http://mharris.ca
                       Proud Canadian.