Jesse Keating jkeating@redhat.com writes:
On Fri, 2009-04-10 at 12:27 -0400, Tom Lane wrote:
I'm still not seeing proof, just speculation.
Well, let's see:
1. I've traced the mysql build failure far enough to know that it is dumping core inside a calloc call.
2. The crash happens with glibc-2.9.90-12, not with -14, again with -15.
3. It happens with mysql 5.1.32 as well as 5.1.33, meaning I've got a due-to-be-released-if-I-can't-update-it package that FTBFS.
Now, I do not have proof at this point that it's really a glibc bug and not a glibc behavioral change that is within spec but tickles some previously unseen mysql bug. It's highly unlikely that I could prove anything either way before Tuesday's freeze, especially if I harbor any illusions about being able to take the holiday off. Whichever way it falls out, we've got an unnecessary change in a core system library destabilizing things just a few days before "final freeze", and I am not happy about that.
regards, tom lane