On Mon, 2015-03-30 at 16:48 -0400, Josh Boyer wrote:
Dictating use of BZ is usually a futile effort, in Fedora. We have
> a
> policy on it which is in practice rarely observed by anyone. What
> should or should not be the cause is pretty much moot: what *is*
> the
> case is that it makes sense to some of our BZ users to not treat
> CLOSED in the way you advocate. BZ is, fundamentally, a tool, and
> tools usually get used in the way that makes sense to the user.
Ah, so true. Except "user" is just as ambiguous as bug state. The
component owner is just as much a user as the reporter.
That is the definition I was working with, yeah. I tend to see the
people who own the bugs as the main users of Bugzilla, not the people
who report them.
It might make
sense to one user to not discuss bugs in CLOSED where to another user
it does.
Indeed, this is sort of the point I was trying to make; that it's not
necessarily safe to make universal assumptions like 'if the bug's
closed, no-one will want notifications about requested actions'.
Having the metadata around the bug editable by anyone is
really kind of a bad design. So much confusion.
It's not actually editable by anyone, in fact there's rather a complex
permissions system somewhat hidden behind the scenes. If you're a
packager you don't much notice it because you mostly have permission
to do most things (though not quite *everything*, see e.g. review
flags) on any bug (you have 'editbugs' permissions). People who aren't
packagers (or QA team members, or a couple of other ways you can get
'editbugs') don't have that, and can only make changes (besides adding
comments and attachments) to bugs they submitted (or bugs that are
assigned to them, but just about anyone who can own bugs has
'editbugs' anyhow).
TLDR; bugzilla is terrible (but it's the best thing we have).
The
only data that actually matters is that which is contained in the
comments section, and even that is pretty suspect most days.
Sure. Basically what I'm saying is, if you're scripting interactions
with Bugzilla, you need to be aware of the fact that there is no
single universal workflow, different 'users' (however you define that)
use it differently and your tool/script should account for that.
When all's said and done, maybe the thing that makes most sense for
this 'notify of requested actions' *is* 'don't send any notifications
for closed bugs', but even if it is, that's a 'least worst option'
decision, not a 'this is obviously correct' decision.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net
http://www.happyassassin.net