13 Sep
2010
13 Sep
'10
8:55 a.m.
On 13/09/10 00:46, Michael Hope wrote:
Your thoughts?
The only real regressions are the vectorizer tests, and I'm not worried about those because a) They're only missed optimizations, not correctness issues, and b) I know there are also a number of progressions in the vectorizer tests, so what we lose in one place, we win somewhere else. We should find out if the tests are bogus, or there is a real problem - in an ideal world all the tests would pass.
I tested glibc and there were no regressions, so I think we can make the releases safely.
Andrew