launchpad-dev team mailing list archive

[Martin Pool <mbp@xxxxxxxxxxxxx>]

On 15 April 2011 16:08, Robert Collins <robert.collins@xxxxxxxxxxxxx> wrote:
> On Fri, Apr 15, 2011 at 5:11 PM, Martin Pool <mbp@xxxxxxxxxxxxx> wrote:
>> There are other simplifications that I think would probably improve
>> the user model more though, like unifying fixreleased and
>> fixcommitted, or confirmed and triaged, or various other old bugs.
>> One reason I think they would have a higher payoff is that everybody
>> dealing with bugs hits them, whereas using multiple tasks is a more
>> advanced use.
>
> I have a few thoughts here. Firstly I hesitate to prioritise things
> purely on user impact - sometimes the balance between effort and
> rewards is linked, sometimes its not :).

I agree.  I think multi-series bugs are a bit more of an edge case
than your numbers suggest, but it is still worth improving, especially
if it makes Launchpad faster or internally cleaner.

I am pretty sure most of the bugs with multiple tasks do not have
multiple tasks on different series in the same project, but rather
they affect multiple projects or they affect Ubuntu and also an
upstream.

I suspect too (with less confidence) that many projects have multiple
series for the sake of making releases off them or hanging branches of
them, even if they never target bugs to them: all the projects in your
"snapshot" class and indeed I think also Launchpad itself.

Of the ten hottest bugs in Ubuntu, only one has more than one valid
series task in the same pillar, and that is <http://pad.lv/450569>,
which was backported to karmic-updates.

The other interesting one there is <http://pad.lv/410407> which has
three screenfuls of people nominating it to be fixed in everything
under the sun and is a great example of something to do with bugtask
understandability.  Quite possibly that would be helped by this change
if it eventually leads to a clearer expression of "fix in the current
devel release" vs "backport it".

All the above is just jam-style "interesting to consider" rather than
reasons not to do it.

Martin