[RFC] Bugtracker workflow

[Christian Robottom Reis]

On Thu, Aug 17, 2006 at 05:27:02PM +0200, Wouter van Heyst wrote:
> > You could explicitly reject the bug (Rejected is intended to split  
> > off into something like Wontfix and Invalid) in the 0.9 series to say  
> > you won't fix it there, and mark it fixed in the 0.10 series.
> > 
> > Will they be different series though?
> 
> Yes, but there might be point releases, like 0.8.1 and 0.8.2 currently.
> How hard would it be to split up a series even further when the need
> arises?

Aren't these separate concerns, and won't conflating the two into the
release management system just cause more confusion? We need to keep in
mind thatseries and releases are distinct things. The problem /here/ is
that there are a few different things we mean when think of bugs, series
and releases:

    1. The release the bug reporter detected the bug in.
    2. The other releases the bug [is known to] affect.
    3. The release in which the bug was fixed.
    3. The series which the bug [is known to] affect.
    4. The series in which we intend to fix the bug.

The current Launchpad model is that series information is used for
management of work to be done on the bug: deciding for which future
release this bug will be fixed, and deciding for which releases the fix
will be backported.

There was a proposal of "infestations", a system intended to capture the
information related to versions. Infestations would be data added by
end-users that essentially represented the data in point 2 above. There
is no implementation of infestations, currently, however. There are
concerns whether this is an actually effective system, since it requires
a lot of work on behalf of the end-users to indicate all the versions
affected -- which wouldn't happen for smaller projects or less-used
packages. It would at best be an informative knob -- this bug is known
to affect at least versions: 0.8.1, 0.8.2.

For now, I would suggest that this information be captured in the bug
description and comments. I would be happy for us to do some extra work
to help this: perhaps adding text suggesting that versions affected be
added in the initial report or in follow-up comments, or something
slightly more involved.
-- 
Christian Robottom Reis | http://async.com.br/~kiko/ | [+55 16] 3376 0125