On Fri, Oct 26, 2018 at 11:02 PM Christian Dersch <lupinix.fedora@gmail.com> wrote:

Having that criterion " if there's an RC specific failure to output images that didn't cause nightlies to fail, then we have to find out what's going on." would be really nice.

Greetings,
Christian

Christian, I absolutely agree, that having the Spins built is desirable, and especially when they were "OK all the time before" it is extra weird when a build is just plain canceled (just to make it clear, I did not have anything to do with that, neither know who did).The thing with having a criterion like that, with the current way the compose process works would most probably mean whole new compose being created, thus basically invalidating the testing done on the blocking images in that compose.
There is also the question of "who's going to be responsible for that investigation?" (as it seems to be in the domain of already pretty busy releng, and not qa. Correct me if I'm wrong) and "how is that investigation gonna be performed?" (release criteria IMO should be clearly actionable).
The better way (I'd even go as far as saying the "correct", but I'm not that well-versed in releng stuff), would be pushing not for the "easy way out" (criterion) but for the tooling to change in such a way, that spins can be built separately off of the "main" isos.
Just my $.02
J.