Re: c++/3099: cygwin: Many g++ EH failures

["Joseph S. Myers" <jsm28@cam.ac.uk>]

On Sun, 11 Nov 2001, Jason Molenda wrote:

> Now that you've found the problem, I can lamely provide the story behind
> it.  The original gnatsweb maintainers didn't think that the GNATS_ADDR
> should be listed on the To line, cf my original note about this:
> 	http://sources.redhat.com/ml/bug-gnats/1999-08/msg00075.html
>
> although everyone else agreed that this should be done.  It's been
> a local patch on gcc.gnu.org/sourceware ever since--it must have
> been accidentally dropped in one of the upgrades.  (it's clear that
> the modification has needed to change over time, and it's always
> a little tricky to merge those in correctly.)

In that particular case, the mail to interested parties link, it seems the
maintainers may eventually have agreed, since that particular case works
and isn't marked as GCC LOCAL.  The problem is just the PR status change
messages.  (PR status changes used to appear twice in the PR logs, once in
the audit trail entry and once when the status change message got to
gcc-gnats and was logged itself, but that was just a minor issue, whereas
replies to those messages not getting filed in GNATS is a serious
problem.)

(I agree with the point made in that thread that it should be the
responsibility of GNATS to generate these mails and audit trail logs, not
gnatsweb - is this going to be fixed in GNATS 4?  This would avoid any
such non-obvious cause for such a change as this one - which I hadn't
expected to be a gnatsweb problem since I'd supposed that GNATS did deal
with these messages itself.)

-- 
Joseph S. Myers
jsm28@cam.ac.uk