Joe Buck wrote: > I agree that the matter should have been raised far earlier, and that > glibc decisions of this kind should be coordinated with gcc, and in this > case the issue should have been discussed far earlier.
Yes, I completely agree. In fact, I think everyone agrees; Roland has suggested as much in his messages as well. As I've indicated before, I'm not pleased with this situation either. It was as much a surprise to me as anyone. There is no question that this change is not in keeping with our branch policy. However, the PowerPC GNU/Linux community seems to want this feature very badly, and has suggested that failure to incorporate these patches in GCC 4.1 would be very bad. My feeling is that it is the PowerPC community which will be harmed if they get the patches wrong, so it's somewhat reasonable to let those folks balance risk and reward. I would not be inclined to hold up the release for these changes, but we're not ready to release yet, so this is not yet an issue. Also, at the time these changes were suggested for 4.1, there were none (minimal?) objections; at this point, the developers have been working on the changes for quite some time. If there were significant objections, they should have been made immediately, and, if necessary, the SC involved at that point. Jakub has already indicated that the libstdc++ changes will not go on the 4.1 branch. I, too, believe those changes are too risky. -- Mark Mitchell CodeSourcery [EMAIL PROTECTED] (650) 331-3385 x713