My New Year's resolution is to complain (constructively) whenever I
get a spurious build failure email from a buildbot. For new or
infrequent contributors especially, they can be extremely confusing
and unnecessarily alarming.
This one below is the first build ever attempted by the builder, so
how on earth can it have come up with a meaningful blame list? And in
any case, surely we should only get emailed when something breaks,
i.e. on a transition from green to red? In this case there was no
previous build whatsoever, let alone a successful one.
Please bear in mind that quite a lot of buildbot-related issues you’ll see are, well, caused by bugs/features of buildbot source code rather than the configs used for the LLVM master.
I’m not familiar with the LLVM master configs per se, but from my experience setting up buildbots for other projects, there are dozens of features in buildbot you’ll find weird. And in most cases, those are not actionable for LLVM developers, so not sure I support your resolution
Flaky tests and/or undecipherable LIT test outputs are good reasons to complain though.
In this particular case, it looks like the bot has just been added [and is unlikely to be green for the first few cycles/days], so it seems reasonable to just disable email notifications for it.
I don’t know who has added the bot, so adding Galina and Kostya to CC, they should know.
This should hopefully be fixed by http://reviews.llvm.org/D6934 (by Will)