Blocker Bug Voting and Conversation

[Tim Flink]

On Tue, 4 Dec 2012 08:22:12 -0500 (EST)
Kamil Paral <kparal at redhat.com> wrote:

> > Affected Voters:
> >  - Do you vote on blocker status in bug comments?
> 
> If necessary, but I don't like it much. In my experience the
> discussion in the meeting is often very helpful to understand the
> nature of the bug, and it can shift my opinion substantially.

I'm not trying to do away with the meetings, per se but I would like to
enable a bit more asynchronous action. One of the problems with doing
blocker review meetings is that they happen at the time they're
scheduled. Finding a time where enough people are able to attend is
difficult and someone is always going to be unable to attend.

I also think that pinging devs during the meeting is also sub-optimal.
Assuming that someone is available during the meetings for whenever a
bug assigned to them comes up, it's disruptive to be interrupted just
because we happen to be discussing something right now. I'd like to find
a solution where we weren't quite so dependent on simultaneous real-time
interaction.

Is it the meeting itself which you find helpful or the discussion and
information during the review meetings? Could you see asynchronous
conversation (exact method TBD) being as useful?

> Also I don't like spamming bugzilla with irrelevant data. It makes
> the whole bug report less readable. 

In my mind, the trick is finding a good balance. I think that keeping
blocker status completely separate from the relevant bug is worse than
polluting the bug with too much blocker conversation.

> >  - Would you vote on blocker status more often if you could easily
> >  vote
> >    outside of meetings?
> 
> If we are not in a rush, I'd keep everything in the meeting. If we
> are in a rush (like now), I'd move some "obvious" (or controversial,
> those might be good candidates too) items outside of the meeting, but
> not into the bugzilla.

I'm trying to keep on top of that for now as more of a manual process
but I think that one person doing the sorting isn't great. It's a bit
of a pain for the sorter, there are going to be mistakes and it doesn't
get around the issue of "where do we have the conversation?".

> An email thread on the test list is much
> better. It can contain long discussion without obfuscating bugzilla.
> Bugzilla can contain just a single comment with a hyperlink to the
> discussion, so that anyone interested can join. Once consensus is
> reached on the list, one of QA guys can update bugzilla status.

Adam said something similar but I really don't want to use email alone
for this. While it might technically work, I also find email threads
like that to be rather difficult/painful to parse. It would also make
any process automation (bug comments, status changes, vote counting
etc.) much more difficult, adding to the amount of required human
intervention.

> There is one important drawback, and that is the necessity to be
> subscribed to the list.

Yeah, I think that's going to be a potential issue no matter what we
end up with. Not necessarily list subscription but making sure that
everyone who needs to be in the loop is in the loop while keeping the
signal-to-noise ratio up and the annoyance factor down.

Tim

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: not available
URL: <http://lists.fedoraproject.org/pipermail/devel/attachments/20121204/2055d536/attachment.sig>