MABs and priority (was: minutes of ESC call ...)

[Bjoern Michaelsen]

Hi,

On Thu, Jan 16, 2014 at 06:21:48PM +0000, Michael Meeks wrote:
>     + All Most Annoying Bugs -> priority Highest (Bjoern)

This is done now too, so right now, all NEW bugs(*) with priority highest should
also block a MAB and all NEW bugs blocking a MAB should be priority highest.

I updated the MAB instructions to include setting "priority: highest" when
adding a bug to MAB.

for QA, this means two things, one short term, one long term.

== short term ==

To keep the highest=MAB equivalence clean, some regular (weekly?) checking for
bugs that are priority:highest and not a MAB would be needed. Such bugs could
be seen as "proposed MABs" and either be:
- promoted to a MAB with the usual procedure (rationale etc.)
- or respectfully rejected and bumped to priority:high

Someone volunteering for this task?

If so, it would be interesting:
- how many such bugs there are each week
- what is their quality (as in: how many are good MABs? how many are well triaged?)

== long term ==

If the above short term activity show that the number of false positives is
low, QA and development might investigate simply using priority highest directly.
QA would then need make sure the total number of priority:highest bugs is
ideally somewhere between 50-100 (exact number would be discussed on the ESC).
It would do so by watching for example the bugs that were added to
priority:highest in the last week with a query like this(**):

 https://bugs.freedesktop.org/buglist.cgi?priority=highest&list_id=383855&resolution=---&chfieldto=Now&query_format=advanced&chfield=priority&chfieldfrom=-7d&bug_status=NEW&product=LibreOffice

Development would simply query for bugs with priority highest, or, if that has
to much noise by stuff not sanity checked by the QA, only those bugs in
priority:highest that have been in that state for more than 7 days (and thus
likely checked by QA), e.g.:

 https://bugs.freedesktop.org/buglist.cgi?priority=highest&f1=priority&list_id=383863&o1=changedafter&resolution=---&n1=1&query_format=advanced&bug_status=NEW&v1=-7d&product=LibreOffice

So, lets see how this evolves! Personally, I would love to see us move away from
the MAB tracker bug patchwork solution to something simpler and cleaner.

Best,

Bjoern


(*) Unresolved bugs are mostly uninteresting for tracking MABs, but while
    checking for this, I found a few unresolved MAB that I missed to put in
    priority highest because they where not in NEW:
    - 2  NEEDINFO
    - 2  ASSIGNED
    - 11 REOPENED
    I would assume the assigned ones to resolve themselves soonish. The
    needinfo ones seem a bit dubious to me: A bug that isnt completely triaged
    doesnt seem like a good MAB. We will see what happens to the REOPENED ones --
    the high count of them seems curious.

(**) Note that this query is currently providing so many results because of the
     bulk move.