Re: bug autoclosing NMU annoyance

[Paolo Molaro <lupus@lettere.unipd.it>]

On Sat, Feb 05, 2000 at 10:05:55PM -0500, Joey Hess wrote:
> The automatic big closing whena package is installed into the archive has a
> nice side feature. If it thinks you made a NMU to fix the bug, it doesn't
> close it, but just sets its severity to fixed. The heuristic it uses to
> determine this seems to be if the uploader name and email do not match the
> maintainer name and email, it's a NMU.
[...]
> I wonder if others who regularly upload such packages are annoyed by
> dinstall's behavior as much as I? If so, perhaps something should be done.

I think the bug closing stuff should be made more explicit and flexible.
The proposal is to extend the perl re to understand a priority instead
of only allowing closing bugs: sometimes a new vesrion may reduce the
priority of a bug while the bug itself still persists.

/closes:\s*(bug)?\#\d+(,\s*(bug)?\#\d+)*/gi

would become:

$priority = '(fixed|normal|critical|important|grave|whishlist)?';
/closes:\s*(bug)?\#$priority\d+(,\s*(bug)?\#$priority\d+)*/gi

or, maybe better, we can add another re:

$priority = '(fixed|normal|critical|important|grave|whishlist)';
/bugpriority=$priority:\s*(bug)?\#\d+(,\s*(bug)?\#\d+)*/gi

and handle this stuff more cleanly in both dinstall and
dpkg-*, without guesses.

The only drawback is that the 'closes' keyword in .changes is no more 
proper. We can leave that as is and add another field:

Bugs: fixed 33 445 667, normal 663 479, critical 44564 5342

to automatically set fixed priority for bugs 33, 445 and 667,
normal for 663 and 479 and so on.

Comments welcome.

lupus