Hi, Having gcc-4.2 as default on i386 and amd64, while most other architectures have gcc-4.3 means most maintainers never test their packages with gcc-4.3. This is making life miserable for those ports that actually use gcc-4.3:
1) people upload packages with known gcc-4.3 bugs and not bothering to fix them. 2) people applying gcc 4.3 bugfixes without actually testing them with gcc-4.3. Often these turn out to be incomplete fixes. On Sun, Mar 23, 2008 at 08:30:54PM +0100, Pierre Habouzit wrote: > Sigh, could we avoid the same discussion over and over, people are not > supposed (we never asked it in the past, and I see no valid reason to do > so) to update to the last stable point release before upgrading to > stable+1. Dist-upgrade will not fail due to people still having this bug in their kernels, so there is no need for users to upgrade kernel _before_ upgrading to lenny. After upgrade reboot to the new lenny kernel is normal procedure. > THe _BEST_ example of that are buildd's that for now run etch (even > some sarge not so long time ago) and have a sid chroot to build. Not > keeping the CLD patch means that we break our own buildd infrastructure. > Yay. This is making a mountain of a molehill. We have one amd64 and one i386 buildd. Certainly installing a (security!) kernel update on two buildd's is a better solution than having old version of gcc on the two most used archs we have.. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
