| harrison (~quassel@adsl-69-209-220-254.dsl.chcgil.ameritech.net) left irc: Read error: Connection reset by peer | 00:10 | |
| harrison (~quassel@adsl-69-209-220-254.dsl.chcgil.ameritech.net) joined #pypy. | 00:10 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/737 | 00:12 |
|---|---|---|
| voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace | 00:13 | |
| voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy. | 00:22 | |
| voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host | 00:22 | |
| voidspace (~anonymous@python/psf/voidspace) joined #pypy. | 00:22 | |
| overminder (~overminde@wf119-103.ust.hk) joined #pypy. | 00:27 | |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 260 seconds | 00:27 | |
| derdon (~derdon@pD9E1DEEC.dip.t-dialin.net) left irc: Remote host closed the connection | 00:36 | |
| amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 248 seconds | 00:43 | |
| Circlefusion (~cf@74-138-199-109.dhcp.insightbb.com) left irc: Ping timeout: 252 seconds | 00:44 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Remote host closed the connection | 00:44 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy. | 00:45 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Read error: Connection reset by peer | 00:46 | |
| aurojit (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy. | 00:46 | |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 00:47 | |
| Circlefusion (~cf@74-138-199-109.dhcp.insightbb.com) joined #pypy. | 00:58 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/163 | 01:08 |
| Alex_Gaynor | responsible thing: split writing paper over two nights (at least). gaynor thing: write ten page paper in one night, start at 10:30 | 01:11 |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 258 seconds | 01:14 | |
| rguillebert | never happened to me 0:) | 01:14 |
| whitelynx (~whitelynx@75.110.189.230) left irc: Ping timeout: 244 seconds | 01:18 | |
| Alex_Gaynor | err, wrong channel :) | 01:19 |
| voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace | 01:21 | |
| gutworth (~benjamin@dhcp-128-36-84-91.central.yale.edu) left irc: Ping timeout: 240 seconds | 01:29 | |
| stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) joined #pypy. | 01:31 | |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 01:31 | |
| whitelynx (~whitelynx@75.111.197.204) joined #pypy. | 01:34 | |
| Nick change: Gulaway -> Gulopine | 01:35 | |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1881 | 01:37 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/618 | 01:37 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/635 | 01:37 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1146 | 01:37 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/298 | 01:37 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/820 | 01:37 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1462 | 01:37 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/742 | 01:37 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/286 | 01:37 |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds | 01:41 | |
| aurojit (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Remote host closed the connection | 01:46 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy. | 01:47 | |
| etrepum (~bob@accessnat4.mochimedia.net) left irc: Ping timeout: 240 seconds | 01:49 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Ping timeout: 244 seconds | 01:51 | |
| gutworth (~benjamin@dhcp-128-36-84-91.central.yale.edu) joined #pypy. | 01:57 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/286 | 01:58 |
| mtigas (~mtigas@users.spokesman.com) left irc: Quit: mtigas | 01:59 | |
| stakkars | arigo: I checked the buildbot, but did not find your changes checked in. | 01:59 |
| brads (~brads@S0106602ad06e67e1.ed.shawcable.net) joined #pypy. | 02:02 | |
| etrepum (~bob@75-101-96-144.dsl.static.sonic.net) joined #pypy. | 02:16 | |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 02:27 | |
| linq (~ident@24-246-25-39.cable.teksavvy.com) joined #pypy. | 02:29 | |
| stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) left irc: Quit: schnarch | 02:29 | |
| squiddy (~squiddy@f053080138.adsl.alicedsl.de) left irc: Quit: Leaving | 02:31 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy. | 02:32 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Remote host closed the connection | 02:35 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy. | 02:36 | |
| harrison (~quassel@adsl-69-209-220-254.dsl.chcgil.ameritech.net) left irc: Ping timeout: 240 seconds | 02:36 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Ping timeout: 240 seconds | 02:40 | |
| Shanita (~John@osbk-4db1538a.pool.mediaWays.net) joined #pypy. | 02:40 | |
| Kaskuka (~John@osbk-4db16ffd.pool.mediaWays.net) left irc: Ping timeout: 240 seconds | 02:41 | |
| brads_ (~brads@S0106602ad06e67e1.ed.shawcable.net) joined #pypy. | 02:49 | |
| brads (~brads@S0106602ad06e67e1.ed.shawcable.net) left irc: Read error: Connection reset by peer | 02:49 | |
| Nick change: brads_ -> brads | 02:49 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/968 | 02:51 |
| zain (~textual@c-67-160-201-63.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep. | 02:56 | |
| overminder (~overminde@wf119-103.ust.hk) left irc: Quit: Leaving | 03:01 | |
| zain (~textual@adsl-99-27-43-230.dsl.pltn13.sbcglobal.net) joined #pypy. | 03:01 | |
| thinker` (~thinker@203.69.99.17) joined #pypy. | 03:02 | |
| gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated! | 03:06 | |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 248 seconds | 03:11 | |
| zain (~textual@adsl-99-27-43-230.dsl.pltn13.sbcglobal.net) left irc: Ping timeout: 248 seconds | 03:14 | |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 03:14 | |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy | 03:18 | |
| rokujyouhitoma (~rokujyouh@173.227.61.4) left irc: Remote host closed the connection | 03:21 | |
| zain (~textual@adsl-99-27-43-230.dsl.pltn13.sbcglobal.net) joined #pypy. | 03:23 | |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 260 seconds | 03:29 | |
| tilgovi (~randall@couchdb/developer/tilgovi) left irc: Read error: Connection reset by peer | 03:38 | |
| zain (~textual@adsl-99-27-43-230.dsl.pltn13.sbcglobal.net) left irc: Quit: Computer has gone to sleep. | 03:45 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1881 | 03:51 |
| zain (~textual@74-95-7-92-SFBA.hfc.comcastbusiness.net) joined #pypy. | 03:59 | |
| linq (~ident@24-246-25-39.cable.teksavvy.com) left irc: Ping timeout: 240 seconds | 04:05 | |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy. | 04:11 | |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Client Quit | 04:13 | |
| apanda (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) joined #pypy. | 04:45 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1462 | 04:50 |
| zain (~textual@74-95-7-92-SFBA.hfc.comcastbusiness.net) left irc: Ping timeout: 252 seconds | 04:54 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/820 | 04:56 |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy. | 05:00 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/635 | 05:04 |
| pybot joined #pypy. | 05:15 | |
| -ChanServ (ChanServ@services.)- [#pypy] Welcome! | 05:15 | |
| ThomasWaldmann (~twaldmann@moinmoin/coreteam/thomas) joined #pypy. | 05:15 | |
| apollo13 (~apollo13@unaffiliated/apollo13) joined #pypy. | 05:15 | |
| mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) joined #pypy. | 05:15 | |
| mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) left irc: Excess Flood | 05:15 | |
| dD0T (~dD0T@unaffiliated/dd0t) joined #pypy. | 05:15 | |
| -ChanServ (ChanServ@services.)- [#stackless] Welcome to #stackless. This channel is for the discussion of matters relating to Stackless Python. Remember that others in this channel may not be in the same timezone as you, so stick around for an answer (even if only in your channel history) if you ask a question. The mailing list is also recommended as being a faster place to get help (http://www.stackless.com/mailman/listinfo/stackless). | 05:16 | |
| Eruquen (Eruquen@server3.raumopol.de) joined #pypy. | 05:16 | |
| xorAxAx (~alexander@moinmoin/coreteam/alexander) joined #pypy. | 05:17 | |
| mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) joined #pypy. | 05:17 | |
| mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) left irc: Excess Flood | 05:17 | |
| alexsuraci (u1962@gateway/web/irccloud.com/x-fmgkunpvlhywhrbp) joined #pypy. | 05:18 | |
| mitsuhiko (~mitsuhiko@hammett.srv.pocoo.org) joined #pypy. | 05:18 | |
| mitsuhiko (~mitsuhiko@hammett.srv.pocoo.org) left irc: Changing host | 05:18 | |
| mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) joined #pypy. | 05:18 | |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-jhyiwpemvoorwycm) joined #pypy. | 05:18 | |
| #pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services. | 05:18 | |
| mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) left irc: Excess Flood | 05:19 | |
| mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) joined #pypy. | 05:20 | |
| mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) left irc: Excess Flood | 05:20 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/618 | 05:23 |
| mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) joined #pypy. | 05:23 | |
| apollo13 (~apollo13@unaffiliated/apollo13) left irc: Ping timeout: 268 seconds | 05:23 | |
| apollo13 (~apollo13@unaffiliated/apollo13) joined #pypy. | 05:25 | |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-jhyiwpemvoorwycm) left irc: Excess Flood | 05:26 | |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-rgtzudjklhivhojf) joined #pypy. | 05:26 | |
| #pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services. | 05:26 | |
| stargaming (stargaming@server3.raumopol.de) joined #pypy. | 05:33 | |
| setmeaway (~setmeaway@119.201.52.190) left irc: Quit: Leaving | 05:45 | |
| setmeaway (~setmeaway@119.201.52.190) joined #pypy. | 05:46 | |
| ericflo (~ericflo@c-50-131-58-97.hsd1.ca.comcast.net) joined #pypy. | 05:48 | |
| oal (u4126@gateway/web/irccloud.com/x-nmdnucfmlpdbesxp) joined #pypy. | 05:51 | |
| asmeurer_____ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy. | 05:55 | |
| asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Ping timeout: 240 seconds | 05:56 | |
| Nick change: asmeurer_____ -> asmeurer | 05:56 | |
| Ademan (~dan@adsl-71-141-224-79.dsl.snfc21.pacbell.net) joined #pypy. | 06:03 | |
| thinker` (~thinker@203.69.99.17) left irc: Remote host closed the connection | 06:12 | |
| Nick change: Gulopine -> Gulaway | 06:16 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/742 | 06:22 |
| aleksi (~aleksi@85.235.191.82) joined #pypy. | 06:24 | |
| RiaanFV (~riaanfv@41-133-33-231.dsl.mweb.co.za) joined #pypy. | 06:25 | |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Read error: Connection reset by peer | 06:28 | |
| ericflo (~ericflo@c-50-131-58-97.hsd1.ca.comcast.net) left irc: Quit: ericflo | 06:39 | |
| RiaanFV (~riaanfv@41-133-33-231.dsl.mweb.co.za) left irc: Quit: Leaving | 06:42 | |
| brads (~brads@S0106602ad06e67e1.ed.shawcable.net) left irc: Ping timeout: 268 seconds | 06:43 | |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy. | 06:45 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/298 | 06:57 |
| fijal (~fijal@41.52.94.74) joined #pypy. | 07:14 | |
| fijal | why is gmail black today? | 07:20 |
| Alex_Gaynor | fijal: it's normal for me | 07:21 |
| fijal | the gmail logo is black | 07:23 |
| rokujyouhitoma (~rokujyouh@12.51.221.130) joined #pypy. | 07:23 | |
| fijal | good | 07:26 |
| fijal | now only thing that's left is to turn what I have into an actual solution | 07:26 |
| fijal | and use matplotlib from pypy | 07:26 |
| fijal | done | 07:26 |
| brutal_chaos | fijal: yay | 07:26 |
| brutal_chaos | you're around | 07:26 |
| fijal | brutal_chaos: yes, why? | 07:26 |
| brutal_chaos | what should I send to the dev ml? | 07:26 |
| fijal | about what? | 07:27 |
| brutal_chaos | (benchmarks) | 07:27 |
| fijal | sorry, I'm having my first coffee, my IQ is halfed | 07:27 |
| fijal | halved even | 07:27 |
| fijal | "hello, those are the benchmarks on clang and shadowstack, discuss"? | 07:27 |
| fijal | machine spec please as well | 07:27 |
| fijal | and compiler versions | 07:27 |
| brutal_chaos | fijal: kk, do I attach the results.json? | 07:28 |
| fijal | can't you just paste the stdout? | 07:28 |
| brutal_chaos | yes, sure. | 07:28 |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 07:28 | |
| brutal_chaos | wasn't sure which was necessary. on it. | 07:28 |
| fijal | stdout is essentially a human-readable version of json | 07:29 |
| fijal | I don't think json has more info innit | 07:29 |
| fijal | Alex_Gaynor: it's kinda neat hack, I'm proud ;-) | 07:29 |
| asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Quit: asmeurer | 07:34 | |
| K-os (~gabriel@port-18238.pppoe.wtnet.de) joined #pypy. | 07:35 | |
| brutal_chaos | Report on Linux infinity 3.1.1-gentoo #1 SMP PREEMPT Sun Nov 20 03:57:03 PST 2011 x86_64 AMD Turion(tm) II Dual-Core Mobile M520 | 07:35 |
| brutal_chaos | Total CPU cores: 2 | 07:35 |
| Alex_Gaynor | fijal: it's quite nice :) | 07:35 |
| brutal_chaos | good enough info? | 07:35 |
| brutal_chaos | fijal? ^ | 07:36 |
| asmeurer_ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy. | 07:37 | |
| hongminhee (u295@gateway/web/irccloud.com/x-juuehcrfxwjrnmtl) joined #pypy. | 07:38 | |
| fijal | brutal_chaos: yop | 07:39 |
| fijal | brutal_chaos: personally I would like to know your cache size | 07:39 |
| fijal | and what pypy thinks about it :) | 07:39 |
| fijal | cat /proc/cpuinfo | 07:40 |
| fijal | cat /proc/cpuinfo | grep cache | 07:40 |
| asmeurer_ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Quit: asmeurer_ | 07:42 | |
| brutal_chaos | fijal: sent. :) | 07:47 |
| fijal | cool | 07:47 |
| fijal | did not receive yet | 07:47 |
| fijal | brutal_chaos: so what are you comparing against? | 07:52 |
| fijal | vs asmgcc one? | 07:53 |
| brutal_chaos | nothing yet I guess. Going to run the tests against your prebuilt one tonight. | 07:53 |
| fijal | no no | 07:53 |
| fijal | how did you run it? | 07:53 |
| fijal | it always compares something with something else | 07:53 |
| fijal | what was the command line? | 07:54 |
| brutal_chaos | not sure, one sec | 07:54 |
| brutal_chaos | pypy ./runner.py | 07:55 |
| brutal_chaos | that's all | 07:55 |
| fijal | ok well | 07:55 |
| fijal | I think you compared pypy to pypy | 07:55 |
| fijal | a bit useless | 07:55 |
| brutal_chaos | lol | 07:56 |
| brutal_chaos | fun | 07:56 |
| brutal_chaos | fijal: how was supposed to run the benchmarks? I'll rerun them tonight | 07:57 |
| fijal | ./runner.py --baseline=one -p twop | 07:58 |
| fijal | where one and two are your things to compare | 07:58 |
| fijal | --baseline defaults to whatever interpreter you use to run runner | 07:58 |
| brutal_chaos | pypy ./runner.py --baseline=/usr/bin/pypy -p /usr/src/python | 07:59 |
| brutal_chaos | ? | 07:59 |
| fijal | for example | 08:00 |
| brutal_chaos | kk, I'll run that tonight | 08:00 |
| fijal | but a comparison with cpython is not as exciting as pypy with shadowstack vs pypy with asmgcc | 08:00 |
| brutal_chaos | ok, compare it to your prebuilt one? | 08:01 |
| brutal_chaos | fijal: your prebuilt Linux 64bit one | 08:01 |
| fijal | sounds good | 08:03 |
| brutal_chaos | fijal: btw, has that email showed up yet? | 08:04 |
| fijal | yes | 08:04 |
| brutal_chaos | shown* | 08:04 |
| brutal_chaos | cool | 08:04 |
| fijal | Alex_Gaynor: hey | 08:14 |
| fijal | Alex_Gaynor: imagine that you do | 08:14 |
| fijal | pip install scipy_pypy | 08:14 |
| fijal | and you get all the scipy routines just like that | 08:15 |
| fijal | via this bridge | 08:15 |
| fijal | ain't it awesome? | 08:15 |
| Alex_Gaynor | pretty cool | 08:15 |
| fijal | that's quite some work | 08:15 |
| Alex_Gaynor | you need to disable cpyext with your pypy though right? | 08:15 |
| fijal | but I want to write a blog post | 08:15 |
| fijal | yes | 08:15 |
| Alex_Gaynor | embed CPython is a much cooler solution anyways :) | 08:15 |
| Alex_Gaynor | fijal: which first, this one or benchmarks? | 08:16 |
| Action: Alex_Gaynor is writing another paper | 08:16 | |
| fijal | I dunno | 08:16 |
| fijal | benchmarks require quite a few things that are hard ;-) | 08:16 |
| Alex_Gaynor | pypy is good at solving hard problems | 08:17 |
| Alex_Gaynor | or transforming them into trivial problems | 08:17 |
| fijal | yeah well | 08:17 |
| fijal | I dunno, I make my presentation today | 08:17 |
| Alex_Gaynor | Want to provide runtime feedback into the JIT on a per-function basis deciding if loops should be unrolled? Add a decorator! | 08:17 |
| fijal | so probably none of them today | 08:17 |
| Alex_Gaynor | presentation to a PUG? | 08:17 |
| Alex_Gaynor | Show them the benchmarks and this stuff, it's seriously cool. | 08:18 |
| rokujyouhitoma (~rokujyouh@12.51.221.130) left irc: Remote host closed the connection | 08:21 | |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy | 08:23 | |
| stargaming (stargaming@server3.raumopol.de) left irc: Changing host | 08:27 | |
| stargaming (stargaming@unaffiliated/stargaming) joined #pypy. | 08:27 | |
| fijal | yes | 08:28 |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy. | 08:35 | |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Client Quit | 08:35 | |
| sunoano (~sa@unaffiliated/sunoano) left irc: Quit: Leaving. | 08:37 | |
| arigato (~arigo@82.113.99.213) joined #pypy. | 08:39 | |
| `fox` (~fox@host91-171-dynamic.245-95-r.retail.telecomitalia.it) joined #pypy. | 08:44 | |
| Nick change: Guest86987 -> Fluxid | 08:47 | |
| cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) joined #pypy. | 08:47 | |
| antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy. | 08:48 | |
| Alex_Gaynor | cfbolz: do you think consistency with prolog is an important property for Python code? | 08:50 |
| cfbolz | for me, yes | 08:50 |
| cfbolz | :-) | 08:50 |
| Alex_Gaynor | hehe, I rather liked prolog, I just can't imagine writing actual programs in it | 08:51 |
| cfbolz | Alex_Gaynor: that feeling goes away after you've written your first medium-sized program in it | 08:51 |
| Alex_Gaynor | yes, I only wrote small programs like maze solvers and parsers | 08:52 |
| mikefc | fijal: scipypy?? ooooo! :) | 08:52 |
| Alex_Gaynor | mikefc: did you see his matplotlib example? | 08:52 |
| mikefc | nope. where at? | 08:52 |
| Alex_Gaynor | mikefc: http://baroquesoftware.com/~fijal/extreme_hacks.png | 08:52 |
| mikefc | that's pretty fucking awesome | 08:53 |
| mikefc | dare i ask: speed? | 08:53 |
| mikefc | is scipypy using numpypy for array stuff there? | 08:54 |
| Action: mikefc gets excited | 08:54 | |
| Alex_Gaynor | basically he's embedding CPYthon inside of pypy | 08:54 |
| Alex_Gaynor | using ctypes | 08:54 |
| Alex_Gaynor | and then passing the array address to CPython, where it's loaded up and displayed | 08:54 |
| Alex_Gaynor | so you're sharing the raw memory between the two of them | 08:54 |
| mikefc | oh. nice. a handy bridge. | 08:54 |
| Action: mikefc wants native scipypy and numpypy :) | 08:54 | |
| Alex_Gaynor | yeah, works nicely with anything that just reads the arryay contents | 08:54 |
| mikefc | native = pypy | 08:55 |
| K-os (~gabriel@port-18238.pppoe.wtnet.de) left irc: Ping timeout: 248 seconds | 08:55 | |
| arigato | Alex_Gaynor: it should work also if the array content is modified, no? | 08:55 |
| Alex_Gaynor | in theory you can write all the slow parts of your code in pypy, and then just do the display (which isn't a time sink) in the bridge | 08:55 |
| Alex_Gaynor | arigato: well, as long as you don't need to change the shape or anything like this | 08:55 |
| arigato | yes | 08:55 |
| Alex_Gaynor | you can change elements, but nothing else | 08:55 |
| arigato | yes | 08:55 |
| arigato | but you can also build new arrays and pass them in the other direction, too | 08:55 |
| Alex_Gaynor | true | 08:56 |
| kenaan | 12arigo default 11104e7a656174 15/pypy/module/clr/boxing_rules.py: Comment out this hack. I guess it was used for "checkmodule", which likely doesn't work any more with the clr module. | 09:00 |
| squiddy (~squiddy@g224195234.adsl.alicedsl.de) joined #pypy. | 09:02 | |
| amaury (amaury_@nat/google/x-aroxchifqyoidfxw) joined #pypy. | 09:02 | |
| G2P (~G2P@fw-asn1.ornis.com) joined #pypy. | 09:11 | |
| mikefc | woo! | 09:12 |
| mikefc | Fatal RPython error: AssertionError | 09:12 |
| mikefc | Abort trap | 09:12 |
| Action: mikefc goes "bug case" hunting | 09:13 | |
| arigato | mikefc: in which pypy? | 09:14 |
| mikefc | nightly from <24hrs ago | 09:14 |
| arigato | feel free to paste more information :-) | 09:15 |
| mikefc | I'll narrow down my test case first | 09:15 |
| aurojit (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) joined #pypy. | 09:19 | |
| apanda (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) left irc: Ping timeout: 244 seconds | 09:21 | |
| arigato (~arigo@82.113.99.213) left irc: Quit: See you | 09:23 | |
| thinker` (~thinker@114-24-13-172.dynamic.hinet.net) joined #pypy. | 09:29 | |
| lucian (~lucian@78-86-111-127.zone2.bethere.co.uk) joined #pypy. | 09:36 | |
| voidspace (~voidspace@python/psf/voidspace) joined #pypy. | 09:46 | |
| ojii (~ojii@40-34.60-188.cust.bluewin.ch) joined #pypy. | 09:50 | |
| mikefc | RPython traceback: | 09:50 |
| mikefc | File "jit_metainterp_optimize.c", line 159, in optimize_loop | 09:50 |
| mikefc | File "jit_metainterp_optimize.c", line 510, in _optimize_loop | 09:50 |
| mikefc | File "jit_metainterp_optimizeopt_unroll.c", line 13528, in UnrollOptimizer_propagate_all_forward | 09:50 |
| mikefc | File "jit_metainterp_optimizeopt_optimizer.c", line 3962, in Optimizer_propagate_all_forward | 09:50 |
| mikefc | File "jit_metainterp_optimizeopt_pure.c", line 2459, in OptPure_optimize_default | 09:50 |
| mikefc | File "jit_metainterp_optimizeopt_fficall.c", line 1740, in OptFfiCall_optimize_CALL | 09:50 |
| mikefc | File "jit_metainterp_optimizeopt_fficall.c", line 7416, in OptFfiCall_do_getsetarrayitem | 09:50 |
| mikefc | File "jit_metainterp_optimizeopt_fficall.c", line 8000, in OptFfiCall__get_interior_descr | 09:50 |
| mikefc | Fatal RPython error: AssertionError | 09:50 |
| mikefc | Abort trap | 09:50 |
| mikefc | sorry. bit too much :( | 09:51 |
| mikefc | i'll make it a bug report. sorry | 09:53 |
| Alex_Gaynor | do you have a script to reproduce this? | 09:53 |
| mikefc | data = numpy.zeros([100,100], dtype='uint8') | 09:53 |
| mikefc | for row in xrange(100): | 09:53 |
| mikefc | " ".join([str(x) for x in data[row]]) | 09:53 |
| Alex_Gaynor | I see which assert it is, but I need the value that triggers it | 09:53 |
| Alex_Gaynor | what th ehell... how does that cause it | 09:53 |
| Alex_Gaynor | oh, I guess I see it | 09:53 |
| Alex_Gaynor | it's probably simpler than that | 09:53 |
| K-os (~gabriel@dhcp-205.zbh.uni-hamburg.de) joined #pypy. | 09:54 | |
| mikefc | pypy nightly: Python 2.7.1 (4162bc8b5f4c, Dec 05 2011, 05:42:01) | 09:54 |
| Alex_Gaynor | ok, I see it :/ | 09:55 |
| Alex_Gaynor | no need to file a bug | 09:55 |
| mikefc | it's a feature? :) | 09:56 |
| mikefc | "Would it be possible to have an interpreter that could support both Python 2 and Python 3 modules in the same runtime?" ... and a pony? | 09:58 |
| arigato (~arigo@fwstups.cs.uni-duesseldorf.de) joined #pypy. | 10:01 | |
| JaRoel|4d | :p | 10:02 |
| K-os (~gabriel@dhcp-205.zbh.uni-hamburg.de) left irc: Quit: Leaving | 10:04 | |
| setmeaway2 (oosool3@119.201.52.190) joined #pypy. | 10:05 | |
| setmeaway (~setmeaway@119.201.52.190) left irc: Read error: Connection reset by peer | 10:05 | |
| arigato | cfbolz: floatstrategy is responsible for nbody_modified getting slower (~10%) | 10:07 |
| Alex_Gaynor | arigato: if you have a moment, could you take a look at the XXX here http://paste.pocoo.org/show/517922/ | 10:09 |
| K-os (~gabriel@dhcp-205.zbh.uni-hamburg.de) joined #pypy. | 10:09 | |
| arigato | Alex_Gaynor: that looks wrong | 10:12 |
| Alex_Gaynor | :) | 10:13 |
| arigato | if kind == 'u', then is_signed=False (and the other is_xxx too) | 10:13 |
| Alex_Gaynor | ah indeed, the tests passes with them all False | 10:14 |
| arigato | you need to change your test | 10:14 |
| Alex_Gaynor | how so? | 10:14 |
| arigato | store a value that is between 128 and 255 | 10:14 |
| Alex_Gaynor | ah, makes sense | 10:14 |
| arigato | also, if only data[0] is non-null, then the jit will never see it | 10:15 |
| arigato | ah, sorry | 10:16 |
| arigato | data has only 1 item, the loop is not over the index | 10:16 |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection | 10:17 | |
| Alex_Gaynor | arigato: great thanks | 10:18 |
| fijal | arigato: it's more complex than that | 10:18 |
| fijal | float strategy changed some unrolling rules | 10:18 |
| fijal | uh maybe not | 10:19 |
| Alex_Gaynor | fijal: as i said, it broke UNPACK_SEQUENCE optmization wtih lists | 10:19 |
| kenaan | 12alex_gaynor default 11617e1464542c 15/pypy/jit/metainterp/: make the dynamic interiorfield optimization work correctly with unsigned values | 10:19 |
| kenaan | 12alex_gaynor default 1126bac7f772ca 15/pypy/module/clr/boxing_rules.py: merged upstream | 10:19 |
| Alex_Gaynor | mikefc: ok fixed on trunk | 10:19 |
| fijal | Alex_Gaynor: I think carl claimed it was something else | 10:19 |
| arigato | fijal: I have two consecutive versions of pypy, so if needed, I can really look at the JIT code produced by each | 10:19 |
| Alex_Gaynor | fijal: no, I'm positive it's unpack sequence | 10:19 |
| fijal | arigato: ok, but IMO it's unpack sequence that makes it slow | 10:20 |
| fijal | but anyway, look at traces, they're nonsense | 10:20 |
| arigato | unpacksequence? | 10:20 |
| Alex_Gaynor | arigato: UNPACK_SEQUENCE the opcoe | 10:20 |
| Alex_Gaynor | opcode* | 10:20 |
| fijal | like | 10:21 |
| fijal | a, b, c = l | 10:21 |
| Alex_Gaynor | fijal: look at space.fixedview for W_ListObject | 10:21 |
| fijal | where l is a list of floats | 10:21 |
| Action: fijal has no time | 10:21 | |
| fijal | going to physio then walking | 10:21 |
| fijal | will do some numpy in the evening | 10:21 |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 10:23 | |
| ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) joined #pypy. | 10:23 | |
| voidspace (~voidspace@python/psf/voidspace) left irc: Quit: Uhm... gotta go | 10:23 | |
| Action: arigato tries to use jitviewer again | 10:23 | |
| fijal | arigato: it works great for me *most of the time* | 10:24 |
| fijal | times where it does not work happen though | 10:24 |
| arigato | it failed to work whenever I tried, usually for some easy-to-fix reason | 10:24 |
| fijal | rarely enough for me not to care | 10:24 |
| fijal | half of the blame goes on python code objects being sloppy at getting to source | 10:24 |
| fijal | the other part about jitviewer not being able to run without source | 10:24 |
| fijal | should be better now | 10:24 |
| fijal | but well | 10:25 |
| fijal | I'm sorry, next incarnation will be better whenver I'll get to it :) | 10:25 |
| arigato | :-) | 10:26 |
| fijal | arigato: I want my tests to run faster under pypy :/ | 10:26 |
| arigato | [yay, it works in this case :-)] | 10:26 |
| fijal | that's a bit cheating, because I used it on nbody_modified before ;-) | 10:26 |
| fijal | arigato: the slow test runs are the last thing that make me not move | 10:27 |
| witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy. | 10:28 | |
| thinker` (~thinker@114-24-13-172.dynamic.hinet.net) left irc: Ping timeout: 252 seconds | 10:28 | |
| witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) left irc: Client Quit | 10:28 | |
| witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy. | 10:28 | |
| oal | Hmm, why isn't Pypy in the debian benchmark game? | 10:28 |
| witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy. | 10:29 | |
| lucian | oal: because they don't want it there, and it's not very relevant | 10:29 |
| arigato | uh, expanding the list of floats to W_Objects is done by two residual calls? | 10:29 |
| arigato | inefficient | 10:29 |
| oal | lucian: ok :) | 10:29 |
| Alex_Gaynor | arigato: with much allocating and copying | 10:29 |
| arigato | yes | 10:30 |
| Alex_Gaynor | there should be a getitems_unroll or so, that fixedview can all | 10:30 |
| lucian | oal: it's literally a benchmark game, and irrelevant for most workloads | 10:30 |
| brutal_chaos | fijal: hmm, ugh. The ssl you build against is old... | 10:32 |
| voidspace (~anonymous@python/psf/voidspace) joined #pypy. | 10:35 | |
| brutal_chaos | how can I test pypy to make sure it'll run properly? | 10:35 |
| kkris (~kris@93-82-40-215.adsl.highway.telekom.at) joined #pypy. | 10:37 | |
| cfbolz | Alex_Gaynor: we'll ask lukas to look into it | 10:40 |
| Alex_Gaynor | ok. thanks | 10:41 |
| kenaan | 12arigo default 1115d89540a5c3 15/pypy/: Add a failing test for fixedview() on a list of floats that expands to too much stuff for now. | 10:44 |
| squiddy (~squiddy@g224195234.adsl.alicedsl.de) left irc: Quit: Leaving | 10:44 | |
| cfbolz | arigato: you checked in some random stuff | 10:50 |
| arigato | oups | 10:50 |
| thinker` (~thinker@111-248-41-131.dynamic.hinet.net) joined #pypy. | 10:50 | |
| hruske (~Gasper@188-230-156-183.dynamic.t-2.net) joined #pypy. | 10:51 | |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 10:52 | |
| kenaan | 12arigo default 113ff9c88dca39 15/pypy/jit/codewriter/codewriter.py: Revert this, committed by accident. (The other accidental commit is actually good and a complete detail, so I'm no... | 10:55 |
| thinker` (~thinker@111-248-41-131.dynamic.hinet.net) left irc: Read error: Connection reset by peer | 11:09 | |
| Arach (~arach@95-30-210-133.broadband.corbina.ru) left irc: Ping timeout: 244 seconds | 11:15 | |
| thinker` (~thinker@111-248-41-131.dynamic.hinet.net) joined #pypy. | 11:20 | |
| lmoura (~lmoura@186.215.206.130) left irc: Ping timeout: 240 seconds | 11:25 | |
| lmoura (~lmoura@186.212.130.54) joined #pypy. | 11:25 | |
| Arach (~arach@2.95.139.234) joined #pypy. | 11:28 | |
| antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds | 11:44 | |
| stakkars_ (~tismer@i59F744D8.versanet.de) joined #pypy. | 11:48 | |
| stakkars_ | whow! I did the first bootstrap. | 11:50 |
| lucian (~lucian@78-86-111-127.zone2.bethere.co.uk) left irc: Ping timeout: 240 seconds | 11:51 | |
| bfirsh (u1308@gateway/web/irccloud.com/x-vertixxxctpxdezm) joined #pypy. | 11:52 | |
| stakkars_ | python2.6 built pypy-win32 in 16500 s | 11:53 |
| stakkars_ | pypy-win32 builds itself in 7750 s | 11:55 |
| thinker` (~thinker@111-248-41-131.dynamic.hinet.net) left irc: Ping timeout: 244 seconds | 11:55 | |
| stakkars_ | the memoty requirement was interesting: | 11:55 |
| stakkars_ | python 2.6 needed 1.9 Gb | 11:56 |
| stakkars_ | pypy-win32 needed 2.13 Gb | 11:57 |
| stakkars_ | so my first attempt crashed even in python.exe with memory error, and I had to enable it for more memory | 11:58 |
| stakkars_ | using editbin. We should add this step for all win32 builds. | 12:01 |
| stakkars_ | interesting was the step "compile_c". This was even slower in PyPy! | 12:02 |
| xiaochen (~chobits@119.80.124.179) joined #pypy. | 12:03 | |
| stakkars_ | obviously because pypy is restarted for every source file to process it, and the Jit forgets all of its knowledge between the sessions. | 12:03 |
| stakkars_ | This is a thing to think about: How can we save Jit knowledge? | 12:04 |
| lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy. | 12:04 | |
| lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Remote host closed the connection | 12:04 | |
| stakkars_ | perhaps with shared memory, forking and a copy-on-write semantic? | 12:04 |
| lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy. | 12:04 | |
| lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Client Quit | 12:06 | |
| lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy. | 12:06 | |
| stakkars_ | all in all a wonderful experience. I will now generalize my Win64 branch to let it run on top of a 32bit interpreter, like the fast pypy-c. That is a cross-bootstrap. | 12:06 |
| stakkars_ | compile_c: python2.6 3000 s | 12:12 |
| stakkars_ | compile_c: pypy-32 3125 s | 12:13 |
| jacob22 | stakkars_: Anto did experiments with dumping the state of the executable at the sprint. That way you can conserve JIT state. | 12:17 |
| stakkars_ | jacob22: ah, I thought he had no real success. | 12:18 |
| stakkars_ | I think that could be an important feature, for instance for running the test suites, which partially are starting python vry often. | 12:19 |
| jacob22 | stakkars_: His problem was that he came back exactly to the dumped state and didn't have any control over the resumption. | 12:19 |
| stakkars_ | ok, that is what I expected. One needs to write something which gives a restart point, at interpreter toplevel. | 12:20 |
| jacob22 | stakkars_: But if you pick up the name of files to translate through an environment variable or some other external mechanism, you can steer the resumed program. | 12:20 |
| jacob22 | Now, I don't know if the warmup time for the translation toolchain warrants such hacks, or if the time saved will be minimal. | 12:22 |
| stakkars_ | I would not even write a dump to a file, but use shared memory, fork the process and let it forget everything but what it learned. | 12:23 |
| stakkars_ | likewise, resetting pypy into an initial state, based on the same virtual address space | 12:25 |
| stakkars_ | (but I fear this exactly will be hard to impossible on Windoze) | 12:25 |
| brutal_chaos | fijal: I cannot compare against your prebuilt pypy | 12:27 |
| brutal_chaos | you compile against an old libssl and libcrypto. | 12:28 |
| brutal_chaos | 0.9.8 vs 1.0.0 | 12:28 |
| stakkars_ | well, I need to think more about this. maybe we can dump a bit more abstractly. | 12:29 |
| nedbat (~nedbat@python/psf/nedbat) left irc: Disconnected by services | 12:31 | |
| nedbat_ (~nedbat@python/psf/nedbat) joined #pypy. | 12:31 | |
| brutal_chaos | fijal: gentoo aparently has 0.9.8 slotted (i can have both versions installed at the same time). I am building it now and will hope the tests run. | 12:31 |
| squiddy (~squiddy@w4322.dip.tu-dresden.de) joined #pypy. | 12:34 | |
| stakkars_ | jacob22: Anto's problem is actually quite similar to what I needed to do for stackless tasklet pickling. That needs a controlled way, by calling a special function. | 12:42 |
| aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy. | 12:42 | |
| mikefc | Alex_Gaynor: thanks for the fix | 12:45 |
| stakkars_ | a last comment: The memory usage was only 10% difference between python and pypy. I expected much more. | 12:47 |
| stakkars_ | IOW, PyPy is quite conservative at the moment, and I think it could preserve things a bit longer, or whatever gets faster by using nore memory. | 12:48 |
| stakkars_ | (this is the warm-up for a blog post) | 12:49 |
| arigato | $ py.test --no-assert | 12:55 |
| arigato | ERROR: assertion options conflict | 12:55 |
| arigato | thank-you-very-much | 12:55 |
| nedbat_ (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds | 12:58 | |
| sunoano (~sa@mail.heartinternet.co.uk) joined #pypy. | 13:00 | |
| sunoano (~sa@mail.heartinternet.co.uk) left irc: Changing host | 13:00 | |
| sunoano (~sa@unaffiliated/sunoano) joined #pypy. | 13:00 | |
| squiddy (~squiddy@w4322.dip.tu-dresden.de) left irc: Quit: leaving | 13:02 | |
| stakkars_ (~tismer@i59F744D8.versanet.de) left irc: Quit: schnarch | 13:09 | |
| jnoller (~jnoller@vpn.nasuni.com) joined #pypy. | 13:10 | |
| harrison (~quassel@adsl-69-209-212-18.dsl.chcgil.ameritech.net) joined #pypy. | 13:11 | |
| mikefc | When are the new nightly releases compiled? i.e. what time GMT? | 13:13 |
| arigato | around 2AM | 13:13 |
| stakkars_ (~tismer@i59F744D8.versanet.de) joined #pypy. | 13:14 | |
| kenaan | 12arigo SpecialisedTuples 11363bac62f6dc 15/pypy/objspace/std/: Rework the source code. Kill a few features that don't really make sense to have; write by hand a fast d... | 13:14 |
| kenaan | 12arigo SpecialisedTuples 1138ecd0cebaa2 15/pypy/objspace/std/specialisedtupleobject.py: translation fix | 13:14 |
| mikefc | arigato: thanks. i've reached an impasse and need to wait for Alex_Gaynor's fix in trunk. | 13:14 |
| Action: mikefc checks watch and does conversion | 13:15 | |
| mikefc | noon tomorrow. | 13:15 |
| arigato | if you want a new release, you can force one right now, too | 13:15 |
| arigato | then you'll get it in 45 minutes | 13:15 |
| mikefc | really? how? | 13:15 |
| arigato | http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32 | 13:16 |
| arigato | (example for linux 32) | 13:16 |
| mikefc | (not that i need it just at the moment. it's nearly midnight, and I don't want to stay up too too late) | 13:16 |
| arigato | give your name in the form, and press "Force build" | 13:16 |
| arigato | :-) | 13:16 |
| mikefc | ah. nice. | 13:16 |
| mikefc | it's not protected from denial of service ? anyone on the net can request a build? | 13:18 |
| stakkars_ | arigato: on win32, this takes 7750 s! How comes? | 13:18 |
| arigato | yes, I agree it sounds strange, but we didn't have any unexpected build at all, so far | 13:18 |
| arigato | stakkars_: I can't really know | 13:19 |
| stakkars_ | I used "pypy-c translate.py -Ojit" | 13:19 |
| arigato | yes | 13:19 |
| stakkars_ | which created all that it could. Do you do this as well? | 13:20 |
| arigato | and also "python translate.py -Ojit" takes ~1h30 on linux | 13:20 |
| arigato | yes | 13:20 |
| arigato | less than 2h for sure | 13:20 |
| arigato | python2.6 | 13:20 |
| arigato | (and python2.7 is much faster than python2.6, according to fijal) | 13:21 |
| stakkars_ | python 2.6 took over 4 hours | 13:21 |
| stakkars_ | pypy took over 2 hours. Should virtualbox be so slow? Will see. | 13:22 |
| arigato | "pypy translate.py -O2 --no-allworkingmodules" takes 20 minutes on my laptop | 13:22 |
| kenaan | 12arigo SpecialisedTuples 111f3ae2d48981 15/pypy/config/pypyoption.py: Add 'withspecialisedtuple' by default, at least for testing the result. | 13:24 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1147 [12arigo, SpecialisedTuples] | 13:24 |
| bbot2 | 13Exception: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1147 [12arigo, SpecialisedTuples] | 13:27 |
| arigato | aaaAaah bah | 13:31 |
| arigato | 2.7/sre_compile.py | 13:31 |
| arigato | is using a ton of ' if x is "mark":" | 13:32 |
| arigato | ah, that has been fixed on trunk | 13:34 |
| Action: arigato merges it into SpecialisedTuples | 13:34 | |
| kenaan | 12arigo SpecialisedTuples 114f1c5b700152 15/: hg merge default | 13:38 |
| kenaan | 12arigo SpecialisedTuples 1124d1d9d80c96 15/pypy/objspace/std/tupletype.py: clean-up. | 13:39 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1148 [12arigo, SpecialisedTuples] | 13:45 |
| stakkars_ | maybe I need a true windows machine to check. This also helps to check certain time related bugs (bad on a virtual) | 13:46 |
| mattip (c1a94669@gateway/web/freenode/ip.193.169.70.105) joined #pypy. | 13:48 | |
| mattip (c1a94669@gateway/web/freenode/ip.193.169.70.105) left #pypy. | 13:53 | |
| Nick change: Gulaway -> Gulopine | 13:57 | |
| gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) joined #pypy. | 13:58 | |
| bivab (~david@ip-62-143-59-5.unitymediagroup.de) joined #pypy. | 13:59 | |
| kenaan | 12bivab arm-backend-2 11e70ac50eb3a3 15/pypy/jit/backend/arm/: refactor FINISH to directly store all values to the failboxes. | 14:00 |
| kenaan | 12bivab arm-backend-2 1110bf505d8a69 15/pypy/jit/backend/arm/opassembler.py: translation fix, missing case | 14:00 |
| kenaan | 12bivab arm-backend-2 119512dcdd4967 15/pypy/jit/backend/arm/opassembler.py: Reverse this BKPT back to a NOP here, because this is not always patched | 14:00 |
| stakkars_ (~tismer@i59F744D8.versanet.de) left irc: Quit: schnarch | 14:09 | |
| sven_hager (~sven@dslb-092-072-028-071.pools.arcor-ip.net) joined #pypy. | 14:09 | |
| whitelynx|work (~whitelynx@63.241.75.144) joined #pypy. | 14:16 | |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 14:23 | |
| kenaan | 12hager ppc-jit-backend 11103beedd8034 15/pypy/jit/backend/ppc/ppcgen/: begin reimplementation of CALL on PPC64 | 14:39 |
| stakkars_ (~stakkars@82.113.121.173) joined #pypy. | 14:41 | |
| Nick change: Gulopine -> Gulaway | 14:42 | |
| stakkars_ | arigato: -O2 --noallworking | 14:43 |
| stakkars_ | 1993 s | 14:44 |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1148 [12arigo, SpecialisedTuples] | 14:48 |
| nedbat (~nedbat@python/psf/nedbat) left irc: Disconnected by services | 14:49 | |
| nedbat_ (~nedbat@python/psf/nedbat) joined #pypy. | 14:50 | |
| ousado_ (~ousado@31-18-121-45-dynip.superkabel.de) joined #pypy. | 14:53 | |
| ousado (~ousado@unaffiliated/ousado) left irc: Ping timeout: 240 seconds | 14:53 | |
| Nick change: ousado_ -> ousado | 14:53 | |
| ousado (~ousado@31-18-121-45-dynip.superkabel.de) left irc: Changing host | 14:53 | |
| ousado (~ousado@unaffiliated/ousado) joined #pypy. | 14:53 | |
| cpets (c358b60a@gateway/web/freenode/ip.195.88.182.10) joined #pypy. | 14:58 | |
| santagada (~leonardo@201.47.217.214) joined #pypy. | 15:02 | |
| stakkars_ (~stakkars@82.113.121.173) left irc: Quit: Colloquy for iPhone - http://colloquy.mobi | 15:04 | |
| Ademan (~dan@adsl-71-141-224-79.dsl.snfc21.pacbell.net) left irc: Quit: leaving | 15:07 | |
| kennethreitz (~kennethre@ip65-46-252-62.z252-46-65.customer.algx.net) joined #pypy. | 15:09 | |
| aleksi (~aleksi@85.235.191.82) left irc: Remote host closed the connection | 15:10 | |
| xiaochen (~chobits@119.80.124.179) left irc: Quit: leaving | 15:16 | |
| JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 15:16 | |
| roelbruggink (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 15:17 | |
| fprimex (~fprimex@brent-macbook.sc.fsu.edu) joined #pypy. | 15:17 | |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 240 seconds | 15:19 | |
| JaRoel (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 240 seconds | 15:20 | |
| Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy. | 15:29 | |
| fijal | stakkars: seems compile_c took tons of time | 15:31 |
| fijal | if it's multicore it can be sped up by quite a bit | 15:31 |
| rokujyouhitoma (~rokujyouh@12.51.221.130) joined #pypy. | 15:33 | |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy. | 15:34 | |
| fijal | fzzzy: hello | 15:43 |
| fzzzy | hey fijal | 15:44 |
| fijal | howzit? | 15:44 |
| fzzzy | good, what's new with you | 15:44 |
| fzzzy | why am i up so early though :( | 15:45 |
| fijal | don't ask me ;-) | 15:46 |
| fijal | fzzzy: what stops you guys from using pypy? | 15:47 |
| fzzzy | fijal: what do you mean? for what? | 15:52 |
| fijal | for mozilla stuff | 15:52 |
| fijal | I wonder if you did some experiments | 15:52 |
| fzzzy | there's plenty of python inside mozilla but not in my department | 15:53 |
| fijal | ah ok | 15:53 |
| fijal | so what do you work on, if it's not a secret? | 15:53 |
| fijal | some sort of JS? | 15:53 |
| fzzzy | javascript | 15:53 |
| fijal | I've seen your pavel | 15:53 |
| fijal | I assume this sort of stuff | 15:53 |
| fzzzy | yeah, i'm working on dom.js mostly | 15:53 |
| fijal | ah cool | 15:53 |
| fzzzy | also most of the dudes I work with are working on rust | 15:54 |
| fijal | that does improve the testing story quite a lot | 15:54 |
| fijal | is rust implemented mostly in rust? | 15:54 |
| fijal | or mostly in C++? | 15:54 |
| fzzzy | rust is implemented in rust | 15:55 |
| fzzzy | there's small amounts of c++ for libuv and other standard library type stuff | 15:55 |
| fijal | :( | 15:56 |
| fijal | does rust come with an ffi? | 15:56 |
| fzzzy | yes | 15:56 |
| fijal | ok, so why c++? | 15:56 |
| voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace | 15:57 | |
| fzzzy | instead of what? c? | 15:57 |
| fzzzy | there's not very much c++ | 15:58 |
| bivab (~david@ip-62-143-59-5.unitymediagroup.de) left irc: Quit: bivab | 15:59 | |
| sven_hager (~sven@dslb-092-072-028-071.pools.arcor-ip.net) left irc: Quit: Leaving | 15:59 | |
| fijal | ok | 16:00 |
| fijal | cool | 16:00 |
| fzzzy | fijal: what about #define though? that seems to be a huge problem for ffis | 16:00 |
| fijal | well, I have a solution for that | 16:01 |
| fijal | which is to compile small snippets of C and to introspect the platform | 16:02 |
| fijal | works fine for pypy | 16:02 |
| exarkun | it's a sensible solution, but I expect that when normal people start trying to use PyPy on Windows, some complaints will arise | 16:02 |
| fzzzy | how do you know what the small snippets of c are? | 16:03 |
| arigato | exarkun: we don't do it at runtime, only at translation time | 16:03 |
| fijal | fzzzy: that's what you declare in your interface files | 16:03 |
| fzzzy | I'm surprised there is so much c++ in rust actually: https://github.com/graydon/rust/tree/master/src/rt | 16:03 |
| fijal | I can imagine some sort of helpers for parsing .h file would help | 16:03 |
| exarkun | arigato: ah well, that's even better :) | 16:03 |
| fijal | exarkun: yeah | 16:04 |
| fijal | exarkun: rffi_platform works quite well for us | 16:04 |
| fzzzy | fijal: you declare small snippets of c? that sounds reasonable | 16:04 |
| kenaan | 12hager ppc-jit-backend 11e83a924bb5d8 15/pypy/jit/backend/ppc/ppcgen/opassembler.py: remove comments | 16:04 |
| kenaan | 12hager ppc-jit-backend 11976ed819b3c4 15/pypy/jit/backend/ppc/ppcgen/ppc_assembler.py: remove debug stuff | 16:04 |
| fijal | trying to expose it at applevel failed miserably | 16:04 |
| fijal | fzzzy: either that or you specify it somehow | 16:04 |
| fijal | it's a matter of convinience | 16:04 |
| fijal | but we don't have say hardcoded sizes of structs | 16:04 |
| fijal | etc. | 16:04 |
| fzzzy | yes, they have been discussing whether it is possible to use clang to discover what the macro structure is like | 16:04 |
| fijal | arigato: would you mind if one of the nightly build will be cpyext-less | 16:07 |
| fijal | ? | 16:07 |
| roelbruggink (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection | 16:08 | |
| arigato | which one? why? and likely yes | 16:08 |
| fijal | one that we have redundant | 16:08 |
| fijal | to make my embedding of cpython work | 16:08 |
| lucian | fzzzy: i would guess rewriting all that C++ in rust itself might be a good test of the language? | 16:08 |
| fijal | I think we have both 32 and 64bit linux that are redundant | 16:08 |
| fijal | and also: | 16:09 |
| fijal | how do I create a code object from char* in CPython's C API? | 16:09 |
| arigato | you can use a -jitnojit, I suppose, but I still have no clue why | 16:09 |
| thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Remote host closed the connection | 16:09 | |
| hruske (~Gasper@188-230-156-183.dynamic.t-2.net) left irc: Quit: Leaving | 16:09 | |
| fzzzy | lucian: the point is to have as much implemented in rust as possible | 16:09 |
| gutworth | fijal: what is the char *? | 16:09 |
| fijal | gutworth: python code | 16:10 |
| fijal | arigato: I want a jit version, we have like 3 different 32bit builds a night | 16:10 |
| fijal | I want to embed cpython in pypy | 16:10 |
| gutworth | Py_Compile* and friends | 16:10 |
| fijal | that requires not using cpyext | 16:10 |
| fijal | gutworth: Py_Compile or PyCompile? | 16:11 |
| arigato | fijal: why not? | 16:11 |
| fijal | why it requires not having cpyext? | 16:11 |
| gutworth | it's not called Py_Compile | 16:11 |
| arigato | fijal: yes | 16:11 |
| gutworth | like Py_CompileSimpleStringFlags | 16:11 |
| gutworth | look at pythonrun.h | 16:12 |
| fijal | arigato: because you get symbol clashes if you try to load libpython2.6.so | 16:12 |
| fijal | using ctypes | 16:12 |
| arigato | ah, that's obscure | 16:12 |
| arigato | can't you load with a flag that means "overwrite all symbols"? | 16:12 |
| fijal | arigato: if you tell me how, then yes, wouldn't that make cpyext stop working though? | 16:13 |
| arigato | yes | 16:13 |
| squiddy (~squiddy@g224195234.adsl.alicedsl.de) joined #pypy. | 16:14 | |
| fijal | ok | 16:14 |
| fijal | well I guess that is fine, but how do I do that? | 16:14 |
| fijal | well, "fine" | 16:14 |
| arigato | maybe RTLD_DEEPBIND? | 16:14 |
| fijal | which I have to pass all the way down since ctypes does not know about it... | 16:15 |
| fijal | that's however fine | 16:15 |
| kenaan | 12arigo default 11062e9d06c908 15/pypy/jit/backend/: Hack: record at least some partial information about which frame locations are freed, and if a hint is present, try... | 16:15 |
| exarkun | fijal: Just write your own linker | 16:16 |
| fijal | exarkun: sounds like a great idea thanks! | 16:16 |
| exarkun | (couldn't be any harder than writing PyPy) | 16:16 |
| fijal | arigato: you've seen my hacks btw? | 16:16 |
| ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) left irc: Quit: Leaving... | 16:18 | |
| arigato | fijal: the screenshot? yes | 16:18 |
| fijal | ok | 16:18 |
| arigato | it looks like a good idea | 16:18 |
| fijal | it works | 16:18 |
| rokujyouhitoma (~rokujyouh@12.51.221.130) left irc: Read error: Connection reset by peer | 16:19 | |
| kenaan | 12arigo SpecialisedTuples 117ad77a6f050c 15/pypy/module/cpyext/tupleobject.py: Fix. Previously, we would risk getting a specialised tuple if enabled. That makes little sense: we can o... | 16:20 |
| Action: fijal wants gateway.applevel | 16:20 | |
| fijal | arigato: I have to reinvent all the cool APIs that objspace has in order to have anything semi-usable | 16:20 |
| arigato | fijal: re symbol clashes: the very obscure alternative is to ensure that pypy+cpyext defines different symbols | 16:23 |
| arigato | by #define'ing them all | 16:24 |
| fijal | yes | 16:24 |
| arigato | e.g. #define PyTuple_GetItem PyPyTuple_GetItem | 16:24 |
| fijal | but this sounds like it goes slightly beyond my hacks now | 16:24 |
| arigato | :-) | 16:24 |
| fijal | although we have support for that | 16:24 |
| fijal | for running inside | 16:24 |
| arigato | uh | 16:25 |
| arigato | ok | 16:25 |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 16:25 | |
| fijal | later :) | 16:27 |
| arigato | SpecialisedTuples fails because hashes are not always correct | 16:31 |
| cfbolz | uh??! | 16:32 |
| cfbolz | arigato: how? | 16:32 |
| kenaan | 12l.diekmann default 11f19d06a5972d 15/pypy/objspace/std/: (l.diekmann, cfbolz): Be more careful about unrolling getitems. also make one less copy | 16:32 |
| kenaan | 12l.diekmann default 116acad4874739 15/pypy/jit/backend/: merge | 16:32 |
| Alex_Gaynor | cfbolz: I believe that's why they call it a bug :) | 16:32 |
| kenaan | 12l.diekmann type-specialized-instances 11e481a093e056 15/pypy/objspace/std/mapdict.py: fixes for type-specialized-attributes | 16:33 |
| K-os (~gabriel@dhcp-205.zbh.uni-hamburg.de) left irc: Ping timeout: 252 seconds | 16:34 | |
| arigato | I have two t=(1,2,3) that are a W_TupleObject (why??) and a W_SpecialisedTupleObject_ooo (why??) | 16:34 |
| arigato | and they don't have the same hash | 16:34 |
| Action: pedronis any sufficiently tested code base will have only obscure bugs left | 16:34 | |
| arigato | :-) | 16:34 |
| Action: fijal just realised his hacks are untestable on CPython | 16:34 | |
| fijal | :-) | 16:35 |
| dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) joined #pypy. | 16:35 | |
| Alex_Gaynor | fijal: sure they are, you just need to use the shared object for the current proc, instead of another one, I think | 16:35 |
| Alex_Gaynor | probably a mess with GIL and refcounts as well | 16:35 |
| arigato | ah, I got at least the 2nd "why??" | 16:36 |
| arigato | bah, indeed, the test fails if I just write it in test_specialisedtupleobject | 16:37 |
| asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy. | 16:38 | |
| pedronis | Alex_Gaynor, at least cpython ctypes has PyDLL to deal with the GIL | 16:39 |
| Alex_Gaynor | pedronis: does that work if you're trying to embed CPython inside of itself? I gues sthat's inherently a mess | 16:39 |
| pedronis | Alex_Gaynor, I think it's partly meant for that, but yes you are a bit on your own and need to be very careful | 16:40 |
| pedronis | Alex_Gaynor, the documentation says as much: http://docs.python.org/library/ctypes.html#ctypes.PyDLL | 16:40 |
| fijal | pedronis: I think I'm past "dragons 25k" roadsign now | 16:41 |
| cfbolz | arigato: about the first why, are we sure that nobody is explicitly instantiating W_TupleObject? | 16:41 |
| pedronis | fijal, they still sell petrol it seems | 16:42 |
| fijal | :] | 16:42 |
| arigato | waa | 16:42 |
| kenaan | 12arigo SpecialisedTuples 11526f61d70b8f 15/pypy/objspace/std/test/test_specialisedtupleobject.py: Kill. | 16:43 |
| kenaan | 12arigo SpecialisedTuples 11f0cc6a3ff8e9 15/pypy/objspace/std/: Test and fix. | 16:43 |
| arigato | cfbolz: I checked, and I found only cpyext | 16:44 |
| cfbolz | arigato: "yay" | 16:44 |
| arigato | cfbolz: ah no, there is also the delegation from | 16:44 |
| arigato | W_SpecializedTuple to W_TupleObject | 16:44 |
| arigato | unsure how it could happen to make a long-lived W_TupleObject | 16:44 |
| arigato | ah yes | 16:45 |
| arigato | if we write e.g. (1, 2) * 1 | 16:45 |
| Rhy0lite (dje@nat/ibm/x-rprjxbrujixjwwyx) joined #pypy. | 16:45 | |
| arigato | should we care? I guess no | 16:45 |
| cfbolz | pfff, I guess we will find out that we *do* care :-( | 16:46 |
| arigato | well right now, "tup*1 is tup" is not true any more | 16:47 |
| arigato | the code is actually written, I just commented it out | 16:47 |
| arigato | instead of reviewing and rpythonifying it | 16:48 |
| ltratt (~ltratt@px238.dcs.kcl.ac.uk) joined #pypy. | 16:50 | |
| kenaan | 12arigo SpecialisedTuples 11d896533b28af 15/pypy/objspace/std/: Re-add and review this code. Needed to ensure that "tup*1 is tup", in case obscure code depends on it. | 16:55 |
| G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving. | 16:55 | |
| kenaan | 12arigo default 118de6f245c959 15/pypy/objspace/std/iterobject.py: Kill this __init__, which nowadays contains just a duplicate of setting to w_seq --- already done in the parent __i... | 16:55 |
| Action: arigato away | 16:56 | |
| arigato (~arigo@fwstups.cs.uni-duesseldorf.de) left irc: Quit: See you | 16:57 | |
| voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy. | 16:57 | |
| voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host | 16:57 | |
| voidspace (~anonymous@python/psf/voidspace) joined #pypy. | 16:57 | |
| neunon (~snoonan@unaffiliated/neunon) left irc: Quit: Leaving... | 17:01 | |
| kennethreitz (~kennethre@ip65-46-252-62.z252-46-65.customer.algx.net) left irc: Quit: Textual IRC Client: http://www.textualapp.com/ | 17:06 | |
| ltratt | Hi all. I seem to having a problem with RPython finalizers and the minimark GC at -O3. One of my finalizers seems to be called while the object is still live (it free's memory, which is later used by a function in that object, causing a core dump). With semispace at -O3, for example, everything is OK. Are there any known gotchas in this area? | 17:07 |
| khs (~khs@151.43.202.84.customer.cdi.no) joined #pypy. | 17:07 | |
| cfbolz | ltratt: can the finalizer resurrect the object? ie store it to a globally reachable position? | 17:08 |
| ltratt | nope, it contains only lltype.free(self.data) | 17:08 |
| cfbolz | ok | 17:08 |
| cfbolz | then I have no clue, sorry | 17:08 |
| cfbolz | also, am about to leave | 17:08 |
| ltratt | thanks anyway :) | 17:08 |
| ltratt | i noticed some commits by fijal on lightweight finalizers that might be relevant? | 17:09 |
| ltratt | certainly if i stick a print statement in there, it seems to delay finalization | 17:09 |
| lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Ping timeout: 240 seconds | 17:09 | |
| fijal | ltratt: hey | 17:20 |
| fijal | ltratt: what do you mean by object is still alive btw? | 17:21 |
| fijal | can you show me the relevant piece of rpython? | 17:22 |
| fijal | for example locals don't keep the object alive | 17:22 |
| Alex_Gaynor | fijal: what? yes it does, if it's a local that can still be used it's kept alive | 17:22 |
| fijal | Alex_Gaynor: not in RPython | 17:23 |
| Alex_Gaynor | fijal: you're syaing def f(n): i = [] while len(i) < n: i.append(3); return i doesn't keep list alive | 17:23 |
| ltratt | fijal: i'm not sure how to extract a useful subset of my code | 17:23 |
| fijal | Alex_Gaynor: you know what I'm saying | 17:23 |
| Alex_Gaynor | no I'm pretty confused | 17:24 |
| fijal | if you get the raw pointer out of a field | 17:24 |
| ltratt | I can send you a whole wodge of stuff if you'd like :) | 17:24 |
| jterrace (~jterrace@aegis.CS.Princeton.EDU) joined #pypy. | 17:24 | |
| fijal | the existance of local does not keep it alive | 17:24 |
| fijal | ltratt: can you start with explaining what you're doing maybe? | 17:24 |
| fijal | is it python? is it rpython? | 17:24 |
| ltratt | It's RPython, part of the Converge VM | 17:25 |
| fijal | http://paste.pocoo.org/show/518032/ | 17:25 |
| fijal | you're probably doing something like that | 17:25 |
| ltratt | Aha! I think you might be right | 17:25 |
| ltratt | let me try... | 17:26 |
| fijal | yeah, that's wrong | 17:26 |
| fijal | if you had gc collection in between it would explode anyway | 17:26 |
| ltratt | Is this stuff written down somewhere? Sometimes I worry that I'm violating subtle constraints that I could only know if I was one of you guys. | 17:27 |
| JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 17:29 | |
| antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy. | 17:29 | |
| ltratt | I have a resizing function which does this http://pastebin.com/XgG7mePD | 17:29 |
| fijal | tests | 17:30 |
| fijal | is the only way... | 17:30 |
| ltratt | should 'keepalive_until_here go before or after the free? I guess after, but then does the free work? :) | 17:30 |
| fijal | no, it's probably not written | 17:30 |
| fijal | what does your finalizer do? | 17:31 |
| ltratt | the finalizer is just lltype.free(self.data, flavor="raw") | 17:31 |
| fijal | note that keepalive has nothing to do with raw memory | 17:31 |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 252 seconds | 17:31 | |
| ltratt | aha | 17:31 |
| ltratt | ok | 17:31 |
| fijal | so it's about keeping alive self | 17:32 |
| ltratt | yes, i think so | 17:32 |
| ltratt | but i wonder how the reference to self can be lost, given that it's later called by other data? | 17:33 |
| fijal | that needs to be kept alive after free() | 17:33 |
| fijal | it can be inlined somewhere? | 17:33 |
| fijal | somehow? | 17:33 |
| ltratt | possible, but unlikely. i suspect there's a reference to that object in the heap (it's a "Converge language level object" if that makes sense) | 17:34 |
| ltratt | of course, I may have made a hugely obvious mistake elsewhere, but I'm damned if I can see it at the moment | 17:34 |
| lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) joined #pypy. | 17:35 | |
| lesshaste | just found my first short piece of code which is faster in pypy than shedskin :) | 17:35 |
| lesshaste | thanks! | 17:35 |
| ltratt | fijal: if you want, i can zip it up and e-mail you my working copy to poke and prod... | 17:36 |
| gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated! | 17:36 | |
| fijal | ltratt: is it open source? | 17:38 |
| fijal | I can do a checkout | 17:38 |
| ltratt | the main bit is (https://github.com/ltratt/converge/tree/pypyvm), but i'm currently in a mess of half-finished changes which aren't suitable for commit and upload yet | 17:39 |
| fijal | well, make a branch | 17:39 |
| Nick change: nedbat_ -> nedbat | 17:40 | |
| ltratt | i was brought up on a different commit model than PyPy: only commit something that's finished. [It's the OpenBSD model.] like i said, i'm happy to zip the whole kaboodle up... | 17:41 |
| exarkun | Hardly any kittens will die if you commit some incomplete changes to a branch | 17:41 |
| ltratt | exarkun: each to their own. | 17:42 |
| exarkun | You're ignoring a perfectly useful tool for no coherent reason. | 17:42 |
| fijal | ltratt: I would be happy to help you, but I should start adopting a policy to not help people who have code to hide for free | 17:42 |
| fijal | I know it sounds harsh and this is not what you really mean | 17:42 |
| santagada (~leonardo@201.47.217.214) left irc: Quit: Ex-Chat | 17:42 | |
| fijal | but still | 17:43 |
| ltratt | it's just a different policy for source control. You find mine weird, I find PyPy's... well, a bit weird too :) But we should all be able to run along well enough I think. | 17:43 |
| exarkun | I dub that argument "version control relativism" | 17:44 |
| exarkun | And it is as bogus as most of other forms of relativism. | 17:44 |
| Alex_Gaynor | ltratt: on pypy we still try to keep default stable with onyl working stuff, we just uses branches aggressively to publish unfinished wokr, I don't think we're actually that far apart in philosophy | 17:44 |
| fijal | ltratt: indeed, I don't think it's that different in the philosophy. I can help with open source stuff, I won't help with stuff that's not open source (for free) | 17:46 |
| ltratt | Alex_Gaynor: In a sense, yes. I don't want to criticise PyPy's model, as it clearly works very well for you. Similarly, other models work well for other people. | 17:46 |
| ltratt | fijal: it's all open source, just not yet in revision control. | 17:46 |
| fijal | if I can't download it it's not very open to me | 17:47 |
| fijal | but stuff aside | 17:47 |
| fijal | if you can paste a finalizer and a method, I can probably tell you what went wrong | 17:47 |
| fijal | also it should work if you put keepalive_until_here(self) at the very end | 17:47 |
| exarkun | ltratt: What problem do you imagine will arise if you check some code into a brand new branch? | 17:48 |
| fijal | ltratt: in the case of the JIT, the object might stay virtual and it's finalizer is called as soon as it goes away | 17:48 |
| fijal | if everything gets inlined it might go away before the end of the method | 17:48 |
| fijal | ltratt: do I make any sense? | 17:48 |
| ltratt | fijal: ok, here's my current (complete) dump at http://tratt.net/current.tar.gz (warning: ~20MB) | 17:49 |
| Action: fijal sidenotes he has 2G internet/month | 17:49 | |
| fijal | ltratt: can you understand my explanation? | 17:49 |
| ltratt | fijal: sort of, but I don't really understand what gets identified as virtual or lined and what doesn't. | 17:50 |
| ltratt | I'm a naive external user in that sense - my guides so far have been the docs, and grep'ing the pypy code. | 17:50 |
| ltratt | That still leaves huge holes in my knowledge :) | 17:50 |
| fijal | well, pypy is a bit undocumented | 17:50 |
| fijal | let me find an example | 17:50 |
| fijal | can someone tell me what's wrong with this: | 17:50 |
| fijal | http://paste.pocoo.org/show/518038/ | 17:50 |
| fijal | ? | 17:50 |
| fijal | gutworth: ? | 17:50 |
| ltratt | overall, my progress with rpython has been very fast, but it would have been crazily fast with more docs. such is life :) | 17:51 |
| fijal | http://paste.pocoo.org/show/518040/ | 17:52 |
| fijal | ltratt: how about that? | 17:53 |
| ltratt | that latter one could be what's happening | 17:53 |
| Alex_Gaynor | fijal: no, JIT doesn't virtualize objs with del | 17:53 |
| ltratt | ah, then it isn't ;) | 17:53 |
| fijal | Alex_Gaynor: even with light finalizers? | 17:54 |
| Alex_Gaynor | fijal: it shouldn't at least, because then __del__ would never be called | 17:54 |
| aurojit (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) left irc: Remote host closed the connection | 17:56 | |
| ltratt | fijal: http://tratt.net/current.tar.xz is an 8MB download, if that helps... | 17:56 |
| apanda (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) joined #pypy. | 17:56 | |
| fijal | Alex_Gaynor: yes it'll | 17:56 |
| fijal | Alex_Gaynor: I did work on it | 17:57 |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds | 17:57 | |
| ltratt | this problem is manifesting at -O3 rather than -Ojit, so is the JIT virtual stuff relevant at that point? [This may be a very silly question on my part.] | 17:57 |
| fijal | ah | 17:58 |
| fijal | ok | 17:58 |
| fijal | but that might be inlined on a different level | 17:58 |
| fijal | anyway | 17:58 |
| fijal | can you just put keepalive_until_here and see if it helps? | 17:58 |
| ltratt | would i do that on self? | 17:58 |
| ltratt | i'm not really sure where it could reasonably go... | 17:59 |
| apanda (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) left irc: Ping timeout: 240 seconds | 18:00 | |
| fijal | yes | 18:01 |
| fijal | on self | 18:01 |
| kkris1 (~kris@80-123-41-164.adsl.highway.telekom.at) joined #pypy. | 18:01 | |
| ltratt | ok, giving it a go now... | 18:02 |
| kkris (~kris@93-82-40-215.adsl.highway.telekom.at) left irc: Ping timeout: 255 seconds | 18:02 | |
| amaury (amaury_@nat/google/x-aroxchifqyoidfxw) left irc: Ping timeout: 252 seconds | 18:04 | |
| ltratt | attempt #1 didn't work, trying #2 now... | 18:05 |
| ltratt | aha #2 that seems to have worked | 18:08 |
| ltratt | now of course, i just have to work out why ;) | 18:09 |
| fijal | what's #1 and #2? | 18:09 |
| fijal | gutworth: ping? | 18:09 |
| fijal | if Return Value is new reference, do I need to incref it? | 18:09 |
| ltratt | different locations for keepalive | 18:10 |
| ltratt | #2 is just after the point where it segfaults, and it seems to work... but why...? dunno ;) | 18:10 |
| voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace | 18:10 | |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy. | 18:11 | |
| jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Ping timeout: 252 seconds | 18:13 | |
| jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy. | 18:14 | |
| ltratt | fijal: ok, thanks for the pointer. | 18:16 |
| ltratt | i still don't really understand why it's working now, but it does seem to be ok. of course, now i'm worried that because i don't understand things, i've just put a sticking plaster over one, but have left others :) | 18:17 |
| fijal | ltratt: you should always keepalive stuff around such calls | 18:17 |
| ltratt | once i've got the mess sorted out and checked in, i might run it past you to see if I'm being really, really dumb | 18:17 |
| fijal | it's generally dangerous to live such stuff around without a keepalive | 18:17 |
| ltratt | I still don't really understand what keepalive is doing, to be honest, or why i need it. clearly my intuitions about the memory model are wrong. | 18:18 |
| Alex_Gaynor | ltratt: basically, if I follow, you have memory that is "owned" by an object, and you're passing that object somewhere, but the object that owns it isn't necesarily alive as long as the raw memory it holds, so you need to tell the compiler "hey, this owns that memory, so keep it alive so long as I'm using the raw memory" | 18:19 |
| ltratt | Alex_Gaynor: i think the "owning" object is alive longer than the raw memory. at least, that's what should be happening ;) | 18:19 |
| Alex_Gaynor | ltratt: right that's the problem, the owning object *should* be alive longer than it, but it has no references (a local vairable only counts as a reference *if*it can be used later), so you basically create an artificial reference until the raw memory is no longer directly in use | 18:20 |
| jterrace | Alex_Gaynor: you have a sec to help me debug a numpypy patch im working on for tolist() function? it's almost there, but having one problem | 18:21 |
| Alex_Gaynor | jterrace: sure | 18:21 |
| ltratt | hmm, i need to think about this carefully. my gut instinct is that there should be a reference from the heap to this object, but i might be wrong. | 18:21 |
| jterrace | Alex_Gaynor: here's my patch so far: http://pastebin.com/qYkJQPMF it works for individual items and for single-dim arrays | 18:21 |
| jterrace | but for multidim, im getting back [[1,2],[1,2]] instead of [[1,2],[3,4]] which makes me think that self.desc_getitem is not returning the right slice index | 18:22 |
| jterrace | but i dont see why | 18:22 |
| ltratt | fijal and Alex_Gaynor, thanks a lot for your help today - I appreciate it! | 18:23 |
| Alex_Gaynor | I'd hazzard a guess that tolist() doesn't work correctly on slice object | 18:23 |
| Alex_Gaynor | I'd write direct tests for that, as well as virtual arays (i.e. a + a) | 18:23 |
| jterrace | oh, im supposed to use start_iterator instead of just range(self.shape[0]) eh? | 18:23 |
| ojii (~ojii@40-34.60-188.cust.bluewin.ch) left irc: Remote host closed the connection | 18:24 | |
| Alex_Gaynor | and yeah, the iterator are generally the right way to go | 18:24 |
| jterrace | ok | 18:24 |
| Alex_Gaynor | not sure if we have one that iterates by axis thoug | 18:24 |
| jterrace | ill try that, thanks | 18:24 |
| jterrace | oh yeah i think it's a flat iterator | 18:24 |
| ltratt (~ltratt@px238.dcs.kcl.ac.uk) left irc: Quit: Bye! | 18:26 | |
| jterrace | ooh it works | 18:28 |
| Action: jterrace does a dance | 18:28 | |
| jterrace | ill write more tests and submit a pull request | 18:28 |
| Alex_Gaynor | great | 18:28 |
| fijal | haha | 18:32 |
| mtigas (~mtigas@users.spokesman.com) joined #pypy. | 18:34 | |
| hpk (~hpk@hq2.merlinux.eu) left irc: Remote host closed the connection | 18:35 | |
| mtigas_ (~mtigas@users.spokesman.com) joined #pypy. | 18:36 | |
| mtigas_ (~mtigas@users.spokesman.com) left irc: Client Quit | 18:36 | |
| mtigas (~mtigas@users.spokesman.com) left irc: Client Quit | 18:36 | |
| mtigas (~mtigas@users.spokesman.com) joined #pypy. | 18:37 | |
| Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Quit: ZNC - http://znc.sourceforge.net | 18:37 | |
| DasIch__ (~dasich@p4FFDE31D.dip.t-dialin.net) joined #pypy. | 18:37 | |
| voidspace (~voidspace@python/psf/voidspace) joined #pypy. | 18:37 | |
| voidspace (~voidspace@python/psf/voidspace) left irc: Client Quit | 18:38 | |
| gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) joined #pypy. | 18:38 | |
| voidspace (~voidspace@python/psf/voidspace) joined #pypy. | 18:38 | |
| DasIch_ (~dasich@p4FFDD13B.dip.t-dialin.net) left irc: Ping timeout: 240 seconds | 18:40 | |
| JaRoel (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection | 18:44 | |
| apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy. | 18:45 | |
| Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy. | 18:45 | |
| mtigas (~mtigas@users.spokesman.com) left irc: Quit: mtigas | 18:52 | |
| jterrace | Alex_Gaynor: my first attempt at contributing: https://bitbucket.org/pypy/pypy/pull-request/15/added-tolist-function-to-numpypys-scalars | 18:53 |
| jterrace | if anything looks wrong, let me know, hoping to contribute more in the future | 18:53 |
| Alex_Gaynor | jterrace: I need lunch, but I'll give it a review this afternoon | 18:53 |
| jterrace | thanks | 18:53 |
| mtigas (~mtigas@users.spokesman.com) joined #pypy. | 18:54 | |
| kennethreitz (~kennethre@204.14.152.118) joined #pypy. | 18:55 | |
| `fox` (~fox@host91-171-dynamic.245-95-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds | 18:57 | |
| `fox` (~fox@seattle237.riseup.net) joined #pypy. | 18:59 | |
| etrepum (~bob@75-101-96-144.dsl.static.sonic.net) left irc: Quit: etrepum | 18:59 | |
| fijal | er | 19:02 |
| fijal | how do I get builtins into globals? | 19:02 |
| fijal | in CPython C API? | 19:03 |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 19:03 | |
| `fox` (~fox@seattle237.riseup.net) left irc: Ping timeout: 252 seconds | 19:07 | |
| JaRoel|4d (~jaroel|4d@sink.jaroel.nl) joined #pypy. | 19:08 | |
| `fox` (~fox@host91-171-dynamic.245-95-r.retail.telecomitalia.it) joined #pypy. | 19:11 | |
| fijal | ok, now this is disconcerning | 19:13 |
| pjenvey | fijal - hmm? | 19:13 |
| fijal | pjenvey: pypy ctypes | 19:14 |
| espes (~espes@123-243-221-248.tpgi.com.au) left irc: Read error: Connection reset by peer | 19:17 | |
| kenaan | 12arigo SpecialisedTuples 115f96cb15c116 15/pypy/objspace/std/: - fix space.fixedlist/unpackiterable/listview to handle directly tuples of any kind - fix the test to ... | 19:17 |
| espes (~espes@123-243-221-248.tpgi.com.au) joined #pypy. | 19:18 | |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/619 [12arigo, SpecialisedTuples] | 19:20 |
| mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord | 19:26 | |
| espes (~espes@123-243-221-248.tpgi.com.au) left irc: Remote host closed the connection | 19:27 | |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 258 seconds | 19:30 | |
| espes (~espes@123-243-221-248.tpgi.com.au) joined #pypy. | 19:33 | |
| K-os (~gabriel@port-18238.pppoe.wtnet.de) joined #pypy. | 19:35 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy. | 19:36 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host | 19:36 | |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 19:36 | |
| mfoord (~anonymous@python/psf/voidspace) left irc: Client Quit | 19:36 | |
| espes (~espes@123-243-221-248.tpgi.com.au) left irc: Client Quit | 19:38 | |
| espes (~espes@123-243-221-248.tpgi.com.au) joined #pypy. | 19:38 | |
| jterrace (~jterrace@aegis.CS.Princeton.EDU) left irc: Ping timeout: 240 seconds | 19:39 | |
| jterrace (~jterrace@aegis.CS.Princeton.EDU) joined #pypy. | 19:44 | |
| kennethreitz | Is anyone working on getting pypy to work with virtualenv --relocatable? | 19:46 |
| ronny | kennethreitz: you might want to neogiate with antocuni (he made the virtualenv support possible0 | 19:48 |
| kennethreitz | ronny: excellent, thanks | 19:48 |
| Alex_Gaynor | or the virtualenv people | 19:48 |
| Alex_Gaynor | it's expiremental at best on CPython | 19:48 |
| antocuni | kennethreitz: AFAIK, nobody ever tried to use it | 19:50 |
| antocuni | which means that it probably doesn't work | 19:50 |
| kennethreitz | antocuni: oh it def doesn't. looking into fixing, just didn't want to duplicate any work already being done | 19:50 |
| antocuni | ah, that would be cool | 19:50 |
| antocuni | ping me if you need any help | 19:50 |
| kennethreitz | antocuni: thanks! | 19:51 |
| antocuni | I don't even know how it is implemented for cpython | 19:51 |
| kennethreitz | & it's a bit odd. | 19:51 |
| arigato (~arigo@89.204.137.229) joined #pypy. | 19:51 | |
| Alex_Gaynor | heh, that's charitable | 19:51 |
| Action: Alex_Gaynor off | 19:51 | |
| daniloaf (~daniloaf@177.42.189.10) joined #pypy. | 19:52 | |
| daniloaf (~daniloaf@177.42.189.10) left irc: Client Quit | 19:55 | |
| mattip (4fb46e04@gateway/web/freenode/ip.79.180.110.4) joined #pypy. | 19:58 | |
| mattip | Could anyone tell me what the proper protocol is for the compatability bitbucket repository? It seems only the wiki https://bitbucket.org/pypy/compatibility/wiki/Home has any content, the repository itself seems to be empty. | 20:01 |
| ronny | mattip: the wiki repo is in a subdir | 20:02 |
| ronny | mattip: i dont remember the details, fogire it yourself, shouldnt be too hard | 20:03 |
| jterrace | Alex_Gaynor: when you say that descr_tolist should always return a wrapped value, should i unwrap the result of the recursive call to the function, then wrap it again when returning? | 20:03 |
| jterrace | looks like that works - are there any performance implications to that? | 20:05 |
| mattip | ronny: thanks, I was just wondering if the only active par tof the repository is the wiki, it seems there is space for actual compatability code but that part is empty. | 20:05 |
| lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) left irc: Quit: Leaving | 20:08 | |
| mat^2 (~mathias@212.130.113.35) joined #pypy. | 20:10 | |
| etrepum (~bob@64.134.236.31) joined #pypy. | 20:11 | |
| fijal | antocuni: ping | 20:12 |
| antocuni | pong | 20:12 |
| espes (~espes@123-243-221-248.tpgi.com.au) left irc: Read error: Connection reset by peer | 20:13 | |
| fijal | antocuni: so, I have functions that are ctypes.c_void_p | 20:13 |
| fijal | and what I receive are longs | 20:13 |
| fijal | why? | 20:13 |
| antocuni | because there is a bug and it's untested? :-) | 20:14 |
| antocuni | what happens on cpython? | 20:14 |
| ericflo (~ericflo@75.103.8.110) joined #pypy. | 20:14 | |
| fijal | I suppose you get c_void_p | 20:14 |
| espes (~espes@123-243-221-248.tpgi.com.au) joined #pypy. | 20:14 | |
| mattip (4fb46e04@gateway/web/freenode/ip.79.180.110.4) left irc: Ping timeout: 265 seconds | 20:14 | |
| antocuni | internally, _ffi indeed return a python long for pointers | 20:14 |
| antocuni | so it means that probably the ctypes layer "forgets" to wrap it inside a c_void | 20:15 |
| jnoller (~jnoller@vpn.nasuni.com) left irc: Quit: Leaving... | 20:17 | |
| fijal | uh no, cpthon does the same | 20:17 |
| antocuni | "good" | 20:17 |
| fijal | wtf? | 20:17 |
| Action: antocuni looks at the code | 20:18 | |
| fijal | it's int not long but whatever | 20:19 |
| fijal | I bet it would be long in some other circumstances | 20:19 |
| arigato | fijal: I can explain you why if you care | 20:20 |
| antocuni | fijal: look at primitive.py._CData_output | 20:20 |
| fijal | arigato: yes? | 20:20 |
| fijal | antocuni: I can imagine how is it implemented, but why? | 20:21 |
| antocuni | if it's a "primitive" type, it just returns it's .value | 20:21 |
| antocuni | I don't know what is the rationale behind it | 20:21 |
| arigato | it's a primitive type like c_int. if a function is declared to return a c_int, you actually get its .value, which is an int. similarly for c_void_p, whose .value is a long | 20:21 |
| arigato | yes | 20:21 |
| antocuni | but e.g. in this way it means that you get a python string instead of a c_char_p | 20:21 |
| antocuni | which is usually what you want | 20:22 |
| arigato | usually it's nonsense, so don't declare the function as returning c_void_p | 20:22 |
| arigato | use a real pointer type | 20:22 |
| antocuni | we really need to design/implement a better ffi for python :-) | 20:22 |
| hruske (~Gasper@188-230-156-183.dynamic.t-2.net) joined #pypy. | 20:23 | |
| hruske (~Gasper@188-230-156-183.dynamic.t-2.net) left irc: Client Quit | 20:23 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/619 [12arigo, SpecialisedTuples] | 20:23 |
| fijal | arigato: well, yes, but I'm saying PyObject | 20:24 |
| fijal | which is an opaque structure | 20:24 |
| fijal | I'm fine with longs | 20:24 |
| fijal | I did not notice even | 20:24 |
| RiaanFV (~riaanfv@41-132-169-26.dsl.mweb.co.za) joined #pypy. | 20:27 | |
| antocuni | fijal: I think that the proper way is to really define the PyObject struct and create a pointer to it | 20:27 |
| fijal | and what would that structure have? | 20:28 |
| lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy. | 20:28 | |
| Alex_Gaynor | antocuni: I'm designing a better one, still a WIP though | 20:29 |
| antocuni | Alex_Gaynor: I always think about it when designing the pypy's _ffi | 20:30 |
| etrepum_ (~bob@75.94.138.216) joined #pypy. | 20:30 | |
| antocuni | we should probably sit down and do it together :-) | 20:30 |
| Alex_Gaynor | antocuni: so far all I've done is written documentation, I'll push it to a repository- tonight and give you permissions | 20:31 |
| antocuni | I'll look at it, thanks | 20:31 |
| RiaanFV (~riaanfv@41-132-169-26.dsl.mweb.co.za) left irc: Client Quit | 20:31 | |
| Rhy0lite (dje@nat/ibm/x-rprjxbrujixjwwyx) left irc: Quit: Leaving | 20:31 | |
| antocuni | fijal: I wrote the ctypes code to manipulate PyObjects & co. some time ago | 20:32 |
| jterrace | Alex_Gaynor: updated pull request, thanks for the feedback | 20:32 |
| antocuni | but I'm unable to find it :-/ | 20:32 |
| antocuni | fijal: ah, found :-) | 20:32 |
| Alex_Gaynor | jterrace: in class ATM, will review later this evening | 20:32 |
| jterrace | thanks | 20:33 |
| antocuni | http://paste.pocoo.org/show/518096/ | 20:33 |
| fijal | http://paste.pocoo.org/show/518097/ | 20:33 |
| fijal | Alex_Gaynor: ? | 20:33 |
| Alex_Gaynor | fijal: ? | 20:33 |
| etrepum (~bob@64.134.236.31) left irc: Ping timeout: 240 seconds | 20:33 | |
| Nick change: etrepum_ -> etrepum | 20:33 | |
| fijal | Alex_Gaynor: look at the paste | 20:33 |
| Alex_Gaynor | fijal: you're on a 64-bit platform? | 20:33 |
| fijal | yes | 20:33 |
| Alex_Gaynor | fijal: it's an issue of "long" dtype vs 64-bit int dtype | 20:34 |
| fijal | well, they should be identical no? | 20:34 |
| Alex_Gaynor | They aren't identitcal | 20:34 |
| Alex_Gaynor | CPython has teh same behavior | 20:34 |
| fijal | how do I check what sort of dtype is that anyway? | 20:34 |
| fijal | do I need to finish __array_interface__? | 20:34 |
| Alex_Gaynor | fijal: maybe == should work though, evne if `is` doesn't | 20:34 |
| fijal | it knows it's in64 | 20:34 |
| fijal | but I can't get the info | 20:34 |
| Alex_Gaynor | which info do you need? | 20:35 |
| Alex_Gaynor | fijal: looks like it's our bug http://bpaste.net/show/20817/ | 20:35 |
| fijal | Alex_Gaynor: http://paste.pocoo.org/show/518103/ | 20:39 |
| fijal | blog post? | 20:39 |
| Alex_Gaynor | fijal: that's AWESOME, tweet it first :) | 20:40 |
| Alex_Gaynor | add some comments though | 20:40 |
| fijal | like what? | 20:41 |
| Alex_Gaynor | saying it's embedding CPython and automatically sharing memory | 20:42 |
| Alex_Gaynor | fijal: maybe add a nicer helper API so you can do matplotlib.plot directly | 20:42 |
| fijal | yes yes | 20:43 |
| ericflo (~ericflo@75.103.8.110) left irc: Read error: Connection reset by peer | 20:43 | |
| fijal | I'm sure my refcounts are fucked btw | 20:44 |
| Alex_Gaynor | who's aren't | 20:44 |
| Alex_Gaynor | whose* | 20:44 |
| Alex_Gaynor | Py_INCREF all the things | 20:44 |
| fijal | http://paste.pocoo.org/show/518105/ | 20:46 |
| fijal | better? | 20:47 |
| ericflo (~ericflo@75.103.8.110) joined #pypy. | 20:47 | |
| Alex_Gaynor | I usually put comments above the line they refer to, but yeah | 20:47 |
| fijal | http://paste.pocoo.org/show/518106/ | 20:48 |
| fijal | ? | 20:48 |
| Alex_Gaynor | yes, nice | 20:48 |
| khs (~khs@151.43.202.84.customer.cdi.no) left irc: Quit: Leaving | 20:48 | |
| fijal | good, tweeted | 20:49 |
| fijal | I'll write a blog post | 20:49 |
| Alex_Gaynor | awesome | 20:49 |
| Alex_Gaynor | nice | 20:49 |
| fijal | it's all in my hack dir btw | 20:49 |
| fijal | https://bitbucket.org/fijal/hack2 | 20:49 |
| fijal | with some tests | 20:49 |
| kenaan | 12arigo SpecialisedTuples 11bbaae4e2d252 15/pypy/objspace/std/test/test_specialisedtupleobject.py: Fixes for runappdirect on top of pypy. | 20:50 |
| fijal | it's not-quite-yet-working | 20:50 |
| fijal | as in, I'm sure you run sooner or later into some issues | 20:50 |
| fijal | obviously tons of things don't work | 20:51 |
| Alex_Gaynor | of course | 20:51 |
| fijal | like the next thing I would like to do would be to pass a keyword arg or so | 20:51 |
| fijal | but it's just a simple matter of engineering | 20:51 |
| fijal | it's funny | 20:51 |
| fijal | I feel like I did such work already like 5 times | 20:52 |
| fijal | there are so many places where you wrap or unwrap python objects around some interface | 20:52 |
| Alex_Gaynor | abstract all the things | 20:52 |
| fijal | lib/distributed, cpyext, python/rpython | 20:52 |
| fijal | etc. | 20:52 |
| Alex_Gaynor | :) | 20:52 |
| fijal | struct | 20:52 |
| fijal | ctypes | 20:52 |
| fijal | etc. | 20:52 |
| squiddy (~squiddy@g224195234.adsl.alicedsl.de) left irc: Quit: Leaving | 20:52 | |
| fijal | it seriously all sounds very familiar | 20:52 |
| fijal | ok | 20:53 |
| fijal | it's 11pm | 20:53 |
| Alex_Gaynor | how did the PUG go? | 20:53 |
| fijal | I gonna have a glass of wine of victory and will write blog post tomorrow | 20:53 |
| fijal | it's on Sat | 20:53 |
| Alex_Gaynor | ah | 20:53 |
| fijal | I don't have slides, but I do have cool demos :) | 20:53 |
| Alex_Gaynor | assuming our pycon talk is accepted, these demos will be great | 20:53 |
| fijal | yop | 20:53 |
| fijal | I should reply to martijn on pypy-dev | 20:54 |
| fijal | "it's all cool and doable, but I actually have interesting problems to solve" | 20:54 |
| Alex_Gaynor | pretty much | 20:54 |
| fijal | I should write a blog post on my own blog "pypy should do X" | 20:55 |
| etrepum (~bob@75.94.138.216) left irc: Quit: etrepum | 20:56 | |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 21:03 | |
| arigato (~arigo@89.204.137.229) left irc: Quit: See you | 21:03 | |
| zain (~textual@c-67-160-201-63.hsd1.ca.comcast.net) joined #pypy. | 21:09 | |
| zain (~textual@c-67-160-201-63.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep. | 21:24 | |
| fijal | for what is worth RTLD_DEEPBIND does not seem to work | 21:30 |
| Alex_Gaynor | in pypy, in cpython, in both? | 21:30 |
| fijal | with dlopen | 21:31 |
| fijal | I' | 21:31 |
| fijal | m passing mode by hand | 21:31 |
| _pyre_ (~stian@245.232.189.109.customer.cdi.no) joined #pypy. | 21:31 | |
| Alex_Gaynor | err, so it's broken in libc you're saying? | 21:31 |
| etrepum (~bob@accessnat4.mochimedia.net) joined #pypy. | 21:32 | |
| tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) joined #pypy. | 21:32 | |
| tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) left irc: Changing host | 21:32 | |
| tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy. | 21:32 | |
| _pyre_ | just a quick question regardling pypy's try...except, does the general code inside a try except block get slower or is cost of the "try" block only in the setup? | 21:32 |
| fijal | I don't know | 21:32 |
| fijal | I can't really understand what the manpage is talking about | 21:33 |
| fijal | Alex_Gaynor: check man dlopen? | 21:33 |
| fijal | _pyre_: there is no cost for try: except: usually | 21:33 |
| Alex_Gaynor | fijal: http://paste.pocoo.org/show/518127/ you're trying to make it work with cpyext? | 21:34 |
| _pyre_ | hmm, ok | 21:34 |
| _pyre_ | trying to optimize a bit of code for pypy | 21:34 |
| fijal | yeah | 21:35 |
| fijal | _pyre_: you're a bit waaay to close to single asm instructions that way | 21:35 |
| fijal | _pyre_: have you looked at traces? | 21:35 |
| _pyre_ | hmm, no, thats the next step, just now doing some general tricks to it, but there was like 15 try and excepts inside one function | 21:37 |
| Alex_Gaynor | you can't improve performanc by guessing, it's searching for a needle in ahaystakc, not shooting fish in a barrell most of the time | 21:37 |
| rokujyouhitoma (~rokujyouh@173.227.61.4) joined #pypy. | 21:37 | |
| lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Remote host closed the connection | 21:42 | |
| fprimex (~fprimex@brent-macbook.sc.fsu.edu) left irc: Quit: http://www.fprimex.com | 21:50 | |
| fijal | Alex_Gaynor: did you notice that pushes to extradoc are just fine? | 21:51 |
| fijal | it's only pypy that takes forever to push | 21:51 |
| Alex_Gaynor | fijal: were they broken at some point? | 21:51 |
| Alex_Gaynor | fijal: maybe we have too many branches? or something | 21:52 |
| kenaan | 12fijal extradoc 11f6959a758880 15/blog/draft/matplotlib.rst: a draft | 21:52 |
| kenaan | 12fijal extradoc 1137901e468764 15/planning/micronumpy.txt: merge | 21:52 |
| fijal | dunno | 21:52 |
| fijal | Alex_Gaynor: feel like reviewing? | 21:52 |
| Alex_Gaynor | sure-ish I'm in class so I'm only half paying attention | 21:52 |
| fijal | to the class or to the blog? | 21:53 |
| jterrace | you pay attention in class? | 21:53 |
| pjenvey | probably both | 21:53 |
| fijal | jterrace: yeah, he's observably less productive when he's attending classes | 21:54 |
| kkris1 (~kris@80-123-41-164.adsl.highway.telekom.at) left irc: Quit: Leaving. | 21:55 | |
| pjenvey | fijal - 3rd sentence 'right there were' -> 'right that were' | 21:55 |
| fijal | pjenvey: don' | 21:56 |
| fijal | t you have commit rights? | 21:56 |
| pjenvey | let's see | 21:56 |
| fijal | Alex_Gaynor: anyway, feel free to review, I have a screenshot can just post it | 21:57 |
| kenaan | 12alex_gaynor extradoc 112d85277632f9 15/blog/draft/matplotlib.rst: some review | 21:58 |
| Alex_Gaynor | fijal: pushed some review | 21:58 |
| fijal | Alex_Gaynor: why not what? | 21:58 |
| amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy. | 21:59 | |
| Alex_Gaynor | fijal: why doens't this work on other platforms? | 21:59 |
| fijal | because ctypes is a bitch | 21:59 |
| Alex_Gaynor | or should it just say "has only been tested on " | 21:59 |
| fijal | and I don't feel like being deliberate | 21:59 |
| Alex_Gaynor | Ok, so in theory it can work, but it's untested | 21:59 |
| fijal | I *know* it won't work as it is now | 21:59 |
| fijal | but I don't really want to care | 21:59 |
| kenaan | 12alex_gaynor extradoc 1182c71fa181e5 15/blog/draft/matplotlib.rst: clarrify | 22:00 |
| fijal | otherwise good to go? | 22:02 |
| kenaan | 12fijal extradoc 113977810e2b86 15/blog/draft/matplotlib.rst: address the review | 22:02 |
| kenaan | 12fijal extradoc 1184a45e78b80d 15/blog/draft/matplotlib.rst: merge | 22:02 |
| Alex_Gaynor | add the picture :) | 22:02 |
| fijal | yeah, I can't do it in rst | 22:03 |
| Alex_Gaynor | .. image:: | 22:04 |
| Alex_Gaynor | ? | 22:04 |
| fijal | and put what? | 22:04 |
| Alex_Gaynor | put the image in the repo? | 22:04 |
| fijal | no, that's not how blogger does it | 22:04 |
| fijal | don't worry | 22:04 |
| Alex_Gaynor | I want to see the iamge :) | 22:04 |
| kenaan | 12fijal extradoc 113e9cbd0c0d54 15/blog/draft/: add a screenshot | 22:04 |
| kenaan | 12fijal extradoc 1189893594fd04 15/blog/draft/matplotlib.rst: links | 22:05 |
| Alex_Gaynor | fijal: can yo use a screenshot tha tincludes the shell showing it being run with pypy? | 22:05 |
| fijal | like is this any proof of anything... | 22:06 |
| Alex_Gaynor | w/e looks cool IMO though | 22:06 |
| fijal | meh | 22:06 |
| fijal | I'll stick to that one | 22:06 |
| Alex_Gaynor | ok | 22:07 |
| fijal | http://morepypy.blogspot.com/2011/12/plotting-using-matplotlib-from-pypy.html | 22:08 |
| Alex_Gaynor | i'll tweet it | 22:09 |
| fijal | OMG, my horrible spelling | 22:09 |
| Alex_Gaynor | fix? | 22:09 |
| fijal | on tweet? | 22:10 |
| Alex_Gaynor | delete it and RT pypyproject | 22:10 |
| Alex_Gaynor | :) | 22:10 |
| jterrace | that's awesome | 22:12 |
| fijal | jterrace: no, that's broken | 22:13 |
| fijal | but it can be done :) | 22:13 |
| stakkars | fijal: hi. | 22:13 |
| jterrace | it's still awesome even if it's broken | 22:13 |
| stakkars | I can speed it up? | 22:14 |
| fijal | stakkars: speed up what? | 22:14 |
| stakkars | you said that to me, today | 22:14 |
| Action: fijal has no idea | 22:14 | |
| K-os (~gabriel@port-18238.pppoe.wtnet.de) left irc: Quit: Leaving | 22:17 | |
| stakkars | fijal: stakkars: seems compile_c took tons of time | 22:18 |
| stakkars | [3:54pm] fijal: if it's multicore it can be sped up by quite a bit | 22:18 |
| stakkars | [3:56pm] | 22:18 |
| fijal | stakkars: by passing --make-jobs=2 | 22:18 |
| fijal | or so | 22:18 |
| fijal | I don't know if it works on windows | 22:18 |
| stakkars | hard to find. I will toss colloquy | 22:18 |
| mikefc | I thought numpy was just a fancy array library? | 22:19 |
| stakkars | ah?? oh whow! that accelerates the build, then. | 22:19 |
| exarkun | mikefc: "just" | 22:19 |
| fijal | stakkars: only compile_c step | 22:19 |
| `fox` (~fox@host91-171-dynamic.245-95-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds | 22:19 | |
| stakkars | yes yes. But it involves pypy as well, because of the preprocessor analysis, I think | 22:20 |
| fijal | yes, but it's still parallelizable | 22:21 |
| pjenvey | make-jobs is usually determined automatically. not so on windows? | 22:21 |
| fijal | because of asmgcc | 22:21 |
| fijal | yes, I think so | 22:21 |
| stakkars | in any case, my snakepit is slower than I thought, at least with virtualbox. | 22:21 |
| stakkars | asmgc is -Ojit ans -O2 ? | 22:21 |
| stakkars | s/ans/and/ | 22:21 |
| antocuni | fijal: what is RTLD_DEEPBIND supposed to do? | 22:22 |
| fijal | antocuni: good question | 22:22 |
| fijal | antocuni: if you read the dlopen manual and understand, feel free to explain | 22:22 |
| stakkars | anyway, I will try a native Linux build, to clarify the negative picture of my oh so great machine | 22:23 |
| fijal | stakkars: I think so | 22:23 |
| stakkars | general question: | 22:23 |
| stakkars | does it make sense to run the tests with pypy-c? | 22:23 |
| fijal | depends how many you want to run | 22:24 |
| fijal | usually no | 22:24 |
| stakkars | or would I this way perhaps miss a hidden bug? | 22:24 |
| aboudreault (~alanb@osgeo/member/aboudreault) left irc: Quit: Leaving | 22:24 | |
| fijal | you should not | 22:24 |
| antocuni | fijal: it might have something to do with LD_PRELOAD. E.g., usually if you LD_PRELOAD a lib then its symbols have precedence over the newly loaded ones | 22:24 |
| stakkars | usually no means the tests like "own-xxx" are run with CPython? | 22:25 |
| fijal | stakkars: yes | 22:25 |
| antocuni | DEEPBIND seems to mean that you ignore the already loaded symbols, and use the actual ones contained the in lib | 22:25 |
| Action: mikefc thinks google should dump some $$$ into pypy | 22:25 | |
| fijal | antocuni: that's not what it does | 22:25 |
| Action: mikefc empties the moths and dustbunnies from his wallet | 22:25 | |
| fijal | mikefc: why would they care btw | 22:25 |
| fijal | ? | 22:25 |
| mikefc | fijal: google do a lot of python in house | 22:26 |
| stakkars | so only for building, anyway. sniff :'-( | 22:26 |
| fijal | I don't think any of it, except youtube is performance critical | 22:26 |
| wulczer (~wulczer@wulczer.org) joined #pypy. | 22:26 | |
| mikefc | I'm just dreaming here :) | 22:27 |
| fijal | mikefc: getting money is tricky | 22:27 |
| kennethreitz (~kennethre@204.14.152.118) left irc: Ping timeout: 252 seconds | 22:27 | |
| Action: mikefc robs a bank | 22:27 | |
| fijal | mikefc: for one, I would not mind someone giving me some paid stuff to work on | 22:27 |
| stakkars | ok, I agree, until I have a pypy-c that ultimatively passes all standard tests, right? | 22:27 |
| fijal | stakkars: pypy-c -A should work btw | 22:28 |
| mikefc | all the people bitching about numpypy sound like those bitching about windows linux. "We're not moving to your platform or even thinking it's good until it's compatible with a decade of other stuff" | 22:28 |
| stakkars | I mean, at some point we want to use pypy as the standard and only occasionally use cpython. | 22:28 |
| fijal | stakkars: yes, but we're not quite there yet | 22:29 |
| fijal | we have too slow interpreter/too slow jit compilation | 22:29 |
| fijal | one of those | 22:29 |
| fijal | or both | 22:29 |
| stakkars | pypy-c -A is what? (I cannot read or remember so many things) | 22:29 |
| fijal | py.test -A | 22:29 |
| fijal | run tests directly | 22:29 |
| fijal | apptests | 22:29 |
| stakkars | directly, meaning what, as opposed to what? | 22:31 |
| fijal | as on top of py.py | 22:32 |
| stakkars | don't understand, need to look it up | 22:33 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/164 | 22:36 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/738 | 22:36 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/969 | 22:36 |
| kennethreitz (~kennethre@204.14.152.118) joined #pypy. | 22:36 | |
| antocuni | stakkars: all tests inside "class AppXXXTest" are run by the pypy interpreter written in rpython | 22:37 |
| antocuni | if you use -A, they are run directly on the python interpreter used by py.test | 22:37 |
| antocuni | so, py.test runs applevel tests on py.py | 22:37 |
| antocuni | py.test -A runs applevel tests on top of cpython | 22:38 |
| antocuni | pypy-c py.test -A runs applevel tests on top of pypy-c | 22:38 |
| Action: fijal goes to bed now | 22:38 | |
| fijal | cool it works :) | 22:38 |
| antocuni | fijal: I think you should really rename embed.emb before doing the blog post | 22:39 |
| fijal | antocuni: it's in my hack directory!!!! | 22:39 |
| fijal | it's name resembles the stability state it is in | 22:39 |
| stakkars | yes, just found that. That means, -A makes very much sense for testing pypy-c, right | 22:39 |
| antocuni | still, it's ugly :-) | 22:39 |
| fijal | so? | 22:39 |
| stakkars | antocuni: thanks | 22:39 |
| fijal | it's not as ugly as it's contents | 22:39 |
| fijal | and refcounts are wrong | 22:40 |
| fijal | and tons of other stuff | 22:40 |
| fijal | it's only a proof it works | 22:40 |
| fijal | you can't even run tests without setting PYTHONPATH | 22:40 |
| antocuni | as long as it it's only in your hack dir, it's all fine | 22:40 |
| mikefc | fijal: matplotlib hack is awesome! | 22:40 |
| antocuni | but if you put it on the blog, I'd prefer a nicer name | 22:40 |
| antocuni | anyway, I'm not going to argue more about it :-) | 22:41 |
| antocuni | sorry | 22:41 |
| fijal | antocuni: I would prefer nicer hacks | 22:41 |
| fijal | (seriously) | 22:41 |
| fijal | btw feel free to come up with something better | 22:41 |
| antocuni | that' orthogonal | 22:41 |
| berkerpeksag (~berkerpek@46.196.255.228) joined #pypy. | 22:41 | |
| antocuni | (I'd prefer no more war in the world, fwiw) | 22:42 |
| stakkars | antocuni: I will write a blog post. Do we prepare that as text file before, for safety and review, or directly into blogspot? | 22:42 |
| berkerpeksag (~berkerpek@46.196.255.228) left irc: Client Quit | 22:42 | |
| antocuni | stakkars: we usually put the drafts in extradoc/blog/draft | 22:42 |
| stakkars | ah, good! | 22:42 |
| jterrace | Alex_Gaynor: fijal: is there a reason that numpypy.int8-int64 are exposed but not uint8-uint64? | 22:43 |
| fijal | jterrace: alex forgot | 22:44 |
| jterrace | i dont see tests for uints either? | 22:44 |
| jterrace | the boxes are there though | 22:44 |
| tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 244 seconds | 22:44 | |
| fijal (~fijal@41.52.94.74) left irc: Read error: Connection reset by peer | 22:45 | |
| `fox` (~fox@host91-171-dynamic.245-95-r.retail.telecomitalia.it) joined #pypy. | 22:48 | |
| Nick change: wulczer -> wulczer-afk | 22:49 | |
| tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy. | 22:59 | |
| rokujyouhitoma | I am staying San Francisco. Does someone want to meet me? | 23:00 |
| e1ven (~E1ven@SQ7/ProjectLead/E1ven) joined #pypy. | 23:04 | |
| e1ven | Hey; I'm trying to donate at http://pypy.org/py3donate.html, but I'm getting an error ;() | 23:04 |
| Niedar (gfghfghf@host-173-230-2-133.vanodod.clients.pavlovmedia.com) joined #pypy. | 23:04 | |
| Alex_Gaynor | jterrace: I was exposing them as they were pointed out to me/someone gave me code that broke, I add tests before exposing them (even though it's only one line), so it's not just a bulk thing | 23:05 |
| jterrace | i see | 23:05 |
| jterrace | ill add them | 23:05 |
| jterrace | working on fleshing out numpypy.fromstring and found that uint8 didnt work | 23:06 |
| Alex_Gaynor | see test_dtypes.py for the tests | 23:06 |
| jterrace | yeah i saw the int8-64 tests | 23:06 |
| derdon (~derdon@p5DE8A137.dip.t-dialin.net) joined #pypy. | 23:10 | |
| antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds | 23:11 | |
| jterrace (~jterrace@aegis.CS.Princeton.EDU) left irc: Quit: Leaving. | 23:11 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/738 | 23:11 |
| rokujyouhitoma | so...we use rpython on work. we use it to make a some tool. | 23:11 |
| rokujyouhitoma | but, we dont develop programming language. we develop some command line tools. | 23:12 |
| `fox` (~fox@host91-171-dynamic.245-95-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds | 23:12 | |
| _pyre_ (~stian@245.232.189.109.customer.cdi.no) left irc: Ping timeout: 252 seconds | 23:22 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy. | 23:29 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host | 23:29 | |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 23:29 | |
| amaury_ | fijal is evil | 23:33 |
| amaury_ | with his pypy-embedding-cpython | 23:33 |
| Alex_Gaynor | amaury_: it's awesome :) | 23:33 |
| amaury_ | yes, similar to the remote CPython that was tried several years ago | 23:34 |
| amaury_ | just closer | 23:34 |
| amaury_ | ATM I'm trying with Martijn's idea | 23:36 |
| amaury_ | 2 object space in the same translation | 23:36 |
| Alex_Gaynor | amaury_: since objspace is a PBC, doesn't that break stuff? | 23:36 |
| amaury_ | well, you have 2 PBCs then | 23:36 |
| Alex_Gaynor | ah, os it's specifically 2k and 3k, not N objspace | 23:37 |
| amaury_ | yes | 23:37 |
| stakkars | what is the idea of pypy-embedding-cpython? | 23:37 |
| gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated! | 23:37 | |
| amaury_ | run numpy inside pypy :-) | 23:37 |
| amaury_ | Alex_Gaynor: I have to disable cpyext though | 23:38 |
| amaury_ | for the same reason as fijal's | 23:38 |
| Alex_Gaynor | amaury_: sure | 23:38 |
| amaury_ | Windows would not have this issue | 23:38 |
| stakkars | hmm, don't see the advantage but for tests | 23:38 |
| amaury_ | stakkars: you can share a memory buffer between pypy and numpy | 23:39 |
| amaury_ | so you can fill an array in pypy, and have numpy compute with it | 23:40 |
| Alex_Gaynor | amaury_: well, you don't want numpy to compute with it, because its slow, but you can use it for rendering charts and stuff with matplotlib | 23:40 |
| amaury_ | 2 objspaces in pypy should be doable | 23:47 |
| Alex_Gaynor | amaury_: that doesn't answer the semantic questions though | 23:47 |
| amaury_ | I'm currently chasing self.space out of PBC objects | 23:47 |
| Alex_Gaynor | I knew space arguments were better than self.space :) | 23:47 |
| amaury_ | so do I | 23:48 |
| amaury_ | I removed them from the Function class | 23:48 |
| amaury_ | I think it's ok to have it stored in Frames | 23:48 |
| amaury_ | Alex_Gaynor: if you have 2 object spaces, many things are possible | 23:51 |
| Alex_Gaynor | I like making thigns possible | 23:52 |
| amaury_ | it would be possible that a single object is seen as a bytes string by one space | 23:52 |
| amaury_ | and seen as unicode by the other | 23:52 |
| Alex_Gaynor | that'd certainly be interesting | 23:52 |
| amaury_ | argh, default arguments of interplevel functions are wrapped | 23:56 |
| Alex_Gaynor | yup, why is that an issue? oh it requires a space :/ | 23:57 |
| amaury_ | yes, so I have two functions that are not identical | 23:57 |
| Alex_Gaynor | just make it NoneNotWrapped and deal with it in the function? | 23:58 |
| amaury_ | too many of them :/ | 23:58 |
| Alex_Gaynor | :/ | 23:58 |
| amaury_ | store the default values in the space | 23:59 |
| tilgovi (~randall@couchdb/developer/tilgovi) left irc: Read error: Connection reset by peer | 23:59 | |
| tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy. | 23:59 | |
| --- Fri Dec 9 2011 | 00:00 | |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!
