| tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection | 00:25 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/12 | 00:58 |
|---|---|---|
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/184 | 01:01 |
| rguillebert (~rguillebe@fec76-1-78-234-124-105.fbx.proxad.net) left irc: Ping timeout: 268 seconds | 01:05 | |
| etrepum (~bob@pool-173-73-20-220.washdc.fios.verizon.net) left irc: Quit: etrepum | 01:11 | |
| rguillebert (~rguillebe@2a01:e34:eea7:c690:21f:c6ff:fe12:4dee) joined #pypy. | 01:17 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/756 | 01:19 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1901 | 01:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/651 | 01:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/317 | 01:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/307 | 01:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/656 | 01:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1172 | 01:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/839 | 01:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1491 | 01:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/765 | 01:30 |
| leo2007 (leo@123.123.254.37) left #pypy ("rcirc on GNU Emacs 23.3.50.1"). | 01:54 | |
| derdon (~derdon@p5DE8BF19.dip.t-dialin.net) left irc: Remote host closed the connection | 01:55 | |
| justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Ping timeout: 240 seconds | 01:58 | |
| overmind_ (~overminde@wf100-253.ust.hk) joined #pypy. | 02:26 | |
| nettok (~quassel@190.143.180.74) joined #pypy. | 02:36 | |
| etrepum (~bob@pool-173-73-20-220.washdc.fios.verizon.net) joined #pypy. | 02:36 | |
| Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Ping timeout: 240 seconds | 02:36 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/993 | 02:46 |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 02:55 | |
| etrepum (~bob@pool-173-73-20-220.washdc.fios.verizon.net) left irc: Quit: etrepum | 03:09 | |
| CIA-53 | 03mattip 07roundup * 10#691/pypy-head does not compile under Windows: Can this be closed? Nightly builds consistently succeed on the buildbot * 14https://bugs.pypy.org/issue691 | 03:12 |
| CIA-53 | 03mikefc 07roundup * 10#972/inconsistence with cpython (datetime?): | 03:13 |
| CIA-53 | (just to show that using cpython directly produces the same result as ipython) | 03:13 |
| CIA-53 | ================ running directly on python2.7.2 (macports) | 03:13 |
| CIA-53 | Pyt ... * 14https://bugs.pypy.org/issue972 | 03:13 |
| verte-{nyctea} (~verte@python/site-packages/verte) joined #pypy. | 03:13 | |
| CIA-53 | 03mattip 07roundup * 10#741/nightly build for windows: [chatting] Nightly is consistently building on win32 now * 14https://bugs.pypy.org/issue741 | 03:18 |
| Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 248 seconds | 03:24 | |
| Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy. | 03:30 | |
| Ademan (~dan@adsl-71-141-224-79.dsl.snfc21.pacbell.net) joined #pypy. | 03:33 | |
| mikefc | ugh. "Too many open files in system" | 03:39 |
| mikefc | files are being read with: return open(filename, 'r').read() | 03:40 |
| mikefc | is the GC not closing these straight away? | 03:40 |
| verte-{nyctea} | it isn't, no. | 03:40 |
| mikefc | (and there are quite a few thousand files) | 03:40 |
| verte-{nyctea} | the GC may never close a file, that's why we have the with statement. | 03:40 |
| mikefc | does the 'with' statement really fit in the context of "read a file and return its contents" ? | 03:41 |
| verte-{nyctea} | of course it does | 03:41 |
| Alex_Gaynor | with open(filename, 'r') as f: return f.read() | 03:41 |
| mikefc | vert/alex: thanks. seems a little obtuse. | 03:42 |
| verte-{nyctea} | what python tutorial did you read? | 03:42 |
| mikefc | i've never read a tutorial on 'with'. never used it before. | 03:43 |
| verte-{nyctea} | it hasn't really been considered a good idea to *not* close files for six years or so. | 03:43 |
| mikefc | vert: fair enough. | 03:44 |
| mikefc | it does seem a little counter to the way python handles everything else. | 03:45 |
| verte-{nyctea} | everything else? | 03:45 |
| mikefc | not having to expressly handle things that fall out of scope. | 03:45 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1901 | 03:48 |
| etrepum (~bob@pool-173-73-20-220.washdc.fios.verizon.net) joined #pypy. | 03:50 | |
| verte-{nyctea} | the best resource for this is probably http://old.nabble.com/small-notes-re%3A-waterken-to31045687.html#a31085563 | 03:53 |
| etrepum (~bob@pool-173-73-20-220.washdc.fios.verizon.net) left irc: Ping timeout: 240 seconds | 03:55 | |
| mikefc | verte-{nyctea}: thanks for the link. | 03:59 |
| mcdonc_ (~mcdonc@ip72-209-213-54.dc.dc.cox.net) joined #pypy. | 04:05 | |
| mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) left irc: Ping timeout: 240 seconds | 04:06 | |
| justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) joined #pypy. | 04:10 | |
| stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy. | 04:12 | |
| stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 04:12 | |
| Nick change: stakkars_ -> stakkars | 04:12 | |
| anish_ (anish@nat/redhat/x-esxoxlzyybgoyzhr) joined #pypy. | 04:14 | |
| stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy. | 04:15 | |
| stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 04:15 | |
| Nick change: stakkars_ -> stakkars | 04:15 | |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 276 seconds | 04:20 | |
| tumbleweed_ (~stefanor@corelli.tumbleweed.org.za) joined #pypy. | 04:21 | |
| jell_ (~jell@host-134-23.dmtec.eu) joined #pypy. | 04:21 | |
| anish_ (anish@nat/redhat/x-esxoxlzyybgoyzhr) got netsplit. | 04:26 | |
| jell (~jell@host-134-23.dmtec.eu) got netsplit. | 04:26 | |
| tumbleweed (~stefanor@ubuntu/member/tumbleweed) got netsplit. | 04:26 | |
| Nick change: tumbleweed_ -> tumbleweed | 04:26 | |
| Possible future nick collision: tumbleweed | 04:26 | |
| tumbleweed (~stefanor@corelli.tumbleweed.org.za) left irc: Changing host | 04:26 | |
| tumbleweed (~stefanor@ubuntu/member/tumbleweed) joined #pypy. | 04:26 | |
| espes (~espes@123-243-221-248.tpgi.com.au) left irc: Ping timeout: 244 seconds | 04:26 | |
| justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Quit: Leaving | 04:30 | |
| anish_ (anish@nat/redhat/x-bcwrwiorcbgkfmyd) joined #pypy. | 04:32 | |
| espes (espes@2001:470:1f05:ca1::) joined #pypy. | 04:33 | |
| jell (~jell@host-134-23.dmtec.eu) got lost in the net-split. | 04:37 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1491 | 04:51 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/839 | 04:55 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/656 | 05:11 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1172 | 05:12 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/651 | 05:18 |
| overmind_ (~overminde@wf100-253.ust.hk) left irc: Ping timeout: 252 seconds | 05:52 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/317 | 06:07 |
| overmind_ (~overminde@wf100-253.ust.hk) joined #pypy. | 06:10 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/765 | 06:18 |
| aleksi (~aleksi@85.235.191.82) joined #pypy. | 06:22 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 06:26 | |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy | 06:26 | |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy. | 06:27 | |
| fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Client Quit | 06:27 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Ping timeout: 244 seconds | 06:30 | |
| unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) left irc: Quit: unbit | 06:35 | |
| unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) joined #pypy. | 06:36 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 06:48 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Remote host closed the connection | 06:49 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 06:53 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Client Quit | 06:57 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 07:05 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/307 | 07:06 |
| Nick change: elliottcable -> coob | 07:22 | |
| overmind_ (~overminde@wf100-253.ust.hk) left irc: Ping timeout: 268 seconds | 07:27 | |
| overmind_ (~overminde@wf118-245.ust.hk) joined #pypy. | 07:28 | |
| nettok (~quassel@190.143.180.74) left irc: Remote host closed the connection | 07:29 | |
| `fox` (~fox@host139-108-dynamic.245-95-r.retail.telecomitalia.it) joined #pypy. | 07:46 | |
| fijal (~fijal@196-210-199-8.dynamic.isadsl.co.za) joined #pypy. | 08:14 | |
| espes (espes@2001:470:1f05:ca1::) left irc: Read error: Connection timed out | 08:15 | |
| espes (espes@2001:470:1f05:ca1::) joined #pypy. | 08:15 | |
| antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy. | 08:23 | |
| fijal | antocuni: hi anto | 08:26 |
| fijal | antocuni: want to know something funny? | 08:26 |
| Alex_Gaynor | fijal: why locals() forces is sort of obviuos, you need to populate frame object with locals | 08:27 |
| Alex_Gaynor | I guess you could do it if you unrolled every single loop | 08:27 |
| fijal | is it? | 08:28 |
| verte-{nyctea} | Alex_Gaynor: except that locals() returns a copy when the stack is fast | 08:28 |
| azanella (~azanella@189.6.80.207) joined #pypy. | 08:28 | |
| Alex_Gaynor | verte-{nyctea}: sure, but you're still reading from the frame object, which means it needs to be allocated for the JIT | 08:28 |
| fijal | Alex_Gaynor: no, I don't think it's that simple | 08:30 |
| fijal | probably if you unroll fast2locals it won't do that any more | 08:30 |
| Alex_Gaynor | fijal: it calls ec.topframe_nohidden() IIRC | 08:30 |
| Alex_Gaynor | that has a loop which isn't unrolled | 08:30 |
| fijal | that should just return the virtualizable | 08:30 |
| Alex_Gaynor | well there's a loop | 08:31 |
| Alex_Gaynor | maybe that loop should be unrolled, I dunno | 08:31 |
| fijal | it's not clear to me though that you should necesarilly escape the virtualizable | 08:31 |
| fijal | Alex_Gaynor: seen my tweets btw? | 08:32 |
| Alex_Gaynor | link? | 08:32 |
| fijal | http://paste.pocoo.org/show/525839/ | 08:32 |
| fijal | . is a passing test from test_numarray.py | 08:32 |
| fijal | run on pypy | 08:32 |
| fijal | and ABORT is well, an aborted loop | 08:32 |
| Alex_Gaynor | yes I saw | 08:32 |
| fijal | does not look too rosy | 08:32 |
| Alex_Gaynor | calling gettopframe_nohidden always forces frame | 08:33 |
| verte-{nyctea} | why? | 08:33 |
| Alex_Gaynor | because it isn't inlined and it accesses frame attrs | 08:34 |
| Alex_Gaynor | fijal: fast2locals has an explicit jit.dont_look_inside | 08:35 |
| fijal | Alex_Gaynor: and a loop | 08:37 |
| fijal | for good measure | 08:37 |
| kenaan | 12fijal default 1145a754407e3d 15/pypy/interpreter/eval.py: remove nonsense hints (those functions contain loops) | 08:38 |
| kenaan | 12fijal default 1194934ddf3ae9 15/pypy/interpreter/gateway.py: remove some nonsense, handle_exception always raises | 08:38 |
| fijal | Alex_Gaynor: it's not as you think it is | 08:38 |
| Alex_Gaynor | fijal: I et you this doesn't translate anymore | 08:38 |
| Alex_Gaynor | fijal: it's not as what as I think it? | 08:39 |
| Alex_Gaynor | what's it? | 08:39 |
| fijal | frame forcing | 08:51 |
| fijal | why wouldn't it translate any more? | 08:51 |
| Alex_Gaynor | because there are loops which access frame directly | 08:52 |
| fijal | you mean the dont_look_inside? | 08:52 |
| fijal | but policy will remove it anyway | 08:52 |
| fijal | because they have loops | 08:52 |
| fijal | so no harm done | 08:53 |
| fijal | IMO | 08:53 |
| fijal | Alex_Gaynor: btw, my jit hooks work | 08:53 |
| fijal | you can put a pdb each time tracing is aborted | 08:53 |
| Alex_Gaynor | fijal: try to translate | 08:53 |
| Alex_Gaynor | I bet you it fails | 08:53 |
| Alex_Gaynor | fijal: cool | 08:53 |
| fijal | I'm working on making greenkey available | 08:53 |
| fijal | because you don't know where it started | 08:53 |
| fijal | but we need a plan for pypy tests | 08:54 |
| fijal | my best bet so far is with --jit inlining=0 | 08:54 |
| fijal | 40s vs cpython 27s | 08:54 |
| fijal | for test_numarray.py | 08:54 |
| fijal | there are quite a few bad loops as well | 08:55 |
| Alex_Gaynor | what is a bad loop? | 08:57 |
| fijal | btw, I made myself a website http://baroquesoftware.com | 08:57 |
| fijal | Alex_Gaynor: something that was unused at the time I created better jit hooks branch :) | 08:57 |
| fijal | so no idea | 08:57 |
| fijal | "ask hakan" | 08:57 |
| Alex_Gaynor | fijal: https://bitbucket.org/pypy/pypy/changeset/d4257b46227b and https://bitbucket.org/pypy/pypy/changeset/abce3df08d92 | 08:58 |
| CIA-53 | 03fijal 07roundup * 10#691/pypy-head does not compile under Windows: [resolved] Indeed, closing * 14https://bugs.pypy.org/issue691 | 08:59 |
| Alex_Gaynor | I can't find the issue fo rit though | 08:59 |
| Count_Niedar (~nnscript@ip68-231-248-182.hr.hr.cox.net) left irc: Ping timeout: 240 seconds | 09:00 | |
| hakanardo | Alex_Gaynor: are you talking about the InvalidLoop exception? | 09:01 |
| Alex_Gaynor | hakanardo: Dunno, ask fijal :) | 09:01 |
| hakanardo | :) | 09:02 |
| fijal | hakanardo: no, ABORT_BAD_LOOP | 09:02 |
| fijal | there is an if in pyjitpl.py | 09:02 |
| fijal | which I don't completely understand | 09:02 |
| fijal | anyway, breakfast/lunch for me | 09:02 |
| Niedar (~nnscript@ip68-231-248-182.hr.hr.cox.net) joined #pypy. | 09:03 | |
| hakanardo | ok, that's about retracing finding an unexpected loop | 09:03 |
| hakanardo | it could be handled better now that optimizeopt supprot mutiple intermeditae labels | 09:04 |
| azanella (~azanella@189.6.80.207) left irc: Quit: Ex-Chat | 09:05 | |
| hakanardo | 44b0e2106e2d wont work, you need alias if you emit SAME_AS there | 09:07 |
| hakanardo | otherwise you'll get an key error when if the box is actually imported | 09:07 |
| phrearch (~phrearch_@5ED42DE2.cm-7-5a.dynamic.ziggo.nl) left irc: Remote host closed the connection | 09:10 | |
| hakanardo | ah, no you did not revert close_loop so the SAME_AS is now considered part of the loop and thus the original boxs get imported? | 09:12 |
| Niedar (~nnscript@ip68-231-248-182.hr.hr.cox.net) left irc: Ping timeout: 252 seconds | 09:12 | |
| Niedar (~nnscript@ip68-231-248-182.hr.hr.cox.net) joined #pypy. | 09:13 | |
| azanella (~azanella@189.6.80.207) joined #pypy. | 09:15 | |
| fijal | hakanardo: I have a bit no idea :) | 09:30 |
| hakanardo | I'm making a test to be sure, but I think it should work as it is | 09:30 |
| anish_ (anish@nat/redhat/x-bcwrwiorcbgkfmyd) left irc: Ping timeout: 240 seconds | 09:32 | |
| fijal | it certainly works better now than it did before | 09:32 |
| hakanardo | good, what triggered the bug? | 09:35 |
| fijal | virtualenv -p pypy | 09:39 |
| anish_ (anish@nat/redhat/x-isusnedsholksefw) joined #pypy. | 09:49 | |
| Nick change: jell_ -> jell | 09:54 | |
| Nick change: jell -> Guest40201 | 09:54 | |
| anish_ (anish@nat/redhat/x-isusnedsholksefw) left irc: Ping timeout: 252 seconds | 09:57 | |
| kenaan | 12hakanardo default 11fff6b491e07d 15/pypy/jit/metainterp/optimizeopt/test/: test for 05c2089f5545, it curently fails | 09:58 |
| anish_ (anish@nat/redhat/x-yutscfnqiebssvdb) joined #pypy. | 10:02 | |
| Nick change: Guest40201 -> jell | 10:06 | |
| `fox` (~fox@host139-108-dynamic.245-95-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds | 10:09 | |
| kkris (~kris@188-22-125-82.adsl.highway.telekom.at) joined #pypy. | 10:10 | |
| kenaan | 12hakanardo default 11660488a0ae41 15/pypy/jit/metainterp/optimizeopt/: Since alias tracking is killed the name of the imported box among the label arguments will be the name from the... | 10:10 |
| squiddy (~squiddy@46.128.142.118.dynamic.cablesurf.de) joined #pypy. | 10:13 | |
| fijal | hakanardo: thanks | 10:18 |
| hakanardo | np, it was my fault | 10:18 |
| phrearch (~phrearch_@5ED42DE2.cm-7-5a.dynamic.ziggo.nl) joined #pypy. | 10:20 | |
| phrearch | morning | 10:20 |
| phrearch | whats the correct way again to make a virtualenv with pypy? | 10:20 |
| phrearch | i tried virtualenv -p /usr/bin/pypy env | 10:20 |
| fijal | phrearch: virtualenv -p <path to pypy> | 10:20 |
| fijal | ? | 10:20 |
| fijal | and? | 10:20 |
| fijal | hakanardo: I'm thinking how to make jitviewer work | 10:21 |
| fijal | because obviously right now the labels don't correspond to loop names any more | 10:21 |
| phrearch | http://paste.pocoo.org/show/526004/ | 10:21 |
| hakanardo | arg, translation broke: http://paste.pocoo.org/show/526004/ | 10:21 |
| Alex_Gaynor | hakanardo: yes, fijal broke it and I told him ;) | 10:21 |
| phrearch | not sure, maybe i have an old virtualenv | 10:21 |
| fijal | phrearch: can you run pypy-c? | 10:21 |
| Alex_Gaynor | phrearch: what version of venv do you have? | 10:21 |
| phrearch | fijal: nope | 10:22 |
| Alex_Gaynor | you need 1.6.1 or newer | 10:22 |
| phrearch | 1.5.1 | 10:22 |
| fijal | phrearch: or pypy | 10:22 |
| phrearch | and pypy 1.7 | 10:22 |
| fijal | hakanardo: paste? | 10:22 |
| fijal | phrearch: you need 1.6.1 or newer | 10:22 |
| phrearch | fijal: i see. everything works fine now. thanks :D | 10:22 |
| `fox` (~fox@host139-108-dynamic.245-95-r.retail.telecomitalia.it) joined #pypy. | 10:23 | |
| hakanardo | fijal: http://paste.pocoo.org/show/526003/ | 10:23 |
| fijal | hakanardo: what broke? | 10:23 |
| fijal | Alex_Gaynor: this is not broken how you told me it would be :-P | 10:23 |
| Alex_Gaynor | fijal: heh, indeed, double no cookie for you apparently :) | 10:24 |
| phrearch | does anyone managed to get zope working on pypy by the way? | 10:24 |
| phrearch | im curious whether plone can run in pypy | 10:24 |
| fijal | pffff | 10:24 |
| fijal | phrearch: probably not, ZODB is a C extension | 10:25 |
| phrearch | fijal: ok thanks | 10:25 |
| kenaan | 12fijal default 1121ca3b5a40aa 15/pypy/interpreter/gateway.py: backout 94934ddf3ae9, it's nonsense, but annotator does not understand it and I'm too lazy to convince it | 10:27 |
| fijal | Alex_Gaynor: fixed | 10:28 |
| fijal | hakanardo: ^^^ | 10:28 |
| hakanardo | thx | 10:28 |
| fijal | annotator is stupid | 10:29 |
| Alex_Gaynor | fijal: no annotator is correct here | 10:30 |
| Alex_Gaynor | if e was not one of those types the function would return None | 10:30 |
| Alex_Gaynor | and not raise | 10:30 |
| Alex_Gaynor | it's not annotated as "can never return" | 10:30 |
| Alex_Gaynor | (the annotator supports such a thing IIRC) | 10:30 |
| fijal | no | 10:30 |
| fijal | it's not "return e" | 10:30 |
| fijal | it's raise e | 10:30 |
| Alex_Gaynor | oh pff | 10:30 |
| fijal | so it's either e or any of those wrapped things | 10:30 |
| Alex_Gaynor | yes indeed that function is really "can only raise" | 10:31 |
| fijal | yes | 10:31 |
| fijal | besides, raise None is not that valid | 10:31 |
| Alex_Gaynor | isn't it the same as "raise;"? | 10:31 |
| fijal | TypeError: exceptions must be old-style classes or derived from BaseException, not NoneType | 10:31 |
| fijal | no it's not | 10:31 |
| Alex_Gaynor | nvm, it's a TypeError | 10:31 |
| fijal | so it kinda works by chance | 10:32 |
| fijal | besides, it sometimes forces the virtualizable | 10:32 |
| fijal | (why???) | 10:32 |
| Alex_Gaynor | what forces the virutalizable? | 10:32 |
| fijal | as soon as I'm out of blackhole I'm on line 641 of error.py | 10:33 |
| fijal | so I have no clue | 10:33 |
| fijal | one of those constructs | 10:33 |
| Alex_Gaynor | eh, which error.py | 10:33 |
| fijal | interpreter/error.py | 10:33 |
| Alex_Gaynor | its only 465 lines | 10:33 |
| phrearch | is there still a seperate simplejson branch for pypy? | 10:34 |
| fijal | uh no | 10:35 |
| fijal | gateway.py, sorry | 10:35 |
| phrearch | or whats the best json encoder out there atm? | 10:35 |
| fijal | phrearch: the builtin json encoder is good | 10:35 |
| verte-{nyctea} | phrearch: the branch was for improving the performance | 10:35 |
| fijal | import json | 10:35 |
| fijal | this one | 10:35 |
| phrearch | aha ok thanks | 10:35 |
| fijal | phrearch: the decoder is still slow though | 10:35 |
| Alex_Gaynor | fijal: this is running tests? you end up at 641 there on py.py? | 10:36 |
| fijal | phrearch: it's about 1-2 days of work, I would welcome bids :) | 10:36 |
| fijal | Alex_Gaynor: yes :/ | 10:36 |
| fijal | with a forced virtualizable | 10:36 |
| fijal | precisely | 10:36 |
| fijal | wtf | 10:36 |
| Alex_Gaynor | fijal: the comment says it isn't possible | 10:36 |
| fijal | surprise | 10:36 |
| fijal | the comment had no tests :) | 10:36 |
| fijal | Alex_Gaynor: the tool is great, I spent half of the yesterday thinking "Wtf" | 10:36 |
| Alex_Gaynor | fijal: we apparently need yet another tool to explain the wtf | 10:37 |
| phrearch | fijal: yea that would be very useful indeed | 10:37 |
| fijal | Alex_Gaynor: yes | 10:38 |
| fijal | phrearch: well, I'm willing to get hired :) | 10:38 |
| fijal | but I don't use json myself | 10:38 |
| fijal | anyway, /me off briefly | 10:38 |
| fijal | phrearch: I even made a website that states that - baroquesoftware.com | 10:38 |
| phrearch | fijal: ill ask around. there are more companies/people that would like to see this happen i imagine | 10:39 |
| phrearch | fijal: got it. thanks | 10:39 |
| espes (espes@2001:470:1f05:ca1::) left irc: Read error: Connection timed out | 10:40 | |
| espes (espes@2001:470:1f05:ca1::) joined #pypy. | 10:41 | |
| verte-{nyctea} | fijal: that's an awesome t-shirt | 10:41 |
| verte-{nyctea} | where'd you get it? | 10:42 |
| Alex_Gaynor | verte-{nyctea}: that looks liek the pycon 2011 tshirt | 10:43 |
| verte-{nyctea} | cool, I should go sometime, maybe 2013 | 10:45 |
| `fox` (~fox@host139-108-dynamic.245-95-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds | 10:57 | |
| stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy. | 11:16 | |
| stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 11:16 | |
| Nick change: stakkars_ -> stakkars | 11:16 | |
| derdon (~derdon@pD9E1DB40.dip.t-dialin.net) joined #pypy. | 11:20 | |
| stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy. | 11:24 | |
| stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 11:24 | |
| Nick change: stakkars_ -> stakkars | 11:24 | |
| arigato (~arigo@xdsl-188-155-166-12.adslplus.ch) joined #pypy. | 11:32 | |
| [mat^2] (mathias@87.55.0.198) left irc: Ping timeout: 240 seconds | 11:35 | |
| etrepum (~bob@pool-173-73-20-220.washdc.fios.verizon.net) joined #pypy. | 11:36 | |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 11:40 | |
| etrepum (~bob@pool-173-73-20-220.washdc.fios.verizon.net) left irc: Quit: etrepum | 11:43 | |
| anish_ (anish@nat/redhat/x-yutscfnqiebssvdb) left irc: Quit: Leaving | 12:00 | |
| kenaan | 12arigo concurrent-marksweep 1105769b6b553e 15/pypy/rpython/memory/gc/: Document and start implementing an idea to make major collection parallel. | 12:21 |
| kenaan | 12arigo concurrent-marksweep 11a0048b726e62 15/: hg merge default | 12:21 |
| kenaan | 12arigo concurrent-marksweep 11130f4c04e5c4 15/pypy/: hg merge default | 12:21 |
| `fox` (~fox@host139-108-dynamic.245-95-r.retail.telecomitalia.it) joined #pypy. | 12:27 | |
| davidpk (~davidpk@5e09151b.bb.sky.com) joined #pypy. | 13:03 | |
| CIA-53 | 03arigo 07roundup * 10#972/inconsistence with cpython (datetime?): | 13:06 |
| CIA-53 | It occurs also on CPython using datetime.py instead of the extension module | 13:06 |
| CIA-53 | datetime.so, so it's really a difference between the two. Anyone int ... * 14https://bugs.pypy.org/issue972 | 13:06 |
| jyp (~interrupt@175.197.245.105) joined #pypy. | 13:08 | |
| jyp (~interrupt@175.197.245.105) left irc: Remote host closed the connection | 13:20 | |
| tumbleweed | is this normal? http://paste.debian.net/150189/ (I can't say that I care about clr failures when I'm not building for CLR) | 13:44 |
| asmeurer (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy. | 13:58 | |
| JaRoel|4d (~jaroel|4d@178-85-196-53.dynamic.upc.nl) joined #pypy. | 13:58 | |
| squiddy (~squiddy@46.128.142.118.dynamic.cablesurf.de) left irc: Remote host closed the connection | 14:01 | |
| davidpk (~davidpk@5e09151b.bb.sky.com) left irc: Quit: davidpk | 14:07 | |
| kenaan | 12arigo default 11d373deb39b4d 15/pypy/translator/generator.py: Simplify the graph before turning it into a generator, to get rid of the extra variables passed around. | 14:08 |
| kenaan | 12arigo default 11c9706c6e6914 15/pypy/rpython/test/test_generator.py: A failing test. | 14:08 |
| kenaan | 12arigo default 11cce139c7a9c6 15/pypy/translator/generator.py: Fix the test. | 14:08 |
| davidpk (~davidpk@5e09151b.bb.sky.com) joined #pypy. | 14:10 | |
| davidpk (~davidpk@5e09151b.bb.sky.com) left irc: Read error: Connection reset by peer | 14:15 | |
| davidpk (~davidpk@5e09151b.bb.sky.com) joined #pypy. | 14:15 | |
| verte-{nyctea} (~verte@python/site-packages/verte) left irc: Quit: ~~~ Crash in JIT! | 14:16 | |
| WildChild (~wildchild@modemcable123.12-202-24.mc.videotron.ca) joined #pypy. | 14:20 | |
| ecm (~lxnch@unaffiliated/lxnch) joined #pypy. | 14:34 | |
| mat^2 (~mathias@87.55.0.198) joined #pypy. | 14:35 | |
| lambacck (~chris@d24-150-124-118.home.cgocable.net) left irc: Quit: lambacck | 14:38 | |
| EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) left irc: Ping timeout: 255 seconds | 14:40 | |
| ecm (lxnch@unaffiliated/lxnch) left #pypy. | 14:45 | |
| Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy. | 14:52 | |
| CIA-53 | 03mattip 07roundup * 10#959/numpypy array operation improperly jitted away: | 14:54 |
| CIA-53 | Latest nightly 44b0e2106e2d runs up to 24000 iterations on linux 32 no longer | 14:54 |
| CIA-53 | exhibits this behaviour, is it fixed? * 14https://bugs.pypy.org/issue959 | 14:54 |
| lambacck (~chris@d24-150-124-118.home.cgocable.net) joined #pypy. | 15:04 | |
| _aleksi (~aleksi@85.235.191.82) joined #pypy. | 15:05 | |
| kenaan | 12arigo extradoc 119091b9fd97f8 15/sprintinfo/leysin-winter-2012/announcement.txt: Add Leysin draft announcement | 15:06 |
| aleksi (~aleksi@85.235.191.82) left irc: Ping timeout: 244 seconds | 15:08 | |
| kenaan | 12arigo extradoc 11fc335a8ec80e 15/sprintinfo/leysin-winter-2012/people.txt: Add the standard people.txt | 15:09 |
| ltratt (~ltratt@ntratt.plus.com) joined #pypy. | 15:16 | |
| kenaan | 12arigo extradoc 111eeeeaed6a6d 15/sprintinfo/leysin-winter-2012/announcement.txt: Add some topics. | 15:16 |
| arigato | fijal, antocuni: ^^^ feel free to add more topics if you want | 15:16 |
| arigato | fijal: should we add e.g. jitviewer as a general topic? | 15:17 |
| JaRoel|4d (~jaroel|4d@178-85-196-53.dynamic.upc.nl) left irc: Remote host closed the connection | 15:19 | |
| mat^2 (~mathias@87.55.0.198) left irc: Ping timeout: 240 seconds | 15:21 | |
| kenaan | 12hakanardo jit-improve-nested-loops 11f75c6c5a133a 15/pypy/: hg merge default | 15:27 |
| kenaan | 12hakanardo jit-improve-nested-loops 11be85523c5872 15/pypy/jit/metainterp/test/test_loop.py: fix test | 15:27 |
| kenaan | 12hakanardo jit-improve-nested-loops 115706bc85eb3b 15/pypy/jit/metainterp/compile.py: more generally ensure that TargetTokens of labels has the correct original_jitcell_token | 15:27 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/766 [12hakanardo, jit-improve-nested-loops] | 15:28 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/652 [12hakanardo, jit-improve-nested-loops] | 15:28 |
| JaRoel|4d (~jaroel|4d@178-85-196-53.dynamic.upc.nl) joined #pypy. | 15:28 | |
| Kaskuka (~John@osbk-4db16f9b.pool.mediaWays.net) left irc: Ping timeout: 240 seconds | 15:40 | |
| Kaskuka (~John@osbk-4db177c4.pool.mediaWays.net) joined #pypy. | 15:40 | |
| kenaan | 12arigo default 11f9932c00e2d2 15/pypy/objspace/std/stringobject.py: Fix comment, and simplify the final logic. | 15:41 |
| mat^2 (~mathias@87.55.0.198) joined #pypy. | 15:47 | |
| davidpk (~davidpk@5e09151b.bb.sky.com) left irc: Quit: davidpk | 15:49 | |
| _aleksi (~aleksi@85.235.191.82) left irc: Remote host closed the connection | 15:49 | |
| mat^2 (~mathias@87.55.0.198) left irc: Ping timeout: 252 seconds | 15:52 | |
| etrepum (~bob@12.130.118.25) joined #pypy. | 16:07 | |
| `fox` (~fox@host139-108-dynamic.245-95-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer | 16:10 | |
| `fox` (~fox@host139-108-dynamic.245-95-r.retail.telecomitalia.it) joined #pypy. | 16:10 | |
| etrepum (~bob@12.130.118.25) left irc: Quit: etrepum | 16:18 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/766 [12hakanardo, jit-improve-nested-loops] | 16:28 |
| overmind_ (~overminde@wf118-245.ust.hk) left irc: Ping timeout: 268 seconds | 16:42 | |
| fijal | arigato: well, yes, but I', not there | 16:46 |
| arigato | sure | 16:46 |
| fijal | tumbleweed: it should not fail but indeed, you should not care | 16:47 |
| fijal | arigato: can you explain why 94934ddf3ae9 broke translation? | 16:51 |
| fijal | note that handle_exception is an always raising function | 16:52 |
| arigato | ah, that's a flow space issue | 16:53 |
| arigato | w_result is not defined afterwards | 16:53 |
| fijal | it's very confusing because handle_exception is always raising, so raise is removed | 16:54 |
| arigato | yes | 16:55 |
| arigato | unsure what to do about it, though | 16:55 |
| fijal | well, I can think about w_result = None | 16:55 |
| fijal | somewhere | 16:55 |
| arigato | yes, that's one solution | 16:55 |
| davidpk (~davidpk@5e09151b.bb.sky.com) joined #pypy. | 16:55 | |
| arigato | alternatively you can write "assert 0 # should always raise" after self.handle_exception() | 16:56 |
| arigato | but that may be a bit heavy | 16:56 |
| kkris (~kris@188-22-125-82.adsl.highway.telekom.at) left irc: Quit: Leaving. | 16:58 | |
| fijal | it's also copy-pasted code :/ | 16:58 |
| fijal | arigato: another interesting observation | 17:02 |
| fijal | I was running pypy py.test module/micronumpy/test/test_numarray | 17:03 |
| fijal | observing aborts | 17:03 |
| fijal | sometimes I get an abort on line 641 in gateway.py | 17:03 |
| fijal | the comment is probably wrong | 17:03 |
| fijal | this is after I'm done blackholing (to have a consistent interpreter state) | 17:04 |
| `fox` (~fox@host139-108-dynamic.245-95-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds | 17:12 | |
| nettok (~quassel@200.119.162.209) joined #pypy. | 17:21 | |
| fprimex (fprimex@80.sub-75-252-40.myvzw.com) joined #pypy. | 17:25 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/652 [12hakanardo, jit-improve-nested-loops] | 17:29 |
| arigato (~arigo@xdsl-188-155-166-12.adslplus.ch) left irc: Quit: See you | 17:37 | |
| MimiEA (~LiliEAm@97-88-28-179.dhcp.roch.mn.charter.com) joined #pypy. | 17:42 | |
| nettok (~quassel@200.119.162.209) left irc: Ping timeout: 252 seconds | 17:45 | |
| nettok_ (~quassel@190.143.178.250) joined #pypy. | 17:45 | |
| mat^2 (~mathias@87.55.0.198) joined #pypy. | 18:13 | |
| fprimex (fprimex@80.sub-75-252-40.myvzw.com) left irc: Quit: Leaving | 18:15 | |
| davidpk (~davidpk@5e09151b.bb.sky.com) left irc: Quit: davidpk | 18:16 | |
| Rhyolite (~dje@pool-108-6-25-71.nycmny.fios.verizon.net) left irc: Ping timeout: 240 seconds | 18:19 | |
| ltratt (~ltratt@ntratt.plus.com) left irc: Quit: using sirc version 2.211+KSIRC/1.3.12 | 18:20 | |
| antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds | 18:31 | |
| CIA-53 | 03fijal 07roundup * 10#973/interpreter/test/test_main.py fails when run on pypy-c: [new] as is pypy-c py.test test_main.py. It looks like a recursion error * 14https://bugs.pypy.org/issue973 | 18:38 |
| phrearch (~phrearch_@5ED42DE2.cm-7-5a.dynamic.ziggo.nl) left irc: Remote host closed the connection | 18:41 | |
| fijal_ (~fijal@196-210-199-8.dynamic.isadsl.co.za) joined #pypy. | 18:46 | |
| fijal_ | Alex_Gaynor: ping | 18:46 |
| Alex_Gaynor | pong | 18:46 |
| fijal (~fijal@196-210-199-8.dynamic.isadsl.co.za) left irc: Disconnected by services | 18:47 | |
| fijal_ | do we have a plan fro numpy? | 18:47 |
| Nick change: fijal_ -> fijal | 18:47 | |
| Alex_Gaynor | you mean "what's next"? | 18:47 |
| fijal | ya | 18:47 |
| Alex_Gaynor | I want to finish complex/custom dtypes, and at some point we want to do vectorization. And then there's whatever is in the planning file | 18:48 |
| kenaan | 12fijal default 11a6b6b5a373a1 15/pypy/interpreter/gateway.py: another attempt to remove a very confusing raise | 18:49 |
| fijal | at some point I want to find out why copying in C is so much faster | 18:49 |
| fijal | and then we can write a blog post | 18:49 |
| Alex_Gaynor | fijal: well, in a contigious multi-dimensional spcae your iterator does multiple memcpys for no reason | 18:49 |
| fijal | yes, of course | 18:50 |
| Alex_Gaynor | for some value of "no reason" | 18:50 |
| fijal | is a thousand calls to memcopy that does a thousand element copy *that* much slower? | 18:50 |
| fijal | only one way to find out | 18:50 |
| fijal | note that the C equivalent is iterating over non-contiguous memory even | 18:51 |
| Alex_Gaynor | probably | 18:51 |
| Alex_Gaynor | memcpy does some stuff to make sure it's aligned I guess | 18:51 |
| fijal | how do I measure time in C? | 18:52 |
| fijal | write 3 different programs? | 18:52 |
| fijal | and use /usr/bin/time? | 18:52 |
| fijal | no, I don't want to measure malloc | 18:52 |
| Alex_Gaynor | I guess yes, and then you have to read ASM | 18:53 |
| fijal | ... | 18:53 |
| Alex_Gaynor | to make sure gcc didn't kill the whole thing | 18:53 |
| fijal | you can't copy memory copying :) | 18:53 |
| fijal | usually | 18:53 |
| fijal | because you can't prove things easily | 18:53 |
| fijal | pointers really prevent you from doing opts | 18:53 |
| Alex_Gaynor | if GCC had virtuals for raw memory you could make | 18:53 |
| fijal | so, how do I measure sub-second time in C? | 18:53 |
| Alex_Gaynor | well, it's only legal to have a pointer inside valid memory | 18:53 |
| fijal | yeah, it does not | 18:53 |
| Alex_Gaynor | :) | 18:54 |
| fijal | gettimeofday? | 18:54 |
| Alex_Gaynor | uh, how is time.time() implemented in Python | 18:54 |
| fijal | yeah, but it's structs and shit | 18:55 |
| MostAwesomeDude | QFT | 18:55 |
| Alex_Gaynor | fijal: from the nonsense department, because we have no stack alloc, time.time() mallocs | 18:56 |
| fijal | Alex_Gaynor: yeah, we should improve that at some point | 18:56 |
| fijal | MostAwesomeDude: ? | 18:56 |
| fijal | Alex_Gaynor: how do I pass "this is C 99"? | 18:56 |
| Alex_Gaynor | -std=c99 | 18:56 |
| MostAwesomeDude | fijal: "Yeah, but it's structs and shit" | 18:58 |
| fijal | Alex_Gaynor: http://paste.pocoo.org/show/526132/ | 19:00 |
| fijal | http://paste.pocoo.org/show/526130/ | 19:00 |
| fijal | looks relatively legit, no? | 19:00 |
| Alex_Gaynor | fijal: no, it does 1000 memcpys | 19:00 |
| fijal | yes, ok | 19:01 |
| fijal | so the C version is 0.23s vs 0.27s | 19:01 |
| fijal | not too bad | 19:01 |
| Alex_Gaynor | I mean this code is fine | 19:01 |
| Alex_Gaynor | but we know our internals are broken | 19:01 |
| Alex_Gaynor | this doesn't use memcpy though | 19:01 |
| Alex_Gaynor | does C vectorize or invert this? | 19:02 |
| fijal | it vectorizes | 19:02 |
| Alex_Gaynor | oh this is double cheating | 19:02 |
| fijal | invert? | 19:02 |
| Alex_Gaynor | you use 2 pointers for the C arrays | 19:02 |
| Alex_Gaynor | but the Python is a contigious block | 19:02 |
| Alex_Gaynor | cheating! | 19:02 |
| fijal | yes, cheating | 19:02 |
| Alex_Gaynor | so benchmark us against a contigious block | 19:03 |
| fijal | single memcpy is 0.16s | 19:04 |
| Alex_Gaynor | better than 50% faster than us | 19:04 |
| fijal | and 1000 memcpys is 0.15s | 19:05 |
| fijal | for what is worth | 19:05 |
| fijal | we somehow mess shit up | 19:05 |
| fijal | http://dpaste.com/678045/ | 19:05 |
| fijal | Alex_Gaynor: does this look like what we're doing? | 19:06 |
| Alex_Gaynor | well, it's not like our setslice doesn't do other cra | 19:06 |
| fijal | how much other crap is my question? | 19:06 |
| fijal | is our interpreter *that* slow? | 19:06 |
| Alex_Gaynor | fijal: yes, + or - the other crap. | 19:06 |
| fijal | last time I looked it was >80% spent in memcpy | 19:06 |
| fijal | on valgrind | 19:06 |
| Alex_Gaynor | fijal: use ltrace -c + time to see how much is in memcpy? | 19:07 |
| fijal | ltrace? | 19:07 |
| Alex_Gaynor | traces calls to the stdlib | 19:07 |
| fijal | valgrind claims 89% | 19:07 |
| fijal | but we're over 2x slower | 19:08 |
| fijal | or around | 19:08 |
| fijal | not 30% | 19:08 |
| Alex_Gaynor | well some of those memcpys can be from copying | 19:08 |
| Alex_Gaynor | copying GC memory that is | 19:08 |
| Alex_Gaynor | instead of he actual array copy | 19:08 |
| fijal | no, it's exactly 100 000 | 19:08 |
| fijal | well, it's not | 19:09 |
| fijal | but 80% of time is spent in ConcreteArray.setslice | 19:09 |
| fijal | valgrind is good at figuring this out | 19:09 |
| Alex_Gaynor | how much in _fast_setslice though? | 19:09 |
| fijal | 80% | 19:09 |
| Alex_Gaynor | 80% of total, or of setslice? | 19:10 |
| fijal | of total | 19:10 |
| Alex_Gaynor | do groups of 1k and see if we get faster once the JIT kicks in? | 19:10 |
| fijal | 99.02 21.386075 157 135412 memcpy | 19:10 |
| fijal | this is ltrace | 19:10 |
| Alex_Gaynor | that percent is out of total lib call time, not total execution time | 19:11 |
| fijal | not particularly | 19:11 |
| fijal | we get slower at some point where jit tries to trace | 19:12 |
| Alex_Gaynor | do we get faster after that? | 19:12 |
| fijal | no | 19:12 |
| fijal | WTF? | 19:12 |
| Alex_Gaynor | is the chunk of memory we copy the right size? | 19:13 |
| fijal | this is the key to understand why our laplace is slow | 19:13 |
| fijal | IMO | 19:13 |
| Alex_Gaynor | ltrace without -c, see if the chunks of memory are the right sizes | 19:13 |
| Alex_Gaynor | ok they're all 8000 for me | 19:14 |
| fijal | 8000 | 19:14 |
| fijal | sounds good | 19:14 |
| fijal | oh | 19:15 |
| fijal | my compiled C does not call memcpy | 19:15 |
| fijal | it inlined it | 19:15 |
| Alex_Gaynor | fijal: I'm going to get lunch, maybe try turning this into a single memcpy is a good starting point | 19:15 |
| fijal | http://paste.pocoo.org/show/526135/ | 19:16 |
| fijal | Alex_Gaynor: ^^^ | 19:16 |
| Alex_Gaynor | fijal: this is why I said check the ASM :) really going now | 19:18 |
| fijal | ok, but the time does not change if it's still a call | 19:19 |
| durin42 (~durin@adium/durin42) left irc: Ping timeout: 252 seconds | 19:19 | |
| durin42 (~durin@adium/durin42) joined #pypy. | 19:21 | |
| fijal | Alex_Gaynor: why do we ever call memcopy with "0" as an argument anyway? | 19:22 |
| DasIch__ (~dasich@p4FFDF3A8.dip.t-dialin.net) joined #pypy. | 19:27 | |
| DasIch_ (~dasich@p4FFDF0B6.dip.t-dialin.net) left irc: Ping timeout: 252 seconds | 19:30 | |
| PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood | 19:39 | |
| Alex_Gaynor | fijal: 0 for size? | 19:40 |
| Alex_Gaynor | I dunno, because nonsense? | 19:40 |
| PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy. | 19:41 | |
| Rhyolite (~dje@pool-108-6-25-71.nycmny.fios.verizon.net) joined #pypy. | 19:41 | |
| Dulak (~michael@unaffiliated/dulak) joined #pypy. | 19:44 | |
| fijal | Alex_Gaynor: note that 1-dim array does not help | 19:45 |
| fijal | WTF | 19:45 |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Ping timeout: 244 seconds | 19:51 | |
| fijal | Alex_Gaynor: I'm sure it's copy_array_contents | 19:51 |
| fijal | we should not really call it if its' 0 | 19:51 |
| Alex_Gaynor | fijal: copy_array_contents? | 19:52 |
| fijal | where memcpy(0) comes from | 19:53 |
| Alex_Gaynor | so this is copying other arrays? | 19:54 |
| kenaan | 12fijal default 11c3142d5c9c35 15/pypy/module/micronumpy/interp_numarray.py: remove some unnecessary copies (they used to be necessary, not any more) | 19:54 |
| mattip (~chatzilla@bzq-79-177-107-57.red.bezeqint.net) joined #pypy. | 19:54 | |
| fijal | yes, I think so | 19:56 |
| fijal | we need a faster interpreter btw :/ | 19:56 |
| fijal | or something interpreter-like | 19:56 |
| mattip | fijal: hi. Is now an OK time for a design pattern question, or should I wait my turn? | 19:57 |
| Alex_Gaynor | fijal: basically you want jaegermonkey ;) | 19:57 |
| fijal | mattip: it's fine | 19:57 |
| fijal | Alex_Gaynor: no, I don't want jaegermonkey :) | 19:57 |
| mattip | Iterators. you built them so .next() returns a new one rather than reusing the old. Why? | 19:58 |
| Alex_Gaynor | mattip: the JIT likes it | 19:59 |
| Alex_Gaynor | the JIT likes immutable things | 19:59 |
| mattip | Ah. So on my numpy-axisop branch I will add that feature to Axisiterator, which currently is not consistent with the others. | 20:00 |
| fijal | mattip: eventually you will have to change it so it does have a consistent interface | 20:01 |
| kkris (~kris@93-82-40-79.adsl.highway.telekom.at) joined #pypy. | 20:01 | |
| fijal | because I believe we would like to JIT that one day | 20:01 |
| fijal | Alex_Gaynor: I *think* I know what's up | 20:01 |
| Alex_Gaynor | fijal: yes? | 20:01 |
| fijal | let me do few more experiments | 20:02 |
| Alex_Gaynor | k | 20:07 |
| fijal | actually | 20:13 |
| fijal | we have a small lang, we don;'t have to translate pypy! | 20:14 |
| fijal | Alex_Gaynor: I dunno, I think allocations around create_slice trash caches badly | 20:14 |
| Alex_Gaynor | fijal: stupid caches | 20:14 |
| fijal | as in processor cache | 20:14 |
| dD0T (~dD0T@unaffiliated/dd0t) left irc: Ping timeout: 252 seconds | 20:15 | |
| Alex_Gaynor | fijal: right | 20:15 |
| Ademan-remote (~yaaic@214.sub-97-159-152.myvzw.com) joined #pypy. | 20:16 | |
| fijal | anyway | 20:16 |
| fijal | coming back to the point - I don't quite understand what's so costly in laplace | 20:17 |
| fijal | maybe something else | 20:18 |
| fijal | but for that I have to fix the jitviewer :/ | 20:18 |
| Alex_Gaynor | measure | 20:18 |
| Alex_Gaynor | fix jitviewer | 20:18 |
| fijal | I don't want to fix jitviewer :/ | 20:19 |
| fijal | I did not break it | 20:19 |
| Alex_Gaynor | well, ok, but no one can use it ATM | 20:19 |
| fijal | yes | 20:19 |
| mikefc | but if i wanted to tune my python code, i'd need to use it! | 20:22 |
| mikefc | (morning) | 20:22 |
| Alex_Gaynor | fijal: perhaps we should add a step to the JIT builders to produce a logfile and have jitviewer process it or something | 20:24 |
| Ademan-remote (~yaaic@214.sub-97-159-152.myvzw.com) left irc: Ping timeout: 268 seconds | 20:28 | |
| fijal | Alex_Gaynor: nice idea | 20:29 |
| Alex_Gaynor | fijal: anyway, if it's the cache issue maybe try adding some unrolling so there's no stuff besides memcpy in the trace | 20:29 |
| fijal | mikefc: how does that keep me motivated? | 20:31 |
| fijal | Alex_Gaynor: there is no trace | 20:31 |
| fijal | btw | 20:31 |
| fijal | unless we JIT the loop with memcpy | 20:31 |
| Alex_Gaynor | fijal: just increase to 1500 iterations | 20:32 |
| Alex_Gaynor | fijal: wtf is list2fixed | 20:32 |
| Action: Alex_Gaynor removes the oopspec | 20:35 | |
| kenaan | 12alex_gaynor default 1121bb1c0bd208 15/pypy/: it's fine to inline this. | 20:36 |
| kenaan | 12alex_gaynor default 11a0b6f54563e6 15/pypy/: merged upstream | 20:36 |
| Nick change: DasIch__ -> DasIch | 20:38 | |
| fijal | Alex_Gaynor: this is a very important thing | 20:39 |
| Alex_Gaynor | fijal: for listcomprehension? | 20:39 |
| fijal | this make say list-comprehension fixed | 20:39 |
| fijal | yes | 20:39 |
| Alex_Gaynor | fijal: ok, but it can be inlined by the JIT fine | 20:39 |
| fijal | yes | 20:40 |
| fijal | did you just spot it in traces? | 20:40 |
| Alex_Gaynor | yes | 20:40 |
| Alex_Gaynor | if you increase to 1500 iterations and look in trace it's there | 20:40 |
| fijal | eh | 20:40 |
| Alex_Gaynor | (please fix jitviewer, reading traces without it sucks) | 20:40 |
| fijal | yeah, I know | 20:40 |
| Action: fijal fixes | 20:40 | |
| Alex_Gaynor | WTF, pygments highlighter is broken too | 20:41 |
| Alex_Gaynor | oh, nvm, that's my fault | 20:42 |
| kenaan | 12fijal jitviewer 111d7244f7b370 15/: move stuff around - helps with setup.py develop | 20:43 |
| fijal | obviously id(token) is not really awesome :/ | 20:44 |
| fijal | hakanardo: :/ | 20:44 |
| fijal | well, works | 20:44 |
| fijal | hakanardo: ping? | 20:50 |
| fijal | Alex_Gaynor: do you know stuff about labels? | 20:50 |
| fijal | can you jump only to a label or can you jump to the beginning of the loop? | 20:50 |
| Alex_Gaynor | fijal: I'm not positive | 20:50 |
| fijal | ok well | 20:50 |
| fijal | now it's obviously broken | 20:51 |
| fijal | because we want to also count times you enter the loop from somewhere else | 20:52 |
| fijal | ARGH | 20:52 |
| Alex_Gaynor | no, it's fine to only count at labels I think | 20:53 |
| Alex_Gaynor | because if you enter at the front you'll still hit the labels | 20:53 |
| fijal | ah no | 20:53 |
| fijal | yes | 20:53 |
| fijal | but the front label is hidden :) | 20:53 |
| fijal | so I did not see it | 20:54 |
| durin42 (~durin@adium/durin42) left irc: Ping timeout: 248 seconds | 20:54 | |
| Alex_Gaynor | fijal: that's the label's fault ;) | 20:55 |
| durin42 (~durin@adium/durin42) joined #pypy. | 20:56 | |
| mattip (chatzilla@bzq-79-177-107-57.red.bezeqint.net) left #pypy. | 20:57 | |
| nickik (~nick@2001:67c:20a1:1018:226:9eff:feef:8369) joined #pypy. | 20:58 | |
| Nick change: coob -> elliottcable | 21:04 | |
| durin42 (~durin@adium/durin42) left irc: Ping timeout: 240 seconds | 21:04 | |
| durin42 (~durin@adium/durin42) joined #pypy. | 21:06 | |
| fijal | eh | 21:09 |
| fijal | this is so broken... | 21:09 |
| kenaan | 12fijal default 1140a95aa4967f 15/pypy/tool/jitlogparser/: some tests and fixes | 21:09 |
| kenaan | 12fijal default 11cc6d58803952 15/pypy/: merge | 21:09 |
| Alex_Gaynor | fijal: broken before or after the tests and fixes? | 21:09 |
| kenaan | 12fijal jitviewer 11c63afaebb8d6 15/: a bit of cleanup, reuse more of jitlogparser | 21:11 |
| kenaan | 12fijal default 11c8ddbb442986 15/pypy/tool/jitlogparser/parser.py: split while doing import | 21:11 |
| davidpk (~davidpk@5e09151b.bb.sky.com) joined #pypy. | 21:13 | |
| fijal | Alex_Gaynor: meh, I should not have let anto near JS code :-) | 21:19 |
| Alex_Gaynor | fijal: it's not like current JS is fantastic, it really should use something like https://github.com/zachwill/pjax_flask | 21:20 |
| fijal | Alex_Gaynor: well, I'm more complaining about JS in __repr__ of traces | 21:21 |
| JaRoel|4d (~jaroel|4d@178-85-196-53.dynamic.upc.nl) left irc: Remote host closed the connection | 21:21 | |
| Alex_Gaynor | as a general practice I wouldn't put JS in __repr__ of anything | 21:21 |
| JaRoel|4d (~jaroel|4d@178-85-196-53.dynamic.upc.nl) joined #pypy. | 21:21 | |
| fijal | I-don't-feel-like-cleaning-this-shit | 21:22 |
| Alex_Gaynor | fijal: I think I might write my own hosted traceviewer + analysis tool | 21:24 |
| Alex_Gaynor | fijal: btw we should version traces somehow | 21:24 |
| fijal | "version traces" | 21:24 |
| fijal | as in? | 21:24 |
| fijal | well, I can help you | 21:24 |
| Alex_Gaynor | put a marker in so you know "this is the version that goes with this" | 21:25 |
| fijal | as in - let's write one instead of two, barely working | 21:25 |
| fijal | yes | 21:25 |
| Alex_Gaynor | ATM whenever we change trace or jitviewer you have to make sure you have precisely the right version | 21:25 |
| fijal | there is a version of jitviewer that works with release of pypy | 21:25 |
| Alex_Gaynor | fijal: I want to play with making some HTML mockups first | 21:25 |
| fijal | at least | 21:25 |
| fijal | okey | 21:25 |
| fijal | the part that's in pypy is probably reusable | 21:26 |
| fijal | most of it anyway | 21:26 |
| fijal | the part that deals with parsing traces, adjusting numbers etc. | 21:26 |
| fijal | the rest is probably unreusable | 21:26 |
| Alex_Gaynor | well, one step at a time | 21:26 |
| fijal | yeah | 21:26 |
| Alex_Gaynor | fijal: basically I want something where I can run PYPYLOG=x.pypylog <my program>, upload the trace, and it gives me graphs and stores it permanently | 21:27 |
| Alex_Gaynor | sort of like a trace-bin :) | 21:27 |
| JaRoel|4d (~jaroel|4d@178-85-196-53.dynamic.upc.nl) left irc: Ping timeout: 248 seconds | 21:27 | |
| Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Ping timeout: 276 seconds | 21:33 | |
| fijal | Alex_Gaynor: ok, I give up for tonight | 21:37 |
| fijal | there are quite a few entrenched assumptions, and essentially the entire loop-parsing stuff has to be retested | 21:37 |
| fijal | I'll fix it tomorrow then | 21:37 |
| fijal | it's impossible to work without the jitviewer | 21:38 |
| Alex_Gaynor | fijal: dgreed, 'night | 21:38 |
| mimi_ (~mimi@97-88-28-179.dhcp.roch.mn.charter.com) joined #pypy. | 21:44 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 21:52 | |
| tito (~tito@pdpc/supporter/active/tito) joined #pypy. | 21:56 | |
| EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) joined #pypy. | 22:14 | |
| azanella (~azanella@189.6.80.207) left irc: Ping timeout: 255 seconds | 22:20 | |
| JaRoel|4d (~jaroel|4d@2001:980:7133:1:596c:c892:3579:7041) joined #pypy. | 22:23 | |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 22:26 | |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/185 | 22:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/757 | 22:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/994 | 22:30 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/13 | 22:30 |
| kkris (~kris@93-82-40-79.adsl.highway.telekom.at) left irc: Quit: Leaving. | 22:43 | |
| davidpk (~davidpk@5e09151b.bb.sky.com) left irc: Quit: davidpk | 22:43 | |
| davidpk (~davidpk@5e09151b.bb.sky.com) joined #pypy. | 22:46 | |
| davidpk (~davidpk@5e09151b.bb.sky.com) left irc: Quit: davidpk | 23:01 | |
| jyp (~interrupt@175.197.245.105) joined #pypy. | 23:07 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Read error: Operation timed out | 23:11 | |
| `fox` (~fox@host2-105-dynamic.54-82-r.retail.telecomitalia.it) joined #pypy. | 23:14 | |
| jyp (~interrupt@175.197.245.105) left irc: Remote host closed the connection | 23:17 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 23:17 | |
| kennethr_ (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 23:22 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Ping timeout: 244 seconds | 23:22 | |
| kennethr_ (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Client Quit | 23:22 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 23:24 | |
| antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy. | 23:25 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Client Quit | 23:29 | |
| mimi_ (~mimi@97-88-28-179.dhcp.roch.mn.charter.com) left irc: Ping timeout: 276 seconds | 23:31 | |
| derfel (~derfel@host44-70-dynamic.0-79-r.retail.telecomitalia.it) joined #pypy. | 23:34 | |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds | 23:34 | |
| derfel_ (~derfel@host44-70-dynamic.0-79-r.retail.telecomitalia.it) joined #pypy. | 23:34 | |
| derfel_ (~derfel@host44-70-dynamic.0-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer | 23:34 | |
| derfel_ (~derfel@host44-70-dynamic.0-79-r.retail.telecomitalia.it) joined #pypy. | 23:35 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 23:35 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Client Quit | 23:36 | |
| derfel_ (~derfel@host44-70-dynamic.0-79-r.retail.telecomitalia.it) left irc: Client Quit | 23:36 | |
| derfel (~derfel@host44-70-dynamic.0-79-r.retail.telecomitalia.it) left irc: Client Quit | 23:36 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 23:37 | |
| LiliEA (~LiliEAm@97-88-28-179.dhcp.roch.mn.charter.com) joined #pypy. | 23:41 | |
| MimiEA (~LiliEAm@97-88-28-179.dhcp.roch.mn.charter.com) left irc: Ping timeout: 255 seconds | 23:42 | |
| mimi_ (~mimi@97-88-28-179.dhcp.roch.mn.charter.com) joined #pypy. | 23:44 | |
| mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord | 23:48 | |
| JaRoel|4d (~jaroel|4d@2001:980:7133:1:596c:c892:3579:7041) left irc: Remote host closed the connection | 23:59 | |
| --- Tue Dec 27 2011 | 00:00 | |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!