597786 - Ubuntu Launchpad bugspamming b.m.o. Linking to (Ancient|Dupped|Wrong Product) bugs.

Status: RESOLVED WORKSFORME

(bugzilla.mozilla.org :: General, defect)

Description

[Philip Chee]

On 19/09/2010 03:48, Reed Loden wrote:
> On Sat, 18 Sep 2010 16:41:06 +0800
> Philip Chee <[e-mail]> wrote:
> 
>> [not sure about the correct newsgroup. Please redirect as necessary]
> 
> Yeah, not sure what newsgroup would be appropriate for this. Maybe
> mozilla.dev.tools or mozilla.dev.mozilla-org? I dunno.
> 
>> The ubuntu launchpad seems to have gone nuts and is randomly updating
>> ancient b.m.o. bugs that have been closed ages ago. Can someone politely
>> ask the launchpad people to stop bugspamming us? Thanks.
> 
> Not sure why Launchpad just started linking bugs again today, but
> it's nothing new, nor is it random. They contacted us a few years ago
> about doing this, but due to some problems early on, it's been very
> slowly rolled out. Basically, they are linking existing Launchpad bug
> reports about Mozilla products with the associated Bugzilla bugs in
> bmo. This also allows for two-way communication with users on
> Launchpad, as comments are linked between the two bug trackers, which
> allows for Ubuntu users to help find and report bugs to us.
> 
> The only real issue I've seen is that in some cases, Launchpad is
> linking to a dupe bug instead of the duped-to bug. I will follow-up
> with the Launchpad guys to see if this can be resolved so the correct
> bug is linked.
> 
> If you know of any other issues where the Launchpad bot isn't working
> correctly, let me know, or you can file a bug under mozilla.org ::
> Bugzilla: Other b.m.o Issues, and I'll make sure the right people at
> Launchpad see it.
> 
> ~reed
> 

On 19/09/2010 04:01, Michael Lefevre wrote:
> 
> I think the issue is, as Philip said, the ancient bugs.
> 
> For example, connecting 3-year-old Firefox 2/3.0 bug reports in 
> launchpad (bug 44987) to 6-year-old bug reports against the Mozilla 
> Suite (bmo bug 144612) seems unlikely to help current users or developers.
> 
> I guess there may actually be a few issues where very old that is 
> useful, but in general it might be an idea not to touch old bugs. I'm 
> not sure what might be a good cutoff - maybe bugs not changed in a 
> couple of years?
> 
> Connecting launchpad Firefox bugs to bugs in the Seamonkey product seems 
> wrong too (although I guess that bug was in a shared component before 
> the Firefox and Seamonkey UI bugs were completely split up, whenever 
> that happened...)

Comment 1

[:Gavin Sharp [email: [e-mail]]]

The two-way linking doesn't seem beneficial to b.m.o at all (even without taking into account the fact that the associations are often bogus). With bugspam taken into account I think it's actually detrimental.

I'm quite tempted to remove the bot's editbugs/canconfirm bits... Why shouldn't we?

Comment 2

[:Gavin Sharp [email: [e-mail]]]

Comment 1 wasn't intended as a loaded question - I really would like to know what the expected benefits of bi-directional linking are, given the demonstrated costs...

Comment 3

[Karl Tomlinson (:karlt)]

I think that's a fair question.
If I want to receive updates on the Launchpad bug, then I subscribe to that bug.

Comment 4

[Chris Coulson]

Note that the issue with linking to duplicate bugs is being tracked in https://launchpad.net/bugs/642418

Comment 5

[Philip Chee]

>           See Also|                       |https://launchpad.net/bugs/642418
Oh now we have a mkanat-bot as well?

Comment 6

[:Gavin Sharp [email: [e-mail]]]

It stopped being noisy, so I guess there's no point keeping this open. If it gets annoying again we can just remove its privileges.