I've now reviewed the open regressions against the GCC 4.1 branch. There are 101 "serious" (P3 or higher) regressions against GCC 4.1, the vast majority of which also apply to 4.2. Therefore, fixing these regressions provides a double benefit: both the release branch and the next release will be improved.
There are 16 P1s, most of which are wrong-code or ICE-on-valid regressions. The C++ front-end is to blame for four of those regressions; the remainder are the middle end and back end. As indicated by their P1 status, I would like to see these issues fixed before the 4.1.1 release. (Of course, it would be nice to get the remaining issues resolved as well.) By our development plan, we'd expect a 4.1.1 release on April 28th, two months after 4.1.0. I don't believe that's achievable, if we are to fix the P1s, but I would hope that May 15th would be achievable; that gives us two weeks to fix regressions, and then two weeks for freeze, release-candidates, etc. Because the release branches tend not to regress, I think it's safe to set May 15th as a relatively firm date; if the P1s are not fixed by then, we will help our users more by releasing 4.1.1 than by waiting indefinitely. -- Mark Mitchell CodeSourcery [EMAIL PROTECTED] (650) 331-3385 x713
