Auto-expiring bugs are getting absurd

[Adam Williamson]

On Fri, 2014-02-07 at 04:26 -0500, Matthew Miller wrote:
> On Thu, Feb 06, 2014 at 11:54:30PM -0800, Adam Williamson wrote:
> > > What is the underlying problem here anyway?
> > I've never been hugely convinced there is one, but the problem people
> > *claim* there is is that closing bugs on EOL releases gives a bad
> > impression to people who report the bugs.
> 
> We're terrible at numbers, but subjectively, people complain to me about
> this at conferences a lot.
> 
> > Why do you think they're any more likely to get a response if the bug
> > stays open?
> 
> Here's one case: a relatively stable package where there are small RFE bugs,
> or spelling fixes, or packaging improvements which are clearly right but
> have low practical impact.

There is a kind of magic trick for this: if you set a bug to be against
Rawhide and give it the FutureFeature keyword (which is our 'official
way' of identifying RFEs), it won't ever be re-based to a stable release
at Branch time, and hence won't ever get EOLed. That's a bit secret
sauce-y, though.

>  These are good things to do, but maybe the
> maintainer doesn't have lot of time for that particular package. A new
> upstream version comes out every couple of years. When that update happens,
> the maintainer might do an update, and look through all the open bugs to
> make sure they're covered. If they got auto-closed, it never happens.
> 
> I know ideally maintainers should be making those changes in rawhide all the
> time as the bugs come in, but.... time.
> 
> > > I've posted about it in 2008 already, and I still think the auto-closing
> > > leads to hiding crap under the carpet.

> > We already don't have enough time to look after all the open bugs we
> > have. Why are things going to be better if we have more?
> 
> Yeah, closed or open, it would be nice to have better triage, but that's a
> huge job with very little reward and extremely high burnout.

Yep. And there always seems to be something more immediately useful to
do.

But even if we triage the bug load, that doesn't mean we have any
*fewer* open bugs, it just means we maybe do a better job of fixing the
most important ones first. The same absolute number of bugs is likely to
go unaddressed as is unaddressed at present.

>  As I said at
> the end of my other message with the handwaving, if someone has a clever way
> to automatically identify the most important candidates from the thousands,
> that would be very useful.

I would also like that. And my golden toilet. :)
-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net
http://www.happyassassin.net