For consideration: bugbot may not be _totally_ useless until all bugs are read-only.
For example, for a while I have been keeping one eye on #bzbot, where bugbot shows all bug status changes.
Admittedly there is not much activity after the recent migration, and I make no claim that this is actually useful.

For reference, bugbot is currently on 39 channels:

banshee, bugbot, bugs, bzbot, conduit, contacts, docs, documents, evolution, geary, gegl, gitg, gnome-br, gnome-builder, gnome-calculator, gnome-calendar, gnome-cz, gnome-flashback, gnome-it, gnome-mud, gnome-nl, gnome-todo, gnomefr, gtk+, gtksourceview, i18n, monodev, monodevelop, monogame, nautilus, newcomers, pdfmod, photos, pygtk, release-team, seahorse, sysadmin, vala, webhackers

I agree with you Edward, I feel we should keep bugbot around until BZ is around for historical purposes, the best solution here would be adding ignores against GitLab bots nicknames. I've just added one for gitlab-bot, please let me know if there's more of them I should include on the list.

The other problem is that most of the remaining use of Bugzilla links will be for other Bugzilla instances, e.g. bugs.webkit.org. It's always annoying when I post a Bugzilla link in #gnome-hackers and bugbot posts BS information by using the bug ID from a separate Bugzilla instance to look up some unrelated bug on GNOME Bugzilla. So I'd prefer to get rid of it anyway. Or make it ignore everything except bugzilla.gnome.org.

There's a default Bugzilla instance flag we can pass on a per channel basis, we used to set that when a new channel requested bugbot in the past. Did we miss any probably? What channel are you seeing this behavior on?

#gnome-hackers

bgo seems to be default on that channel already:

<av> channel #gnome-hackers supybot.plugins.Bugzilla.defaultBugzilla
<bugbot> bgo

I don't understand what you're saying.

The problem is it takes links from unrelated Bugzilla instances, replaces the hostname with bugzilla.gnome.org, and prints a bogus result. For example, in #gnome-hackers:

mcatanzaro	Company: Thanks for your help in https://bugs.webkit.org/show_bug.cgi?id=126907!
bugbot	Bug 126907: Gnumeric, major, High, ---, jody, RESOLVED FIXED, manual recalc breaks redraw, and is not exported

I'd like it to either stop doing that, or go away, at least from #gnome-hackers and #gtk+. It's negative value.

Michael, seems we misunderstood each other on this last item :)

I've CCed Teemu who should be able to land a bugs.webkit.org definition in bugbot's configuration files, that way whenever the bugs.webkit.org URL is mentioned bugbot will query the correct BZ instance.

That would, of course, be an improvement.

Still, no configuration should be required to avoid it posting obviously-wrong information. Why is it following links to unknown hosts anyway? That should be fixed.

Michael,

if a specific BZ instance is not defined, bugbot fallbacks to the GNOME's one as per https://bugzilla.gnome.org/show_bug.cgi?id=796458#c6. By landing the webkit URL on the configuration the problem will get fixed automatically.

(In reply to Andrea Veri from comment #10)
> Michael,
> 
> if a specific BZ instance is not defined, bugbot fallbacks to the GNOME's
> one as per https://bugzilla.gnome.org/show_bug.cgi?id=796458#c6.

Yes, but I complain because that's broken and makes no sense. Why would that ever be desirable? Bugbot already knows the link is not a link to GNOME Bugzilla, it knows it is not a GNOME bug, so why form a bogus link and treat it as such?

bugbot is still a problem in #gtk+, at least.

(In reply to Michael Catanzaro from comment #12)
> bugbot is still a problem in #gtk+, at least.

Actually, it followed a link I posted to WebKit Bugzilla... so I suppose that was the right thing to do. I just didn't notice because I'm so used to ignoring bugbot. :P

Also asked gitlab-bot's maintainer to drop the 'bug' word from the bot's output and rename it to issue to avoid bugbot and gitlab-bot to get in conflict. Thanks!