IRC Logs for #circuits Sunday, 2017-01-29

*** Coldblackice has quit IRC04:10
GitHub121[circuits] prologic force-pushed Issue#133 from 1b72fad to 46589db: https://git.io/vM7N004:38
GitHub121circuits/Issue#133 050fc48 James Mills: [web]: Improve the API for session management and add expire support....04:38
GitHub121circuits/Issue#133 46589db James Mills: Address @spaceone's feedback04:38
*** travis-ci has joined #circuits04:42
travis-cicircuits/circuits#655 (Issue#133 - 46589db : James Mills): The build was broken.04:42
travis-ciChange view : https://github.com/circuits/circuits/compare/1b72fad83f51...46589dbcb31c04:42
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627421404:42
*** travis-ci has left #circuits ()04:42
*** travis-ci has joined #circuits04:44
travis-cicircuits/circuits#655 (Issue#133 - 46589db : James Mills): The build was broken.04:44
travis-ciChange view : https://github.com/circuits/circuits/compare/1b72fad83f51...46589dbcb31c04:44
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627421404:44
*** travis-ci has left #circuits ()04:44
*** travis-ci has joined #circuits04:46
travis-cicircuits/circuits#655 (Issue#133 - 46589db : James Mills): The build was broken.04:46
travis-ciChange view : https://github.com/circuits/circuits/compare/1b72fad83f51...46589dbcb31c04:46
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627421404:46
*** travis-ci has left #circuits ()04:46
prologicoh boy04:49
prologicsorry I've been idle for a while04:49
prologicbeen busy at work04:50
prologicI've caught up with all the nice PR(s) and Commits04:50
prologicanything else I miss?04:50
*** Coldblackice has joined #circuits04:52
GitHub116[circuits] prologic force-pushed process_tasks_faster from be1a2b9 to 77c36d5: https://git.io/vDUCS04:54
GitHub116circuits/process_tasks_faster 77c36d5 James Mills: [core]: Improve performance of coroutines/tasks and fix event.stop()...04:54
GitHub22[circuits] prologic pushed 1 new commit to process_tasks_faster: https://git.io/vDUCF04:58
GitHub22circuits/process_tasks_faster a8c50d3 James Mills: Address @apollo13's feedback04:58
*** travis-ci has joined #circuits05:01
travis-cicircuits/circuits#657 (process_tasks_faster - 77c36d5 : James Mills): The build is still failing.05:01
travis-ciChange view : https://github.com/circuits/circuits/compare/be1a2b9c06f9...77c36d5cdf0705:01
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627559905:01
*** travis-ci has left #circuits ()05:01
*** travis-ci has joined #circuits05:01
travis-cicircuits/circuits#657 (process_tasks_faster - 77c36d5 : James Mills): The build is still failing.05:01
travis-ciChange view : https://github.com/circuits/circuits/compare/be1a2b9c06f9...77c36d5cdf0705:01
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627559905:01
*** travis-ci has left #circuits ()05:01
*** travis-ci has joined #circuits05:13
travis-cicircuits/circuits#659 (process_tasks_faster - a8c50d3 : James Mills): The build is still failing.05:13
travis-ciChange view : https://github.com/circuits/circuits/compare/77c36d5cdf07...a8c50d39850405:13
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627602905:13
*** travis-ci has left #circuits ()05:13
*** travis-ci has joined #circuits05:14
travis-cicircuits/circuits#659 (process_tasks_faster - a8c50d3 : James Mills): The build is still failing.05:14
travis-ciChange view : https://github.com/circuits/circuits/compare/77c36d5cdf07...a8c50d39850405:14
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627602905:14
*** travis-ci has left #circuits ()05:14
*** travis-ci has joined #circuits05:14
travis-cicircuits/circuits#659 (process_tasks_faster - a8c50d3 : James Mills): The build is still failing.05:14
travis-ciChange view : https://github.com/circuits/circuits/compare/77c36d5cdf07...a8c50d39850405:14
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627602905:14
*** travis-ci has left #circuits ()05:14
GitHub179[circuits] prologic pushed 1 new commit to Issue#133: https://git.io/vDUWw05:23
GitHub179circuits/Issue#133 18520d1 James Mills: wtf? Fixed Import order :)05:23
*** travis-ci has joined #circuits05:31
travis-cicircuits/circuits#661 (Issue#133 - 18520d1 : James Mills): The build was fixed.05:31
travis-ciChange view : https://github.com/circuits/circuits/compare/46589dbcb31c...18520d19241e05:31
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627845505:31
*** travis-ci has left #circuits ()05:31
*** travis-ci has joined #circuits05:34
travis-cicircuits/circuits#661 (Issue#133 - 18520d1 : James Mills): The build was fixed.05:34
travis-ciChange view : https://github.com/circuits/circuits/compare/46589dbcb31c...18520d19241e05:34
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627845505:34
*** travis-ci has left #circuits ()05:34
*** travis-ci has joined #circuits05:35
travis-cicircuits/circuits#661 (Issue#133 - 18520d1 : James Mills): The build was fixed.05:35
travis-ciChange view : https://github.com/circuits/circuits/compare/46589dbcb31c...18520d19241e05:35
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627845505:35
*** travis-ci has left #circuits ()05:35
GitHub33[circuits] prologic force-pushed simplify_HTTP_properties from df4d609 to ebc2e4a: https://git.io/vM7oR05:37
*** travis-ci has joined #circuits05:38
travis-cicircuits/circuits#661 (Issue#133 - 18520d1 : James Mills): The build was fixed.05:38
travis-ciChange view : https://github.com/circuits/circuits/compare/46589dbcb31c...18520d19241e05:38
travis-ciBuild details : https://travis-ci.org/circuits/circuits/builds/19627845505:38
*** travis-ci has left #circuits ()05:38
GitHub178[circuits] spaceone pushed 1 new commit to starttls: https://git.io/vDUV212:03
GitHub178circuits/starttls 8eb6c62 ⁣ Florian Best: Merge branch 'master' into starttls12:03
GitHub92[circuits] spaceone force-pushed socket_options from 74a56a7 to b478492: https://git.io/vMAyS12:06
GitHub92circuits/socket_options b478492 SpaceOne: Add support to set additional socket options...12:06
GitHub190[circuits] spaceone pushed 1 new commit to master: https://git.io/vDUwt12:17
GitHub190circuits/master be3405a SpaceOne: Add support to set additional socket options...12:17
GitHub98[circuits] spaceone closed pull request #215: Add support to set additional socket options (master...socket_options) https://git.io/vMNrJ12:17
GitHub26[circuits] spaceone deleted socket_options at b478492: https://git.io/vDUwY12:18
GitHub75[circuits] spaceone force-pushed starttls from 8eb6c62 to 0c8e9dd: https://git.io/vMFxH12:24
GitHub75circuits/starttls 0c8e9dd SpaceOne: Implement STARTTLS event; Fix #113...12:24
GitHub194[circuits] spaceone force-pushed starttls from 0c8e9dd to 27c689e: https://git.io/vMFxH12:50
GitHub194circuits/starttls 27c689e SpaceOne: Implement STARTTLS event; Fix #113...12:50
riothmm, i'm trying to use circuits.app.daemon but get an error from the threads: http://pastebin.com/Ej9JJ1Gb is there anything i should watch out for?14:36
GitHub19[circuits] prologic force-pushed Issue#133 from 18520d1 to 2d07389: https://git.io/vM7N017:56
GitHub19circuits/Issue#133 1cfe87c James Mills: [web]: Improve the API for session management and add expire support....17:56
GitHub19circuits/Issue#133 30f16f6 James Mills: Address @spaceone's feedback17:56
GitHub19circuits/Issue#133 2d07389 James Mills: wtf? Fixed Import order :)17:56
prologicriot: how are you using Daemon() ?17:57
prologicriot: without further context/tracebacks that kind of looks like a Python 3.5 bug :)18:09
riotin my core component, I'm doing this at the end of __init__: http://pastebin.com/bkEzcGQK18:24
riotand the same happens with 3.6 - also: it starts eating memory until my ram is full after forking.18:24
riotin other news: debian python app packaging is a HUGE pita18:29
riotI'm suddenly unhappy with a few crucial design choices.18:29
apollo13riot: bzw since I've been reading the circuits source quite a bit, all you need for zeromq is to write your own poller which you register19:06
apollo13you can more or less copy the select poller, should be quite easy19:06
apollo13no need for extra threads then19:06
rioti stumbled over the poller implementations and had similar thoughts :>19:08
apollo13I mean that is assuming that the python variant exposes http://api.zeromq.org/2-1:zmq-poll19:09
riotwhat i saw was this: https://github.com/zeromq/pyzmq/blob/master/zmq/sugar/poll.py19:11
apollo13ah zmq_pool exists, that makes it easy19:11
apollo13ah and that poller is even part of zmq, then it is really just replacing the poller class of the backend19:13
riotokay, without the zmq-thread i don't get any weird error messages from the thread anymore, it still eats 100% cpu to eat all my ram, after Daemon.register(self)19:14
riotaah, the logfile says more19:22
riotoh, wow, i built an unstoppable forkbomb19:33
riotit keeps forking, then crashes. pid changes every call, obviously.19:34
riothmmm. not so good.19:34
riotcaught that. How do i keep ports open when daemonizing? It closes them. Not so useful :(19:47
rioti think i'll rather go with a stupid systemd-style daemon that is not a real daemon.. *sigh*19:48
apollo13riot: python either sets close on exec by default or closes sockets on forks by default in newer python versions20:03
apollo13riot: https://docs.python.org/3/whatsnew/3.4.html#pep-446-newly-created-file-descriptors-are-non-inheritable20:03
riotso, in essence i can't fix that.20:06
apollo13why not?20:07
riothmm. I'd need to fix circuits - which can be done, again.20:07
riotby giving the TCPServer a new init-argument20:08
riotjust doing socket.set_inheritable(True) seems to do the job.20:08
riotaah, next problem: certificate isn't found after daemonizing. Hoorah20:09
apollo13imo you should open the server after forking unless you have a good reason to do it before20:10
apollo13is there one?20:10
riotyes.20:10
riotworks like a charm if i do not daemonize.20:10
apollo13that is besides the point20:10
riotprobably not anynore because forking changes the directory20:10
apollo13if you daemonize (ie fork) you really don't want the socket in both processes20:10
apollo13which makes it questionable why you are opening the server before the fork in the first place20:11
apollo13always use full paths ;)20:11
rioti am not opening it before. I have three timers: 1 sec after start: daemonize (probably works without a timer as well) 3 secs -> open port, 5 secs -> drop privileges20:13
rioti get an error upon server start "bad file descriptor" but it works anyway.. weird20:13
apollo13why are you using timers for that?20:14
riothad those in there, just copied them, thought it might be safer to wait so everything can settle20:14
apollo13well there are plenty of events like started and ready which are fired and you should be able to use20:15
riotfsck this, its not worthwhile and gives me the creeps about other problems coming up.20:26
rioti play the stupid systemd-variant. Actually much more sane than trying to accomplish all that via true daemon-forking20:27
riot(also i still don't really see the benefits here.)20:27
apollo13I'd probably use systemd myself where applicable :D20:45
apollo13you get nice monitoring/restarts/etc20:45

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!