#jython IRC Log (v0.9)


IRC Log for 2014-09-24

Timestamps are in GMT/BST.

[0:49] * Aylin_Pagac22 (~nodebot@li604-225.members.linode.com) has joined #jython
[1:25] * Aylin_Pagac22 (~nodebot@li604-225.members.linode.com) Quit (Remote host closed the connection)
[1:50] * Arlie36 (~nodebot@li604-225.members.linode.com) has joined #jython
[2:16] * Tayl (~Taylor@unaffiliated/taylor) has joined #jython
[2:19] * Tayl (~Taylor@unaffiliated/taylor) has left #jython
[2:39] * Arlie36 (~nodebot@li604-225.members.linode.com) Quit (Remote host closed the connection)
[3:54] * Ari55 (~selbyk@li604-225.members.linode.com) has joined #jython
[4:06] * mary_ (~ircfnode@moya.miuark.net) Quit (Ping timeout: 246 seconds)
[4:06] * mary_ (~ircfnode@moya.miuark.net) has joined #jython
[4:18] * Ari55 (~selbyk@li604-225.members.linode.com) Quit (K-Lined)
[5:46] * frase (~Fraser@110-174-235-130.static.tpgi.com.au) has joined #jython
[5:58] <frase> I have encountered some bugs in Jython. Since non-subscribers can't send mail to the jython-dev@ and unregistered users cannot create new bugs in the tracker - and since I care to subscribe or create an account - here are some gists:
[5:58] <frase> https://gist.github.com/frasertweedale/c43cda1b4d1efca3022a
[5:58] <frase> https://gist.github.com/frasertweedale/03bf0d0dc69e4461db76
[6:12] <agronholm> frase: jython-standalone is meant for embedding jython into existing java apps -- can you reproduce this with the standard install too?
[6:12] <frase> agronholm: will give that a go shortly
[6:13] <agronholm> I may have created an actual ticket about __dict__ not behaving like a dict
[6:14] <frase> I saw one ticket that may have my issue in scope: http://bugs.jython.org/issue1631
[6:14] <frase> there is one patch already but I'm unsure whether it has provided the dictview methods
[6:40] <frase> agronholm: right you are; the full installation runs ``-m unittest discover`` just fine
[6:41] <frase> except by "fine" I really mean "the test suite does run, but all the tests fail due to the other issue" ^_^
[6:41] <agronholm> :)
[6:42] <agronholm> why are you touching __dict__ btw?
[6:43] <frase> not touching __dict__ directly, actually, but the same object is received as an argument to the metaclass __new__ method
[6:45] <frase> i.e. def __new__(mcs, name, bases, dict): ... <-- that `dict' arg lacks essential methods of dict :)
[6:48] <frase> agronholm: updated the gist with metaclass __new__ details
[13:09] * xemdetia (xemdetia@nat/ibm/x-vkaystztyzaddkxb) has joined #jython
[13:59] * enebo (~enebo@c-75-73-8-169.hsd1.mn.comcast.net) has joined #jython
[14:04] * mike_af (~mike_af@50-77-49-46-static.hfc.comcastbusiness.net) has joined #jython
[14:04] * mike_af (~mike_af@50-77-49-46-static.hfc.comcastbusiness.net) Quit (Changing host)
[14:04] * mike_af (~mike_af@unaffiliated/mike-af/x-5454762) has joined #jython
[14:34] * a42niem (~niem@p578b7daf.dip0.t-ipconnect.de) has joined #jython
[14:38] * enebo (~enebo@c-75-73-8-169.hsd1.mn.comcast.net) Quit (Quit: enebo)
[14:48] * mary_ (~ircfnode@moya.miuark.net) Quit (Quit: Reconnecting)
[14:48] * mary_ (~ircfnode@moya.miuark.net) has joined #jython
[14:49] * mary_ is now known as mary
[14:49] * mary (~ircfnode@moya.miuark.net) Quit (Changing host)
[14:49] * mary (~ircfnode@unaffiliated/mary) has joined #jython
[15:09] * enebo (~enebo@c-75-73-8-169.hsd1.mn.comcast.net) has joined #jython
[17:57] <jimbaker> frase, should be straightforward to add these missing methods to PyStringMap. from an internals perspective it is a different bug than http://bugs.jython.org/issue1631
[17:57] <jimbaker> but we can use the same approach in terms of testing
[17:57] <jimbaker> using test_dict
[18:06] <jimbaker> it is the lack of such testing that causes these issues, since it's easy enough to fix - there's some unfortunate discrepancies because PyStringMap used to implement an alternative hashing strategy in comparison to PyDictionary, which was probably good around the days of java 1.2. it since uses the same underlying ConcurrentHashMap, but there's enough code in jython's runtime that depends on its api that we cannot readily remove
[18:06] <jimbaker> maybe with jython 3
[18:07] * enebo (~enebo@c-75-73-8-169.hsd1.mn.comcast.net) Quit (Quit: enebo)
[18:29] <jimbaker> frase, i update the corresponding bug - http://bugs.jython.org/issue1152612
[18:40] * enebo (~enebo@ has joined #jython
[18:42] * gthank (~gthank@unaffiliated/gthank) has joined #jython
[19:25] <jimbaker> nickmbailey, you may have seen the fix in select is now in trunk
[19:27] <nickmbailey> i didn't see but awesome!
[19:27] <nickmbailey> thanks a ton
[19:27] <nickmbailey> actually speaking of select, at least with twisted and cpython, the select reactor has a 1024 limit on file descriptors. you don't happen to know if that limitation would exist on jython do you?
[19:28] <jimbaker> nickmbailey, well, it's likely the limit is much higher on jython due to the use of underlying netty 4
[19:28] <jimbaker> it would be great to benchmark of course
[19:29] <jimbaker> and see how things possibly blow up :)
[19:29] <nickmbailey> are there alternative jython reactors do you know?
[19:30] <nickmbailey> honestly i'm not clear on where the 1024 limit comes from with the select reactor, but the epoll and kqueue reactors don't have it
[19:30] <jimbaker> nickmbailey, the other thing is, at some point we should also write a reactor specific to jython that uses netty. because there's a lot of overhead in maintaining the fiction that socket has its c style api
[19:30] <jimbaker> so no jython-specific reactors other than what i suggested
[19:31] * teuchter is now known as choochter
[19:31] <jimbaker> but i would expect using netty should get closer to what we see with epoll. lots of stuff out there that suggests it scales well, and twisted's api is a natural fit
[19:32] <nickmbailey> interesting
[19:32] <nickmbailey> i donno if it'll be extremely soon but i'll be able to see how far i can push it and let you know :)
[19:38] <jimbaker> nickmbailey, you might find this doc interesting - https://github.com/jimbaker/socket-reboot/blob/master/README.md
[19:38] <jimbaker> it describes this mapping. you should readily see why a jython-specific reactor would do better :)
[19:39] <jimbaker> however, i like the idea of testing with the select reactor, for all the obvious reasons of stress testing this wrapper
[19:44] * xemdetia (xemdetia@nat/ibm/x-vkaystztyzaddkxb) Quit (Quit: Leaving)
[19:45] <nickmbailey> jimbaker: thanks :) i'll take a look
[20:10] <jimbaker> nickmbailey, another possible source of inspiration: https://github.com/fiorix/niosted
[21:37] * gthank (~gthank@unaffiliated/gthank) Quit (Ping timeout: 258 seconds)
[22:04] * a42niem (~niem@p578b7daf.dip0.t-ipconnect.de) Quit (Quit: Leaving.)
[22:06] * fwierzbicki (~Adium@99-106-169-5.lightspeed.sntcca.sbcglobal.net) Quit (Quit: Leaving.)
[22:13] * enebo (~enebo@ Quit (Quit: enebo)
[22:17] * xemdetia (xemdetia@nat/ibm/x-pewykpaxgzcgzfxs) has joined #jython
[22:39] * mike_af (~mike_af@unaffiliated/mike-af/x-5454762) Quit (Remote host closed the connection)
[22:43] * fwierzbicki (~Adium@78.sub-70-197-14.myvzw.com) has joined #jython
[22:59] * xemdetia (xemdetia@nat/ibm/x-pewykpaxgzcgzfxs) Quit (Ping timeout: 260 seconds)
[23:32] * fwierzbicki (~Adium@78.sub-70-197-14.myvzw.com) Quit (Quit: Leaving.)
[23:44] * robbyoconnor (~wakawaka@guifications/user/r0bby) has joined #jython


These logs were automatically created by JythonLogBot_ on irc.freenode.net using a slightly modified version of the Java IRC LogBot (github).