Launchpad bug statuses

[Scott Kitterman]

On Tuesday 02 October 2007 16:15, Curtis Hovey wrote:
> On Tue, 2007-10-02 at 21:14 +0200, Sebastien Bacher wrote:
> > Le mardi 02 octobre 2007 à 15:04 -0300, Christian Robottom Reis a
> >
> > écrit :
> > > I've written a piece at
> > >
> > >     http://news.launchpad.net/general/of-bugs-and-statuses
> > >
> > > that describes the intended semantics and existing behaviour for
> > > Launchpad bugs. I'd love to hear your questions and comments about
> > > cases which we don't handle well so we can better improve the way the
> > > tool works.
> >
> > * "New (a.k.a. Nobody Has Looked At Me Yet)"
> >
> > Bugs are sometime moved from Incomplete to New when the submitter has
> > provided the required details but the submitter doesn't confirm the
> > issue
> >
> > * "Incomplete (Reporter, Give Us More Information!)"
> >
> > The description doesn't mention the case where bugs are sent back to
> > New. What should a triager do when the submitter replied but he still
> > doesn't know if the bug should be confirmed? Shouldn't launchpad also
> > automatically reopen Incomplete bugs when the submitter replies if those
> > are going to be autoclosed?
>
> I'm not sure. We want to avoid this situation. It is difficult to judge
> progress when a bug moves back and fourth between statues. We will send
> out a notification about the impeding expiration to all parties to
> encourage a timely followup. Anyone who sends a message/leaves a comment
> will implicitly reset the period of inactivity.
>
> We have two intervals to work with, the time of expiration and the time
> of the warning. Sending out a warning 14 days before expiration  should
> be enough time for some person to send an email, Do we need more time?
> Is 60 days not enough time? I believe Ubuntu's policy is 30 days in
> Incomplete before it should be moved to Invalid.

This all sounds like stuff that it would have been good to work out before the 
feature was rolled out.  I think this kind of complexity is an example of why 
the entire auto-close idea is a bad one in my opinion.

It's one thing to have a triager judge that for the circumstances in question 
warrant closing a bug due to inactivity, but those circumstances vary 
considerably from project to project.  Backports bugs often move very slowly 
and sit incomplete for long periods due to lack of someone to do necessary 
testing.  

I don't believe that there is a one size fits all solution for this.

Scott K