Rainer wrote: > Do you really think so, with testsuite results like this: Yes. It certainly makes it easier to identify bootstrap regressions. I can imagine that if a 3.1 branch were taken today you may want to disable it on the branch until the testsuite results are better. AG