Regression information in Bugzilla

["Giovanni Bajo" <giovannibajo at libero dot it>]

Hello RMs,

Right now, regressions on the mainline have their target milestone set to
3.4.0. Regressions on the branch are set to 3.3.3. But regressions on both the
mainline *and* the branch are naively set to 3.3.3.

This is misleading for several reason:

- Gaby was closing bugs targeted to 3.3.3 as WONTFIX if there was no plan to
fix them on the branch. But those might be regressions on the mainline as well,
so they shouldn't be closed altogether. This can easily be checked using the
stamp in the summary, which immediatly shows which versions are affected by the
bug.

- Mark is counting the 3.4 regressions after querying for bugs with milestone
set to 3.4.0. This query is incomplete, because it doesn't report regressions
which are present on both the mainline and the branch, and that are of course
part of the release criteria for 3.4.0 as well. Instead, he should be querying
for bugs containing the word "3.4" in the summary.

I understand that this system is far from being optimal. We are discussing this
issue in these days on the bugzilla-master list, and trying to come up to a
decent solution to handle this. A first stopgap would be to be able to set
multiple target milestones for each gap, as suggested by Daniel Berlin. A more
efficient solution (solving also other problems with version tracking) is at
study.

Thanks,
Giovanni Bajo