So, good news - Greenwave is emitting decision fedmsgs again!

Kinda bad news - it's emitting some pretty odd ones. Like the one I just showed @ralph.

It's a compose decision, but it seems not to have an actual subject. There's a subject type, but no...subject. This is reflected in the way the fedmsg2meta-generated text description of the result is "greenwave says NO-GO on "something" for "rawhide_compose_sync_to_mirrors"" - note the "something" there.

Given that the only satisfied requirement is "compose.cloud.all" - which is the ResultsDB 'testcase' that autocloudreporter uses when forwarding Autocloud results to ResultsDB - I suspect these decisions are being generated in response to those autocloud tests.

I don't remember all the ins and outs of how greenwave decides to emit decision fedmsgs and when, I'll have to look into it again. But there's obviously something a bit off with these decisions.

Emitting weird 'no go' decisions isn't really a huge problem, of course, because they'll probably just kinda get...ignored. But it's probably worth looking into. We'll have to wait for a finished compose and the openQA tests to run to see whether we get a valid decision at the end of all the openQA tests...