On Mon, Jul 20, 2026 at 01:02:10PM +0200, Santiago Vila wrote: > [ replying to the list, please do not CC me on this thread ] > > On Sun, Jul 19, 2026 at 07:26:31PM +0300, Adrian Bunk wrote: > > > The basic rule is that FTBFS on our buildds are RC, and FTBFS that only > > happen in other environments are not RC. > > For the record: > > According to Release Policy, packages which FTBFS with nocheck build > profile are RC.
I already mentioned that indirectly in my email. It is less clear whether it is also RC when the nocheck build is not identical to the buildd build, which is quite relevant here. > Given that there are more than one way in which a build failure is RC > (and more to come in the future) and the buildds only build packages > in one such way, it naturally follows that the "basic rule" is not a > rule at all. It's false. It is an exception when a FTBFS that does not happen on our buildds is considered RC by the release team. And a reason why I suggested reproducible rebuilds should consider switching to nocheck in the future is that it is quite common that the testsuite of a package fails on some architecture based on what hardware the build machine is. > Thanks. cu Adrian
