On Sun, Nov 20, 2005 at 10:25:57AM +1100, grant beattie wrote:
> [moved to tech-pkg]
> 
> On Sat, Nov 19, 2005 at 08:36:00AM +0100, Thomas Klausner wrote:
> 
> > > reply to PRs automatically includes pkgsrc-bugs.
> > > 
> > > personally, I like replies going to pkgsrc-bugs, too, as I like to see
> > > PR followups without having to query each PR individually. I suspect
> > > I'm not alone.
> > 
> > Actually, if you reply to a PR and cc gnats-bugs, it will automatically
> > get forwarded to pkgsrc-bugs (so you'll see it). There is no reason
> > to cc pkgsrc-bugs.
> > 
> > On the other hand, if you reply to pkgsrc-bugs (but not to gnats-bugs),
> > it will end up on the mailing list too, but not in the gnats
> > database, so people looking for the audit trail there won't get
> > all relevant information.
> 
> I always reply to 'all', so at least for my replies, nothing will be
> missed in GNATS.

Actually, that's not true, because of how gnats behaves.
When you edit a PR and change its state or the responsible
person, a mail goes out with the following people in the To:
header:
. the person doing the change
. the (former and new, if it changed) responsible person
. pkgsrc-bugs
. gnats-admin
. the submitter

The Cc and Reply-To headers are empty and not set, respectively.
If you "reply to all" to such a message, gnats-bugs will not get a
copy and the reply will not end up in the gnats database.

A different, but related problem, is that when you reply to a
bug per mail and only mail to gnats-bugs, the submitter won't
get a copy.

So currently the correct procedure for replying to gnats is to
send mail to the submitter, if he should get it, and gnats-bugs.

> thanks for the explanation - I guess that GNATS config should be
> updated to bcc: pkgsrc-bugs instead of cc:, so that it will not be
> included in replies.

That would be an improvement, but not a complete solution.

Cheers,
 Thomas