mozilla :: #bors

17 Jul 2017
15:52* notriddle still can't help but smirk at "not-a-robot[bot]"
15:55notriddlebadboy: Something broke on https://github.com/redis/redis-rb/pull/701
15:55boringbotPR #701 [open]: Get tests passing with frozen-string-literals enabled. - https://github.com/redis/redis-rb/pull/701
15:55notriddleYou might want to check the `heroku logs not-a-robot`
16:12badboybest support ever :D
16:12badboywill do
16:23notriddleBTW, if you update not-a-robot (which will involve an ecto.migrate) you can get error reports within the bors dashboard.
16:29badboyok
16:35badboy2017-07-17T13:04:45.346032+00:00 app[web.1]: Errors
16:35badboy2017-07-17T13:04:45.346032+00:00 app[web.1]:
16:35badboy2017-07-17T13:04:45.346033+00:00 app[web.1]: %{reviewer: [{"can't be blank", [validation: :required]}]}
16:42notriddle... weird.
16:42notriddleI don't think that can be related, though.
16:43badboyI pulled, pushed and migrated
16:43badboylet's see if it happens again
16:43badboyI very much appreciate your remote-debugging though
16:44notriddleI think it has to be https://github.com/redis/redis-rb/pull/661 that did it, but I'm not sure how.
16:44boringbotPR #661 [open]: added REPLACE modifier for restore command - https://github.com/redis/redis-rb/pull/661
16:45notriddleIt can't have been 640 or 701, because not-a-robot got the reviewer for both of those PRs (it put them in the merge commit).
16:46notriddle661 looks like it crashed after you reviewed it, because it sent the status update (that's why the GitHub page says "waiting in queue") but didn't actually get to the point where it ran it.
16:46notriddleI'm not sure how though.
16:49notriddleI would've expected "reviewer can't be blank" from a command like `bors r=` (though the parser is supposed to reject it before it gets to the point where it causes a crash).
16:51badboyah
16:51notriddle`bors r+` should've mined the reviewer from your user account, which is not blank.
16:52badboyprobably not helping that redis-rb tests are flaky and looong
16:54notriddlebors-ng is supposed to store enough information in the database that it can safely restart in the middle of a build
16:54* notriddle is not dismissing the possibility that there's a bug related to that, though.
18:37badboynotriddle: d'oh!
18:38badboyhttps://github.com/redis/redis-rb/blob/master/.travis.yml#L32-L34
18:40badboy(ok more like https://github.com/redis/redis-rb/blob/ea9f1d27f7648142f30bd697028c62a3685cbb45/.travis.yml#L32-L34 )
22:16notriddlebadboy: I think I found the bug.
23:07* notriddle should probably build more test harness for this stuff, to reduce the amount of copy-pasted test code.
18 Jul 2017
No messages
   
Last message: 6 days and 17 hours ago