IRC Logs for #circuits Tuesday, 2017-01-17

*** robert_ has joined #circuits00:58
*** robert_ has quit IRC00:58
*** robert_ has joined #circuits00:58
*** robert_|disconne has joined #circuits07:33
*** robert_ has quit IRC07:33
*** Workster has quit IRC09:57
*** Workster has joined #circuits10:00
*** Workster has quit IRC10:00
*** Workster has joined #circuits10:00
apollo13prologic: ah lol, didn't realize that the manager is running in an extra thread in the testruns, now the code in watcher.wait makes sense again10:38
GitHub26[circuits] apollo13 opened pull request #196: [WIP] Increase test verbosity to debug testfailures. (master...testfailures)
apollo13mhm, the event is not even getting send11:11
apollo13prologic, spaceone: can you look at and tell me anything out of the unordinary you see -- it should (tm) fire an event (return_value) which it does not do (without any error)11:19
spaceoneapollo13: maybe you need to watcher.clear()11:39
spaceoneprior to watcher.wait()11:39
apollo13spaceone: why?11:39
spaceoneto clear the previous wait()-call-results11:39
apollo13spaceone: shouldn't matter, the event is not even triggered11:39
spaceoneprologic: apollo13: there was one gui which requested not to remove python2.6 in the master branch11:39
apollo13at least according to the debugger output11:40
apollo13spaceone: from what I read, they were fine with pinning a version, ie 3.211:40
spaceonedoes it fail on python2.7 ?11:40
apollo13it fails everywhere but only sometimes11:41
spaceoneACTION is @work11:41
apollo13and I cannot repor locally11:41
apollo13it's as if node_server.send() just does not fire the event11:41
apollo13but no event reaches the debugger11:41
spaceoneit's not reproducible locally?11:42
apollo13or maybe I am just lucky locally, let me run in a while loop11:43
spaceonepassed for me, too11:47
apollo13so the main question is literally: how can .fire(event) not end up somewhere11:48
apollo13I even fixed the hashseed to match the one fromt he failing env etc…11:49
spaceonewhat is the value of 'event'?11:52
apollo13119     event = return_value()11:52
apollo13120     app.server.send(event, app.server.get_socks()[0], no_result=True)11:52
apollo13success = True is set also11:53
apollo13gotta add a few prints for the value11:53
spaceoneadd a temporary watcher.clear() before .send()11:55
spaceonemaybe it's the reason why the event is not fired/displayed as fired11:55
apollo13will do, but doubt it11:55
apollo13spaceone: re -- what happens if you do wait([ev1,ev2,ev3]) and ev3 finishes first, will the inner wait for ev3 then hang or does it shor circuit for already finished events?12:02
spaceone'shor' ?12:07
spaceonethe wait-call will be resolved when all events are finished12:07
apollo13ie if the event got already handled does .wait take that into account?12:07
apollo13yes, but basically .wait([ev1, ev2, ev3]) is a shortcut for wait(ev1) wait(ev2) wait(ev3) -- what if the events finish in order of ev3, ev2, ev112:08
spaceonethe order probably doesn't matter. but if you e.g. do:12:08
spaceoneyield self.wait([e1,e2,3])12:09
spaceoneand e* is already finished12:09
spaceonethis will fail12:09
apollo13ah right, but the event instances are passed around and never copied right? so .wait could just check or so?12:10
apollo13ah no, it is the other way12:10
spaceonei need to write a test case for the order12:10
apollo13there is no state on the event object12:10
apollo13might be nice to mark an event as "already handled" though, should fix order problems etc12:11
apollo13hurray, got a local failure too! but only when running the whole suite, never in isolation it seems12:11
apollo13arg /me rage quits12:25
apollo13test_auto_connect does something stupid12:26
apollo13ah, gotcha12:27
apollo13node server do not seem to close on unregister yikes yikes12:28
GitHub9[circuits] apollo13 pushed 1 new commit to master:
GitHub9circuits/master 91443dc Florian Apolloner: Properly closed node server in test_auto_reconnect.12:31
GitHub199[circuits] apollo13 closed pull request #196: [WIP] Increase test verbosity to debug testfailures. (master...testfailures)
*** travis-ci has joined #circuits12:42
travis-cicircuits/circuits#554 (master - 91443dc : Florian Apolloner): The build failed.12:42
travis-ciChange view :
travis-ciBuild details :
*** travis-ci has left #circuits ()12:42
GitHub20[circuits] apollo13 pushed 1 new commit to master:
GitHub20circuits/master 3ac4f61 Florian Apolloner: Simplified node tests....12:45
apollo13spaceone: ok, I am nearly green -- one failure in the web stuff
GitHub167[circuits] apollo13 pushed 1 new commit to master:
GitHub167circuits/master 87af180 Florian Apolloner: Removed verbosity again, travis limits output quite a bit…12:51
*** travis-ci has joined #circuits12:57
travis-cicircuits/circuits#555 (master - 3ac4f61 : Florian Apolloner): The build is still failing.12:57
travis-ciChange view :
travis-ciBuild details :
*** travis-ci has left #circuits ()12:57
apollo13oh cmon, green tests for once please :D12:59
*** travis-ci has joined #circuits13:03
travis-cicircuits/circuits#556 (master - 87af180 : Florian Apolloner): The build is still failing.13:03
travis-ciChange view :
travis-ciBuild details :
*** travis-ci has left #circuits ()13:03
apollo13prologic: can you fix those:
apollo13prologic: everything else seems fixed, but mac is still anooying13:04
spaceoneOSError: [Errno 24] Too many open files13:08
spaceonewhat is the max-open-fd limit?13:08
apollo13no idea, that is on the mac travis boxes13:09
apollo13ACTION has no mac13:09
spaceoneah okay13:11
spaceonebut this seems to be an bug regarding error handling13:11
apollo13yeah, that might also be the case13:11
spaceoneif socket.accept() fails the code should never reach this IndexError function/poller registration13:11
apollo13probably, but having a hard time fixing something I cannot repro :D13:12
apollo13I've added OSX to allowed_failures for now13:13
GitHub7[circuits] apollo13 pushed 1 new commit to master:
GitHub7circuits/master 884b14d Florian Apolloner: Added osx to allowed failures till the errors there are fixed.13:13
apollo13everything else seems to be stable13:13
apollo13and well, pypy, we either seem to be having a threading problem which kills heappush or I do not know13:14
apollo13spaceone: re missing whitespace, I plan to run flake8/isort as travis jobs, thoughts?13:20
spaceonejeah, would be nice13:26
apollo13    273    2641   2282013:28
apollo13could be worse XD13:28
*** travis-ci has joined #circuits13:30
travis-cicircuits/circuits#557 (master - 884b14d : Florian Apolloner): The build was fixed.13:30
travis-ciChange view :
travis-ciBuild details :
*** travis-ci has left #circuits ()13:30
spaceonea lot of them are core/ which is a patch13:31
apollo13spaceone: actually no, -- just stupid stuff13:32
spaceoneah yes13:33
spaceonethis was only a local file of mine13:33
spaceonein circuits/ is only 74 issues13:33
spaceonethis is the important part ^^13:33
apollo13meh, if I get in there I'll fix all of em13:37
spaceoneyou can use autopep8 --select E123 for example13:40
apollo13oh, never used that :D13:41
*** ke4roh__ has joined #circuits13:44
*** robert_ has joined #circuits13:51
*** robert_ has quit IRC13:51
*** robert_ has joined #circuits13:51
*** robert_|disconne has quit IRC13:51
*** ke4roh_ has joined #circuits20:01
*** ke4roh__ has quit IRC20:03
prologichey all20:33
*** ke4roh__ has joined #circuits20:36
*** ke4roh_ has quit IRC20:39
prologicapollo13: thank you for your efforts on circuits!20:44
prologicI hope we can continue to collaborate more ;)20:44
GitHub11[circuits] prologic pushed 1 new commit to master:
GitHub11circuits/master 229ab9c James Mills: Guard against OSError and SocketError when creating the control socket21:33
*** ke4roh__ has quit IRC21:49
apollo13prologic: not going to promise anything, but that is the plan ;) also it is a nice learning experience -- circuits is small enough to dig in, stuff like qt with all it's eventloop abstractions is a whole other story…23:51

Generated by 2.14.0 by Marius Gedminas - find it at!