launchpad-dev team mailing list archive

[Martin Pool <mbp@xxxxxxxxxxxxx>]

2010/1/12 Deryck Hodge <deryck.hodge@xxxxxxxxxxxxx>:

>> How about a survey, on the lp blog, asking people if the hot bugs list
>> works for them?
>
> Perhaps we should do something like this.  The current "hot bugs" is
> just recently touched bugs.  But if people find this useful, we should
> rename it appropriately and keep it.  I really hadn't considered that
> people like the page as it is now, since I consider it a broken
> implementation.
>
> If we kept as is, we could then have a filter link to "hot bugs" or
> have two short lists -- have both "bugs with recent activity" and "hot
> bugs" on the home page.

We used to have two bug lists on that page, and fwiw I think just one
that shows all the columns and is easier to scan is better.

> I think the usefulness of a bug heat indicator is in searches and
> glancing at existing bugs lists, so I'm cool with being more flexible
> about the bugs home and trying something there.  And I do recognize
> that this is just my (one person's) opinion, so the idea is to get
> something working this month and spend the following month polishing
> the feature based on user feedback.

Right, the nice thing about this is that you can tweak it over time.

I realized last night that your algorithm in
<https://dev.launchpad.net/Bugs/BugHeat> is really pretty close to
just being sort-by-affected-users with priority given to private and
security bugs.  It may be reasonable to assume that projects already
treat security bugs as important, at least to the extent of triaging
them first.  So perhaps no more is needed here than finishing the
affects-count bugs?

It would be useful to understand how this is supposed to mesh with
importance.  It could be any of:

 * We have a gazillion new/untriaged bugs; I want to triage the
likely-most-important ones first.  Then once triaged, the importance
is considered authoritative.  This would likely be skewed by the
accuracy of dupe-finding across different types of bug.

 * Within all our high bugs, or bugs tagged easy or doc, I'd like to
do the hottest ones first.

 * As a project manager or spectator I'd like to just get a feel for
the most interesting bugs.  recently-changed does this fairly well.

 * Sometimes we triage incorrectly or at variance with our user's
opinions; I'd like to see how they align.  affects-count does this in
an interesting way; the top bzr one is both fairly large and medium
priority but there's a reality check in seeing that (modulo
dupefinding etc) it is the most-wanted bug.

 * I want to find important but neglected bugs.  (maybe the same as
the previous point.)

 * ...?

-- 
Martin <http://launchpad.net/~mbp/>