Re: c/9390: Segmentation Fault when compiling for athlon with LANG=es_ES@euro or LC_ALL=es_ES@euro

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

The following reply was made to PR c/9390; it has been noted by GNATS.

From: "Joseph S. Myers" <jsm28@cam.ac.uk>
To: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
Cc: <david@aspl.es>,  <gcc-bugs@gcc.gnu.org>,  <gcc-gnats@gcc.gnu.org>
Subject: Re: c/9390: Segmentation Fault when compiling for athlon with
 LANG=es_ES@euro or LC_ALL=es_ES@euro
Date: Mon, 3 Feb 2003 00:01:26 +0000 (GMT)

 On Sun, 2 Feb 2003, Wolfgang Bangerth wrote:
 
 > Well, I am happy to reopen this if you insist. However, there is a more 
 > structural problem, which will not be solved by just finding out the one 
 > message that caused the ICE here, so I don't think there is much value in 
 > just fixing this one. The problem is thet we have tons of messages like
 >   "Don't do %X with %Y using %Z"
 > and translations like
 >   "Versuch %X mit %Z zu tun"
 > Here, the number of %-sequences don't match, which leads to the ICE. I had 
 > identified this sometime back last November (+- 1 month; see the mailing 
 > list archives), and a quick scan showed that we literally hundreds of 
 > these cases.
 
 Were those hundreds reported to the language translation lists?  Every
 such case needs fixing at some point, and only the translators can fix
 them.  There's no point in delaying the fixes for known bugs - which will
 be exactly the same fixes with or without the infrastructure to detect
 those bugs - because more bugs might be detected later.  If the bugs are
 reported to the maintainers of the translations now (presuming that the
 bugs are still present in the .po files on the TP site, not just the old
 versions in GCC CVS) then we can have a more reliable 3.3 release.
 
 > We need to have a process that prevents this up front. Just fixing 
 > individual messages is not sufficient.
 
 Fixing individual messages is however necessary.  Every such bug needs
 reporting to the maintainers of the translations at some point, which may
 as well be when it is detected.
 
 > With respect to this particular case: we already have at least one 
 > prototypical report about this open. I closed this one as kind-of-a-
 > duplicate.
 
 A report about this exact message, for this language?  If so - and only if
 so - it is appropriate to close this as a duplicate.  If for a different
 message or language, this is a separate GCC bug, which should be open
 until it has been fixed by the maintainers of the translations and the
 fixed .po file committed to GCC.
 
 -- 
 Joseph S. Myers
 jsm28@cam.ac.uk