IRC Logs for #circuits Sunday, 2016-07-24

erioluhm... it's already has "ready" label, maybe I miss the PR with the fix?00:00
prologiclemme see00:01
prologicnope no PR yet00:01
prologicgo for it!00:01
*** travis-ci has joined #circuits00:01
travis-cicircuits/circuits#476 (master - c9b9eb0 : Adric Worley): The build has errored.00:01
travis-ciChange view :
travis-ciBuild details :
*** travis-ci has left #circuits ()00:01
prologicready is just a label00:01
eriolI will look at it... I did not used yet. Thanks!00:02
eriolprologic: is there an example of method? After digging a bit, I'm not sure anymore I was doing the right thing. To expire the session, client code has not only to delete self._data[sid], but also clean cookie... or 133 is only about to provide a method to not have to access Sessions object's _data attribute? Presuming client code will have to clean up correctly?01:42
prologicnot sure myself :)03:19
prologicis deleting the Cookie necessary?03:19
*** GeneralAltoids has joined #circuits03:21
GitHub187[circuits] prologic deleted fix_Host_UNIX_web_Server#162 at 55be091:
GitHub161[circuits] prologic deleted fix_travis_builds#166 at 88a7c02:
GitHub132[circuits] prologic deleted add_pull_approve_config at e5c6561:
GitHub82[circuits] prologic deleted fix_broken_Web_UNIX_test_OSX#163 at b7ad135:
GitHub106[circuits] prologic deleted issue_pull_request_templtes at e31f7ce:
*** Osso has joined #circuits03:42
*** Osso has quit IRC03:52
*** robert_ has joined #circuits04:32
*** robert_ has quit IRC04:32
*** robert_ has joined #circuits04:32
prologiceriol: how's #133 going?05:24
Romsterhow's the weekend going prologic05:33
prologicRomster: not bad05:49
prologicwe're meant to be having a mini-Hack-a-thon for circuits05:49
Romsterbeen busy with photocopiers since the boss went away last week.05:50
prologicso far we haven't done much yet :)05:50
Romsteri have yet to have time and mental ability to do much python at all.05:50
GeneralAltoidsprologic: is there any particular reason why BACKLOG (in is capped at 5k concurrent connections?05:55
prologicGeneralAltoids: hi :)05:55
prologicAnd no there is not05:55
prologicwhy do you ask? is there a more sensible default? Or should this be configurable per user application?05:55
Romsterwouldn't you allow as many as the system load dictates over an average?05:56
prologicThe backlog parameter defines the maximum length for the queue of pending connections.  If a connection05:57
prologic     request arrives with the queue full, the client may receive an error with an indication of ECONNREFUSED.05:57
prologic     Alternatively, if the underlying protocol supports retransmission, the request may be ignored so that retries05:57
prologic     may succeed.05:57
Romsterso it's not a max but a backlog buffer max05:58
prologicI feel like this parameter should be configurable05:58
prologicwhere if you fill the queue clients will get an immediate connection refused05:59
prologicbut in a circuits app where things are mostly async and event-driven you may not want this05:59
prologicyou may in fact want an infinite backlog (if that's even possible)05:59
Romsterthings like that should be user configurable since some setups wont need to be as responsive as others05:59
GeneralAltoidsexactly what I was thinking Romster06:00
Romsteruntil system resources are consumed06:00
prologicGeneralAltoids: submit a PR to make this configurable06:00
prologicI'll happily stamp it :)06:00
GeneralAltoidsprologic: well, I'll try, the codebase is a little intimidating for me at the moment. I only found it this existed yesterday after all06:01
prologicwell welcome!06:02
prologicyou'll want to look at circuits/net/sockets.py06:02
prologicand specifically the Server baseclass and TCPServer component06:02
GeneralAltoidsprologic: so this is an interesting case where the default behavior of the underlying socket object differs b/w Python 3.5 and 2.706:10
prologicof course it does06:11
GeneralAltoidsspecifically, in 2.7, the backlog argument is required, and the documentation specifically mentions that the "maximum value is system-dependent (usually 5) [!]"
prologicand in 3.5?06:11
GeneralAltoidswhereas in 3.5, it's been made optional, in which case "a default reasonable value is chosen"
GeneralAltoidsso we have two orthogonal approaches to the concept of backlog across implementations06:12
prologicyeah okay06:13
prologicso I still think that paramter should be configurable06:13
prologicin both versions06:13
prologicfrom our (circuits) end06:13
GeneralAltoidsI think so too, but now I'm wondering what circuits' default value should be, given 3.5's addition "default reasonable value"06:15
GeneralAltoidsapparently the Python developers decided in 3.5 that the default backlog value should be 128, which stands in stark contrast to the 5000 circuits defaults to
GeneralAltoidsso perhaps emulating their default value (in addition to making it configurable on a per-application basis) is desirable06:18
prologicdo it :)06:24
GeneralAltoidsprologic: aye aye captain06:24
prologiccode wins arguments :)06:24
prologicmake sense to me!06:24
GeneralAltoidsprologic: my mother always told me to at least be polite before you go to someone's house and leave a mess :)06:25
prologichaha that's probably good advice06:25
prologicbut circuits has long stopped being "my" project per se :)06:26
prologicit's seen 20+ contributors and 10+ years of effort06:26
prologicso it needs to make sense to the majority :)06:26
Romsteris that 20+ active now or in the past?06:26
prologiccontributors always come and go :)06:27
Romsterindeed they do06:31
Romsteri need to take a look at it again and see if i can do what i need with my code, or hire someone to do it for me06:32
GitHub183[circuits] prologic created improve_test_reliability#174 (+1 new commit):
GitHub183circuits/improve_test_reliability#174 dd0f3fa James Mills: [tests]: Fix test_tcp_lookup failure by using an invalid FQDN06:33
GitHub34[circuits] prologic opened pull request #176: [tests]: Fix test_tcp_lookup failure by using an invalid FQDN (master...improve_test_reliability#174)
prologicRomster: I'm too expensive :)06:34
prologicbut that's an idea for sure06:34
Romsterhow expensive <_<06:34
prologicyou don't wanna know :)06:34
Romsteralso budgies
Romsterwhat $300 an hour?06:34
prologicI've never thought about a consultant rate/hr yet :)06:35
Romstereh i cant explain too well i am best to learn and do it myself06:39
Romsteras you can tell i am more mechanically minded06:40
*** travis-ci has joined #circuits06:46
travis-cicircuits/circuits#477 (improve_test_reliability#174 - dd0f3fa : James Mills): The build has errored.06:46
travis-ciChange view :
travis-ciBuild details :
*** travis-ci has left #circuits ()06:46
GitHub178[circuits] prologic pushed 1 new commit to improve_test_reliability#174:
GitHub178circuits/improve_test_reliability#174 88856db James Mills: [tests]: Fixed test_auto_reconnect node test reliability07:02
*** travis-ci has joined #circuits07:09
travis-cicircuits/circuits#479 (improve_test_reliability#174 - 88856db : James Mills): The build has errored.07:09
travis-ciChange view :
travis-ciBuild details :
*** travis-ci has left #circuits ()07:09
*** robert_|disconne has joined #circuits07:15
*** GeneralAltoids has left #circuits ()07:15
*** robert_ has quit IRC07:17
*** robert_ has joined #circuits07:41
*** robert_ has quit IRC07:41
*** robert_ has joined #circuits07:41
*** robert_|disconne has quit IRC07:43
*** robert_|disconne has joined #circuits07:43
*** robert_ has quit IRC07:44
*** robert_|disconne has quit IRC12:12
eriolprologic: sorry it was to late and I fell asleep... :/ Well, if it's only to expose a method to provide del self._data[sid], it's ready, but I did want to test it so, I was asking for an example for the save method since it would be used in the same way.17:23
GitHub127[circuits] eriol created expire-method-for-web-sessions (+1 new commit):
GitHub127circuits/expire-method-for-web-sessions a0bdf7e Daniele Tricoli: Expose an expire method for Sessions objects17:29
eriolprologic: ↑17:31
*** travis-ci has joined #circuits17:36
travis-cicircuits/circuits#481 (expire-method-for-web-sessions - a0bdf7e : Daniele Tricoli): The build has errored.17:36
travis-ciChange view :
travis-ciBuild details :
*** travis-ci has left #circuits ()17:36
prologiceriol: ahh cool I think I saw your committ?18:11
prologiceriol: you could (if we dont already) write a whole integration test for sessions?18:14
prologicI also think something to keep in mind is subclassing the implementation18:15
prologicso other users can easily have a database backed session(s)18:15
eriolprologic: yes, I want to write tests about it! I did not create a PR since I consider that branch a WIP.19:49
prologicno worries19:51
prologicI'm working in improving tests on OS X19:51
eriolAlso since I'm the less experienced about circuits codebase, I will make a lot of questions, even silly to you and spaceone! :)19:51
eriolprologic: \o/ for OS X tests!19:52
prologicdon't worry :)19:52
eriolsorry for yesterday, it was really to late...19:52
prologicI crahsed too :)19:52
eriolnext time I will take a nap during the afternoon! ;) It was my fist time with 9 hours of difference due to timezones. :)19:54
eriolfor a simple typo in one of the docs examples can I just push to master? I think so...20:05
eriolgit di20:05
eriolsorry wrong terminal :)20:06
prologicyeah go for it :)20:20
GitHub81[circuits] eriol pushed 1 new commit to master:
GitHub81circuits/master e328ed4 Daniele Tricoli: Fix typo in one example20:37
*** travis-ci has joined #circuits20:50
travis-cicircuits/circuits#482 (master - e328ed4 : Daniele Tricoli): The build has errored.20:50
travis-ciChange view :
travis-ciBuild details :
*** travis-ci has left #circuits ()20:50

Generated by 2.14.0 by Marius Gedminas - find it at!