Re: Bugzilla: Version = unspecified vs. trunk?

[Roland McGrath <roland at hack dot frob dot com>]

> A better policy would be to use the "last reconfirmed" field in triage and 
> say it means the bug was confirmed in trunk as of that date; part of 
> triage would be verifying a bug in trunk.  Then, bugs with older "last 
> reconfirmed" dates could be periodically reviewed - but that doesn't 
> depend on what version they were originally reported against.

That makes sense.  I'd forgotten about that field.

> Presumably your idea is that trunk bugs reflect some transient defect in 
> the source tree, but I don't think that's the normal case.

If a bug existed in the most recent release, then its version field should
cite that release rather than trunk.  It's certainly useful for the bug to
say whether it still exists in trunk too as of some date, though I'm not
sure how best to indicate that formally (i.e. queryably).  For bugs freshly
introduced since the last release, I think it is the normal case that it's
a transient defect.  Saying that's not so is to say that regression is normal.


Thanks,
Roland