Jason Merrill wrote: > No. Previously we've branched at <100 regressions, but waited for the > numbers to get better than that before making the release. I'm > suggesting that the release criteria stay about the same as before, just > that we delay the branch until the release is ready rather than have > branch criteria and then release criteria.
Yes, that's what I'm imagining too, under this plan. All we'd be doing differently is delaying Stage 1 until after the release, instead of parallelizing Stage 1 with the final release preparation. -- Mark Mitchell CodeSourcery [EMAIL PROTECTED] (650) 331-3385 x713