bugzilla triage madness :-/

John Poelstra poelstra at redhat.com
Tue Apr 8 04:23:28 UTC 2008

Pekka Savola said the following on 04/04/2008 11:07 AM Pacific Time:
 > On Fri, 4 Apr 2008, Michael Schwendt wrote:
 >> What do I care? When I filed the bug, the product was fresh and
 >> maintained,
 >> wasn't it? But more than a year later I'm no longer willing to spend 
 >> on the same issues without a single sign of life from the package owner.

I think this is completely reasonable.  I'd do the same thing and would 
thus close the bug.

 > That seems like one of the fundamental problems here; I agree with you.
 > Having filed bugs, and seeing them get closed 5 years later with "this
 > product is no longer supported", and no developer response in between is
 > rather disheartening.

This raises several important issues that have been masked as the 
outstanding bug count in Fedora grew and grew until January 2008 when a 
few of us thought something needed to be done.

1) Do we really have enough package maintainers?
   --If not, why?
   --Will we ever?
   --Is this the wrong issue to focus on?

2) Is it reasonable for a bug reporter or a maintainer to expect that 
EVERY legitimate bug filed will be fixed?
    --If not, how does Fedora as project adjust bug reporter expectations?
    --Could Fedora provide guidelines around which types of bug reports 
are more desirable than others--assuming bug reporters have finite time 
as well?

3) With finite resources is it wiser to attempt to fix all the bugs in 
Fedora Core 1 to 6 or polish Fedora 9?
    --The EOL releases have definitely been a distraction and potential 
time waster for new triagers

4) Is it actually a *better* user experience to routinely close open 
bugs that aren't getting fixed then let them sit for several years with 
no response?  In other words is it better to disappoint someone a month 
or two after reporting a bug or a year?

 > A suggestion for a saner bugzilla automation: close packages in
 > _NEEDINFO_ stage unless there is response from the reporter.

How would you do this automatically?  Looking through 12,000 open bugs 
is not worth the time.

 > If the developer believes the issue is fixed or there is a high
 > likelyhood that it has already been fixed in a newer version, s/he
 > should set NEEDINFO state (with some help to automate this if needed)
 > But this should not be done automatically.

How do we motivate people to do this on a large scale?  Since Friday 400 
bugs have been closed soley because of this work.  That is 400 *other* 
bugs that triagers can focus on.

Based on the response rate to the proposals that started this process, 
my guess is that sending an email to fedora-devel-list requesting that 
folks perform these actions on their own would have resulted in about as 
many closed bugs as the comments we received from the five or six times 
we requested input and feedback :)


More information about the fedora-devel-list mailing list