#jython IRC Log


IRC Log for 2013-10-17

Timestamps are in GMT/BST.

[0:17] * lheuer1 (~Adium@f049071174.adsl.alicedsl.de) has joined #jython
[0:19] * lheuer (~Adium@unaffiliated/lheuer) Quit (Ping timeout: 256 seconds)
[0:29] * fwierzbicki (~Adium@99-106-170-162.lightspeed.sntcca.sbcglobal.net) has joined #jython
[0:54] * vIkSiT (~vIkSiT|2@unaffiliated/viksit) has joined #jython
[0:57] * zz_whg is now known as whg
[0:57] <jimbaker> vIkSiT, that's another branch
[0:58] <jimbaker> i spent a day going through that obscure set of test cases
[0:58] <vIkSiT> jimbaker, sorry i got disconn earlier after asking you..
[0:58] <jimbaker> definitely some sort of strange allocation behavior
[0:58] <vIkSiT> did you mean that ssl + large collection iare different branches?
[0:58] <vIkSiT> are*
[0:58] <jimbaker> yes
[0:58] <jimbaker> as i would generally prefer them
[0:59] <vIkSiT> hehe
[0:59] <jimbaker> useful for my dev, less so for you
[0:59] <vIkSiT> yeah!
[0:59] <vIkSiT> hmm, so i'm out of luck..
[0:59] <vIkSiT> unless i refactor code. oh well
[1:00] <jimbaker> vIkSiT, i do plan to fix, with this short term fix and a better long term fix
[1:00] <jimbaker> vIkSiT, so there's one more option for you: import pycimport
[1:01] <jimbaker> i don't know if it will work w/ 2.7 bytecode however
[1:01] <jimbaker> from cpython
[1:01] <jimbaker> but for 2.5, definitely
[1:01] <jimbaker> so if you can build a pyc file, you can get it imported in that way
[1:01] <jimbaker> :)
[1:02] <jimbaker> more of a demonstration of what is possible for the long term case, not so practical. but might be the workaround you need
[1:02] <jimbaker> *not so practical now* - i plan to support this in general in the future
[1:18] * vIkSiT (~vIkSiT|2@unaffiliated/viksit) Quit (Quit: This computer has gone to sleep)
[1:43] * lheuer1 (~Adium@f049071174.adsl.alicedsl.de) Quit (Quit: Leaving.)
[3:12] <sinsnare|zzZZzz> jimbaker: is there a way to pack arbitrary folders into the jar created?
[3:24] * thereisnospoon_ (~thereisno@27-33-1-87.tpgi.com.au) Quit (Remote host closed the connection)
[3:26] * whg is now known as zz_whg
[4:14] * r0bby_ (~wakawaka@guifications/user/r0bby) has joined #jython
[5:17] * lheuer (~Adium@f049071174.adsl.alicedsl.de) has joined #jython
[5:17] * lheuer (~Adium@f049071174.adsl.alicedsl.de) Quit (Changing host)
[5:17] * lheuer (~Adium@unaffiliated/lheuer) has joined #jython
[5:20] <jimbaker> sinsnare|zzZZzz, this should be feasible to do, but we need to work through the details w/ standalone support
[5:20] <jimbaker> which is sort of funky in jython
[5:21] <jimbaker> and needs to be less so :)
[5:22] <jimbaker> later this week, i will take a look to see what can be now, then consider what changes need to be made to standalone
[6:37] * Oti (~ohumbel@adsl-178-39-149-166.adslplus.ch) Quit (Quit: Oti)
[7:23] * jarekps (~jsmiejcza@office.clearcode.cc) has joined #jython
[9:03] * jarekps_ (~jsmiejcza@office.clearcode.cc) has joined #jython
[9:06] * jarekps (~jsmiejcza@office.clearcode.cc) Quit (Ping timeout: 272 seconds)
[9:06] * jarekps_ is now known as jarekps
[10:52] <sinsnare|zzZZzz> i figure so, but just making sure :)
[13:07] * zz_whg is now known as whg
[13:17] * r0bby_ (~wakawaka@guifications/user/r0bby) Quit (Ping timeout: 240 seconds)
[13:17] * thereisnospoon (~thereisno@c58-107-8-85.fitzg4.qld.optusnet.com.au) has joined #jython
[13:43] * enebo (~enebo@c-75-73-8-169.hsd1.mn.comcast.net) has joined #jython
[14:47] * dso (~dso@64-148-242-8.lightspeed.snantx.sbcglobal.net) has joined #jython
[17:02] * lheuer (~Adium@unaffiliated/lheuer) Quit (Quit: Leaving.)
[17:27] * lheuer (~Adium@f049071174.adsl.alicedsl.de) has joined #jython
[17:27] * lheuer (~Adium@f049071174.adsl.alicedsl.de) Quit (Changing host)
[17:27] * lheuer (~Adium@unaffiliated/lheuer) has joined #jython
[17:35] * lheuer (~Adium@unaffiliated/lheuer) Quit (Ping timeout: 260 seconds)
[17:35] * lheuer (~Adium@f048141069.adsl.alicedsl.de) has joined #jython
[17:35] * lheuer (~Adium@f048141069.adsl.alicedsl.de) Quit (Changing host)
[17:35] * lheuer (~Adium@unaffiliated/lheuer) has joined #jython
[19:31] * dso (~dso@64-148-242-8.lightspeed.snantx.sbcglobal.net) Quit (Ping timeout: 245 seconds)
[20:21] * paolo (~Paolo@net-2-36-97-216.cust.dsl.vodafone.it) has joined #jython
[20:40] * dso (~dso@64-148-242-8.lightspeed.snantx.sbcglobal.net) has joined #jython
[21:10] * paolo (~Paolo@net-2-36-97-216.cust.dsl.vodafone.it) Quit (Quit: Leaving)
[21:30] * sinsnare|zzZZzz is now known as sinistersnare
[21:30] * vIkSiT (~vIkSiT|2@unaffiliated/viksit) has joined #jython
[21:30] <jimbaker> sinsnare|zzZZzz, so you can definitely use setuptools to add data files
[21:30] <vIkSiT> jimbaker, hehehe, just saw your pycimport comment
[21:31] <jimbaker> and access w/ pkgutil. ideally you can use pkg_resources, but this is not seeming to be included, nor setuptools
[21:31] <jimbaker> vIkSiT, yes, that would be sort of an interesting workaround
[21:32] <jimbaker> i know people have used it in the past. should just try now w/ 2.7
[21:32] <jimbaker> i suspect that 2.7 changed the "magic number" for bytecode, plus added some instructions
[21:33] <jimbaker> but that should be easy to figure out
[21:34] <jimbaker> i also asked shashank1 about his compiler work, so we can use that to frontend python bytecode compilation. basically python bytecode gets some nice capabilities, such as faster startup time, opportunity to profile, ability to deoptimize, etc
[21:34] <jimbaker> also android support for dynamic code and big method support
[21:35] <jimbaker> it of course is slower than java bytecode, especially when hotspot has kicked in
[21:36] <jimbaker> otherwise, we would just use it :)
[21:37] <sinistersnare> that sounds cool
[21:37] <sinistersnare> i really have no experience with setuptools, so ill do some reasearch on the topic!
[21:42] <jimbaker> vIkSiT, i just tried pycimport and works just fine w/ 2.7 bytecode
[21:46] <vIkSiT> ah
[21:46] <vIkSiT> what are your views on performance implications of jython vs cpython btw?
[21:46] <vIkSiT> rather, moving from the latter to former
[21:46] <jimbaker> best course of action: in cpython, use compileall.compile_dir("my-dir")
[21:46] <jimbaker> remove *.py files, leaving just pyc files
[21:47] <jimbaker> then use import pycimport to enable importing
[21:47] <jimbaker> by using compileall, you can actually use jython code, including java imports, w/o problems - it's just doing bytecode compiling
[21:48] <jimbaker> well, jython is potentially faster than cpython for python code, and possibly even pypy. but??? it isn't right now
[21:49] <jimbaker> it's arguably easier presumably to optimize by hand into java critical performance sections than C
[21:49] <jimbaker> but people use jython because they need to use the java ecosytem
[21:50] <sinistersnare> i think that jython can be much faster; JRuby is much faster than MRI (the reference Ruby interpreter)
[21:50] <sinistersnare> so i think that if we could get some steam on this, it could be great
[21:57] * whg is now known as zz_whg
[21:59] * fwierzbicki (~Adium@99-106-170-162.lightspeed.sntcca.sbcglobal.net) Quit (Quit: Leaving.)
[21:59] <vIkSiT> I see
[21:59] <vIkSiT> so what are the limiting factors that don't make it fast?
[22:00] <vIkSiT> fast == as fast as it could theoretically be
[22:06] <sinistersnare> vIkSiT: im guessing that were not really bringing in a lot of optimizations right now; but im not too savvy on this
[22:07] <sinistersnare> if they spent the time to optimize and maybe implement invokeDynamic, i think the performance would be good
[22:16] <jimbaker> shashank1 did his masters thesis on looking at optimizations for jython
[22:17] <jimbaker> invokedynamic is a key compoennt
[22:17] <jimbaker> there are some students at uc irvine who are now looking at graal/truffle
[22:17] <sinistersnare> thats cool
[22:17] <jimbaker> for jython
[22:17] <sinistersnare> maybe i should do something cool at UMD with jython
[22:18] <jimbaker> sounds like a good idea
[22:18] <sinistersnare> need to get accepted first, but i wrote my essay on how confident i am on getting in there...
[22:18] <jimbaker> :)
[22:23] * enebo (~enebo@c-75-73-8-169.hsd1.mn.comcast.net) Quit (Quit: enebo)
[22:27] * dso (~dso@64-148-242-8.lightspeed.snantx.sbcglobal.net) has left #jython
[22:33] <jimbaker> vIkSiT, so again i would encourage you to look at pycimport. i should also look at running the entire regrtest w/ pyc again
[22:33] * vIkSiT (~vIkSiT|2@unaffiliated/viksit) Quit (Quit: This computer has gone to sleep)
[22:33] <jimbaker> there are a couple of minor differences in floatingpoint representation, but nothing significant
[22:34] <jimbaker> due to how cpython and jython will marshal values differently
[22:35] <jimbaker> given some flexibility in the underlying IEEE standard
[22:36] <jimbaker> (this would go away if we compiled our own python bytecode)
[22:46] * fwierzbicki (~Adium@ip-64-134-232-205.public.wayport.net) has joined #jython
[22:50] * vIkSiT (~vIkSiT|2@unaffiliated/viksit) has joined #jython
[22:51] <sinistersnare> vIkSiT: check the logs, words were said
[22:55] <vIkSiT> thanks sinistersnare, just saw
[22:55] <vIkSiT> looks like the main issues are the pycimport
[23:17] * vIkSiT (~vIkSiT|2@unaffiliated/viksit) Quit (Quit: This computer has gone to sleep)
[23:19] * Scorp1us (~Scorp1us@75-145-94-105-WashingtonDC.hfc.comcastbusiness.net) Quit (Read error: Connection reset by peer)
[23:19] * Scorp1us (~Scorp1us@75-145-94-105-WashingtonDC.hfc.comcastbusiness.net) has joined #jython
[23:27] * fwierzbicki (~Adium@ip-64-134-232-205.public.wayport.net) Quit (Quit: Leaving.)
[23:50] <jimbaker> ahh, i see what's going on. zipped eggs have some additional header info in front of the PK marker, so they don't get recognized
[23:50] <jimbaker> curiously, ZipInputStream will open, then quietly discard


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