[06:25] StevenK: Why shouldn't bug #941926 crash? [06:25] <_mup_> Bug #941926: InvalidTransition: Transition from Completed to Waiting is invalid. [06:27] wgrant: I just thought it should actually give a nice error message rather than an OOPS [06:27] StevenK: Why shouldn't it OOPS? [06:28] Because we have a policy [06:28] Sure [06:28] WHERE "auditor_auditor"."object" IN (foo) GROUP BY (object), (operation), "auditor_auditor"."object", "auditor_auditor"."operation", "auditor_auditor"."date" [06:28] Dear Django, I HATE you [06:28] But I could fulfill the word of that policy with an unconditional except: [06:28] That displays an error page [06:29] Turning data corruption into an error message rather than OOPS is the opposite of progress [06:29] wgrant: It can happen when data isn't corrupted, though [06:29] StevenK: How? [06:30] Complete job with unaccepted queue entry sounds like data corruption to me [06:30] acceptFromQueue() doesn't check, it just forces it to be accepted [06:30] wgrant: Archive Admin A and B both load +queue, A accepts the sync, while B goes to make a coffee. B comes back, the job has completed and submits the form -- OOPS. [06:31] Right, so you're papering over a real and dangerous bug [06:31] Surely the problem is the double accept [06:32] If acceptFromQueue can today be called on a DONE or ACCEPTED entry, we have a data corruption bug [06:32] That is the bug here, not the InvalidTransition OOPS [06:34] Oh, screw you Django [06:38] wgrant: Hm, we already check [06:50] wgrant: Hm, it will croak on ACCEPTED, but not DONE === tasdomas` is now known as tasdomas_afk === tasdomas_afk is now known as tasdomas === mbarnett` is now known as mbarnett === tasdomas is now known as tasdomas_afk [16:59] Ursinha: ping [17:29] cjohnston, pong [17:29] hey Ursinha.. Was there any further progress ever on exporting all subscribers for blueprints? [17:30] cjohnston, not really, I guess we need to start a discussion somewhere to figure out the impact and what to do (I guess that's what was said, iirc) [17:33] hrm.