[Bugsquad] Bugreporting barrier is too low with the new Dr. Konqi

[Darío Andrés]

2009/11/11 Darío Andrés <andresbajotierra at gmail.com>:
> On Tue, Nov 10, 2009 at 6:23 AM, FiNeX <finex at finex.org> wrote:
>> In data martedì 10 novembre 2009 02:53:04, Darío Andrés ha scritto:
>>> Another idea I got in order to reduce the noise to developers:
>>> [...]
>>
>> I like the idea: usually when I have some time for triaging bugs I pick up the
>> latest N bug reported and I read them all. Usually bugs reported from drkonqui
>> are dups or "NEEDSINFO". Anyway they are mostly filled under the right
>> product. Why not set a flag "automatically submitted" so developers can filter
>> them out and triagers can filter them in?
>>
>> Something like "waiting for human check" ?
>>
>
> May be I start liking this idea. Creating the bug report under the
> corresponding product but with a special status
> NEEDSINFO/WAITINGFORAPPROVAL or a special keyword would allow:
> - Developers will get notified about them, but they should be able to
> ignore them if they like to. (just exclude such reports from your
> query searches, and ignore your mails if they have "Reported using
> DrKonqi")
> - Reports will be under the right product from the start (I will look
> forward implementing the online mapping feature)
> - Open bug reports count will not increase until the triagers confirm
> that they are valid issues (pre-filter system)
> - Having a keyword or a special state makes searching for the
> untriaged bugs a lot easier too.
>
> + Improving duplicate handling in DrKonqi itself
>
> All the discussions about changing bugtracking system are a bit beyond
> me so I won't suggest anything yet...
> BTW, thanks to all for the feedback about the current system, it is
> greatly appreciated and I will look forward improving the situation.
>
> @FiNeX: this discussion comes from kde-core-devel, so be sure to send
> your replies to that list too so everyone can see them.
>

Mh, it seems that bugzilla won't accept entering bugs with a closed
state. We could implement a patch or use a keyword instead (something
like "automatic_unverified_report"). What do you think ? Devs can
ignore reports with that keyword in their queries, too bad we can't
fix the statistics :-\ (or may be we can fix them, patching them)

I have been also wondering if we could implement a remote agnostic
interface, so DrKonqi will be able to communicate with bugs.kde.org
system regardless of the bugtracking system (useful if we ever migrate
to another system, and to write a better implementation to retrieve
the bugs' data  (parsing HTTP after POST/GET HTTP requests is ~nasty~
(current way)).
With a custom (and own) interface we could provide a stable remote API
for DrKonqi even for stable (not trunk) KDE versions (bugzilla xmlrpc
is unstable API, we can't rely on it, bugzilla upgrades will break
it).

> Cheers
> Darío A.
>