| mattip | I gotta go, any last questions? | 00:00 |
|---|---|---|
| tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy. | 00:00 | |
| fijal | cool stuff | 00:01 |
| fijal | mattip: no, thanks | 00:01 |
| mattip | ttfn | 00:01 |
| mattip (~mattip@bzq-79-183-105-189.red.bezeqint.net) left irc: Quit: Leaving | 00:02 | |
| kenaan | 12fijal default 11d246547e90d8 15/pypy/module/micronumpy/: (mattip) merge numpypy-reshape-bug | 00:05 |
| kenaan | 12fijal numpypy-shape-bug 1112c695258f06 15/: close merged-branch | 00:05 |
| fijal | hm | 00:06 |
| fijal | is buildbot on wyvern running under venv? | 00:06 |
| fijal | Alex_Gaynor: should we compare against 1.6.1 numpy or numpy 2.0-dev? | 00:08 |
| fijal | bleh | 00:09 |
| fijal | official numpy website refers to svn | 00:09 |
| fijal | while numpy devel happens on github | 00:09 |
| fijal | nice | 00:09 |
| amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 272 seconds | 00:21 | |
| DasIch (~DasIch@p4FFDCB0C.dip.t-dialin.net) left irc: Quit: DasIch | 00:21 | |
| fijal | Alex_Gaynor: are you reviewing the pull request or can I just pull it? | 00:28 |
| fijal | looks relatively innocent | 00:28 |
| fijal | to the best of my knowledge | 00:29 |
| kenaan | 12dripton default 113ed178fb213d 15/lib_pypy/_sqlite3.py: Remove unreachable statement referring to nonexistant variable. | 00:30 |
| kenaan | 12dripton default 11d39f5493dcaf 15/lib_pypy/_sqlite3.py: Remove unused import. | 00:30 |
| kenaan | 12dripton default 11014a02a9d3df 15/lib_pypy/_sqlite3.py: Remove unused local variables. | 00:30 |
| kenaan | 12dripton default 11d49f44c4a62f 15/lib_pypy/_sqlite3.py: Fix old-style raise statements (illegal in Python 3, ugly in Python 2) | 00:30 |
| kenaan | 12dripton default 1101ed002a2b95 15/lib_pypy/_sqlite3.py: Add a comment making it clear this file is modified, for license compliance. | 00:30 |
| kenaan | 12dripton default 116141b7352630 15/pypy/module/test_lib_pypy/test_sqlite3.py: Failing test for issue 996 | 00:30 |
| kenaan | 12dripton default 1180014f10414c 15/lib_pypy/_sqlite3.py: Do not call statement.reset() after executing DDL. Fixes issue996. | 00:31 |
| kenaan | 12dripton default 11fd6e2144e652 15/pypy/module/test_lib_pypy/test_sqlite3.py: Add an assert, as requested by Alex. | 00:31 |
| squiddy (~squiddy@f053087141.adsl.alicedsl.de) left irc: Quit: Leaving | 00:33 | |
| fijal | dripton: done | 00:33 |
| fijal | thanks! | 00:33 |
| asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Quit: asmeurer | 00:50 | |
| asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy. | 00:52 | |
| fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) left irc: Ping timeout: 240 seconds | 00:53 | |
| kvda | Hello | 00:55 |
| kvda | Ah did fijal just leave. | 00:56 |
| bgola (~bgola@c951826e.virtua.com.br) left irc: Ping timeout: 252 seconds | 01:03 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/784 | 01:04 |
| Unhelpful (~quassel@rockbox/developer/Unhelpful) left irc: Quit: http://quassel-irc.org - Chat comfortably. Anywhere. | 01:17 | |
| Unhelpful (~quassel@rockbox/developer/Unhelpful) joined #pypy. | 01:17 | |
| nettok_ (~quassel@190.148.249.220) joined #pypy. | 01:19 | |
| mat^2 (~mathias@212.130.113.35) left irc: Ping timeout: 245 seconds | 01:21 | |
| bgola (~bgola@c951826e.virtua.com.br) joined #pypy. | 01:24 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/40 | 01:29 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/213 | 01:39 |
| tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection | 01:54 | |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1937 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/687 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/347 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/337 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/684 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1209 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/869 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1527 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/798 | 02:00 |
| stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy. | 02:08 | |
| stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 02:08 | |
| Nick change: stakkars_ -> stakkars | 02:08 | |
| nirbheek (~nirbheek@117.197.144.115) joined #pypy. | 02:12 | |
| nirbheek (~nirbheek@117.197.144.115) left irc: Changing host | 02:12 | |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy. | 02:12 | |
| stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy. | 02:13 | |
| stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 02:13 | |
| Nick change: stakkars_ -> stakkars | 02:13 | |
| CIA-17 | 03dripton 07roundup * 10#996/sqlite3 CREATE TABLE always fails: [resolved] That pull request was merged, so I think this is fixed. * 14https://bugs.pypy.org/issue996 | 02:14 |
| stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy. | 02:15 | |
| stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 02:15 | |
| Nick change: stakkars_ -> stakkars | 02:15 | |
| jaffachief (~textual@host86-170-15-228.range86-170.btcentralplus.com) left irc: Quit: Computer has gone to sleep. | 02:25 | |
| etrepum (~bob@75-101-96-144.dsl.static.sonic.net) joined #pypy. | 02:48 | |
| dripton (~dripton@pool-72-86-45-238.clppva.fios.verizon.net) left irc: Quit: Konversation terminated! | 02:50 | |
| Moku (~John@osbk-4db1688f.pool.mediaWays.net) left irc: Ping timeout: 240 seconds | 03:03 | |
| csenger` (~user@dslb-084-058-148-237.pools.arcor-ip.net) joined #pypy. | 03:03 | |
| Moku (~John@osbk-4db14ccb.pool.mediaWays.net) joined #pypy. | 03:03 | |
| csenger (~user@dslb-084-058-191-114.pools.arcor-ip.net) left irc: Ping timeout: 252 seconds | 03:07 | |
| whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) left irc: Ping timeout: 245 seconds | 03:16 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/1023 | 03:27 |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/337 | 03:29 |
| Moku (~John@osbk-4db14ccb.pool.mediaWays.net) left irc: Remote host closed the connection | 03:35 | |
| Kaskuka (~John@osbk-4db14ccb.pool.mediaWays.net) joined #pypy. | 03:38 | |
| Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 255 seconds | 03:38 | |
| Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy. | 03:43 | |
| PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood | 03:48 | |
| PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy. | 03:49 | |
| csenger` (~user@dslb-084-058-148-237.pools.arcor-ip.net) left irc: Quit: ERC Version 5.3 (IRC client for Emacs) | 04:10 | |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 276 seconds | 04:16 | |
| fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) joined #pypy. | 04:22 | |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 04:23 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1937 | 04:29 |
| icrazyhack (~horieyui@183.93.198.19) joined #pypy. | 04:33 | |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 248 seconds | 04:49 | |
| anish_ (anish@nat/redhat/x-kcpqxtjeyognfria) joined #pypy. | 04:50 | |
| papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep. | 05:04 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1527 | 05:31 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/869 | 05:32 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/684 | 05:35 |
| nettok_ (~quassel@190.148.249.220) left irc: Remote host closed the connection | 05:50 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1209 | 05:53 |
| Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 252 seconds | 05:56 | |
| fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) left irc: Quit: 'night! | 05:57 | |
| Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy. | 05:58 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/687 | 06:06 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Read error: Connection reset by peer | 06:12 | |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 06:13 | |
| mat^2 (~mathias@212.130.113.35) joined #pypy. | 06:32 | |
| icrazyhack (~horieyui@183.93.198.19) left irc: Read error: Connection reset by peer | 06:52 | |
| icrazyhack (~horieyui@183.60.101.86) joined #pypy. | 06:53 | |
| mat^2 (~mathias@212.130.113.35) left irc: Ping timeout: 252 seconds | 06:54 | |
| horieyui (~horieyui@183.93.198.19) joined #pypy. | 06:57 | |
| icrazyhack (~horieyui@183.60.101.86) left irc: Ping timeout: 252 seconds | 06:58 | |
| amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy. | 06:59 | |
| tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy. | 07:01 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/798 | 07:06 |
| asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Quit: asmeurer | 07:10 | |
| amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Read error: Operation timed out | 07:13 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/347 | 07:15 |
| mvt (~mvantelli@87.213.45.85) joined #pypy. | 07:31 | |
| fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) joined #pypy. | 07:32 | |
| JaRoel|4d (~jaroel|4d@sink.jaroel.nl) left irc: Remote host closed the connection | 07:32 | |
| csenger` (~user@dslb-084-058-148-237.pools.arcor-ip.net) joined #pypy. | 07:41 | |
| Telugodu (~rambaho@202.3.77.231) joined #pypy. | 07:47 | |
| JaredWigmore (~jared@n112119177158.netvigator.com) joined #pypy. | 07:49 | |
| CIA-17 | 03fijal 07roundup * 10#1008/numpypy: commit the code for empty_like, zeros_like, ones_like: | 07:50 |
| CIA-17 | These functions are implemented directly in applevel python in numpy and should | 07:50 |
| CIA-17 | simply be copied over. Tests are absolutely necessary for the co ... * 14https://bugs.pypy.org/issue1008 | 07:50 |
| Telugodu (~rambaho@202.3.77.231) left irc: Ping timeout: 240 seconds | 07:54 | |
| CIA-17 | 03fijal 07roundup * 10#1011/pypy slower than CPython for recursion: | 08:01 |
| CIA-17 | [resolved] PyPy has some problems with recursion (it can be faster), but the latest nightly | 08:01 |
| CIA-17 | is already a bit faster than CPython, closing. * 14https://bugs.pypy.org/issue1011 | 08:01 |
| squiddy (~squiddy@f053084057.adsl.alicedsl.de) joined #pypy. | 08:08 | |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 252 seconds | 08:14 | |
| CIA-17 | 03fijal 07roundup * 10#1016/RPython AssertionError while running async io greenlet test: | 08:16 |
| CIA-17 | FYI to run without JIT, it's enough to run a jitted version of pypy like this: | 08:16 |
| CIA-17 | pypy --jit off <args> | 08:16 |
| CIA-17 | It's much easier than retranslating :) * 14https://bugs.pypy.org/issue1016 | 08:16 |
| CIA-17 | 03fijal 07roundup * 10#1016/RPython AssertionError while running async io greenlet test: | 08:19 |
| CIA-17 | I get an error: error: [Errno 11] Resource temporarily unavailable | 08:19 |
| CIA-17 | any idea what that means? * 14https://bugs.pypy.org/issue1016 | 08:19 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Quit: pnguyen | 08:35 | |
| fijal | csenger`: hi | 08:37 |
| antocuni_ (~antocuni@host155-125-dynamic.25-79-r.retail.telecomitalia.it) joined #pypy. | 08:44 | |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 08:50 | |
| horieyui (~horieyui@183.93.198.19) left irc: Ping timeout: 240 seconds | 09:00 | |
| AlekSi (~AlekSi@83.149.8.143) joined #pypy. | 09:17 | |
| lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy. | 09:17 | |
| mikefc_ | fijal: zeros_like and ones_like are both wrappers around "empty_like" which is a c func in core/src/multiarray/multiarraymodule.c | 09:23 |
| mikefc_ | which i'd get to if I have more than 30 seconds spare at the keyboard. :( | 09:23 |
| fijal | mikefc_: I made it possible to have app-level code as methods | 09:24 |
| fijal | the same way as numpy does | 09:24 |
| fijal | var is an app-level method now | 09:24 |
| mikefc_ | ah! | 09:25 |
| mikefc_ | that's cool :) | 09:25 |
| fijal | on a branch though | 09:25 |
| fijal | I'm trying to make __str__ back applevel and __repr__ | 09:25 |
| mikefc_ | but empty_like is C code. So we're going to create a pure python version at app level in numpypy? | 09:26 |
| fijal | we can put it on interp-level if it makes sense | 09:29 |
| mikefc_ | not sure. I'm not exactly sure what empty like actually allocates. a simple call to "zeros()" instead is a stopgap. but i'll look at what the numpy's tests say. | 09:30 |
| dmitrey (~quassel@77.121.195.200) joined #pypy. | 09:33 | |
| fijal | dmitrey: hi | 09:34 |
| csenger` | fijal: hi | 09:34 |
| fijal | csenger`: so right now your way (have a separate factory for running cpython) runs cpython benchmarks twice | 09:34 |
| dmitrey | hi fijal, I don't understand what do you mean by "Also tests do that implicitely" | 09:35 |
| fijal | I think we can avoid it, by having say just one that runs jitted version of pypy vs cpython | 09:35 |
| csenger` | I think it runs it as baseline a second time? | 09:35 |
| fijal | dmitrey: look say at pypy/module/test_lib_pypy/numpypy/ | 09:35 |
| kenaan | 12fijal numpy-back-to-applevel 110068fa0d5e01 15/: merge default | 09:36 |
| fijal | dmitrey: those are AppTests which means they'll run on untranslated pypy | 09:36 |
| Telugodu (~rambaho@202.3.77.231) joined #pypy. | 09:36 | |
| fijal | csenger`: it runs cpython and cpython, no? | 09:36 |
| dmitrey | ok | 09:36 |
| csenger` | It starts the runner once, with the interpreter set both as pypy-c and baseline. | 09:37 |
| unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) left irc: Ping timeout: 245 seconds | 09:37 | |
| csenger` | I could not make sense out of it. Does it run the baseline every time to have a reference value? | 09:37 |
| fijal | '-p', cpython_interpreter, | 09:38 |
| fijal | '--baseline', cpython_interpreter, | 09:38 |
| fijal | this is what I see no? | 09:38 |
| fijal | yes | 09:38 |
| csenger` | yes. | 09:38 |
| fijal | it comes from unladen swallow, it sucks | 09:38 |
| fijal | but it's good enough for us not to write a new bench runner | 09:38 |
| sacred_coder (~rambaho@202.3.77.231) joined #pypy. | 09:39 | |
| fijal | so you probably want a factory that both translates pypy and compiles cpython | 09:39 |
| fijal | and then runs them against each other | 09:39 |
| Telugodu (~rambaho@202.3.77.231) left irc: Client Quit | 09:39 | |
| csenger` | We could try to pass in the baseline values we aggregated once as long as the maschine does not change. | 09:39 |
| fijal | the point is running runner twice is wasteful | 09:39 |
| unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) joined #pypy. | 09:39 | |
| antocuni_ | hi | 09:39 |
| fijal | especially with cpython that takes forever | 09:39 |
| csenger` | but we need the relative change compared to a 'know' value. | 09:39 |
| Nick change: antocuni_ -> antocuni | 09:39 | |
| fijal | csenger`: that's dealt with automatically | 09:40 |
| csenger` | Which has to be the results of the same interpreter every time. | 09:40 |
| JaredWigmore (~jared@n112119177158.netvigator.com) left irc: Ping timeout: 240 seconds | 09:40 | |
| fijal | antocuni: hi | 09:40 |
| fijal | csenger`: sorry, I fail to follow? | 09:40 |
| csenger` | don't think that works if the performance of pypy and cpython change on the same day. | 09:40 |
| fijal | no it does | 09:40 |
| fijal | because you have a history | 09:40 |
| fijal | look at speed.pypy.org | 09:41 |
| fijal | you can compare against yesterday's run | 09:41 |
| fijal | speaking of which - how much sense does it make to run 2.7 benchmarks at all? | 09:41 |
| fijal | is there any development happening there? | 09:41 |
| tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection | 09:42 | |
| csenger` | 2.7 was announced as a longer maintained version, but the changes won't be radical. | 09:42 |
| fijal | well | 09:43 |
| fijal | no performance work is being done there right? | 09:43 |
| csenger` | Nevertheless, it's not a burden to run them and it might help some day. | 09:43 |
| fijal | I heavily doubt that | 09:43 |
| fijal | but ok | 09:43 |
| dmitrey (~quassel@77.121.195.200) left irc: Ping timeout: 255 seconds | 09:44 | |
| csenger` | I have a script that can run the benchmark on past revisions of a branch to get historical data. I'd like to run that for the 2.7 branch, and maybe some of legacy-trunk to see how the performance changed. | 09:45 |
| fijal | I can tell you upfront - it didn't | 09:45 |
| fijal | but it's fine, science is better than guesses | 09:46 |
| fijal | csenger`: but still, running cpython twice is pointless | 09:48 |
| G2P (~G2P@fw-asn1.ornis.com) joined #pypy. | 09:49 | |
| stakkars_ (~tismer@i59F77E49.versanet.de) joined #pypy. | 09:50 | |
| Telugodu (~rambaho@202.3.77.231) joined #pypy. | 09:50 | |
| csenger` | yes. I will test if we can get all relevant information into codespeed in one upload so it knows in which repos to look for revision information for both. I have no codespeed running here yet and don't know how that works. | 09:51 |
| arigato (~arigo@adsl-89-217-233-144.adslplus.ch) joined #pypy. | 09:51 | |
| fijal | csenger`: I tell you upfront we can | 09:51 |
| fijal | csenger`: uploading two versions of identical numbers would not change a thing irrelevant to how codespeed works | 09:51 |
| fijal | arigato: morning armin | 09:51 |
| arigato | morning | 09:52 |
| Telugodu (~rambaho@202.3.77.231) left irc: Client Quit | 09:52 | |
| csenger` | fijal: just not sure how to interpret the executable/project information which is only passed once in the url query string. Codespeed is a blackbox for me atm so it would help if I had the right arguments to the runner :) | 09:54 |
| fijal | csenger`: I have a bit no idea how is this implemented right now, but chances are we'll need to change it | 09:54 |
| fijal | csenger`: but as you can see it does a separate upload for base and modified | 09:55 |
| fijal | so you would save data twice | 09:55 |
| csenger` | fijal: btw, I wondered if this is correct: https://bitbucket.org/pypy/benchmarks/src/fad651ead401/runner.py#cl-15 (set the name to pypy-c if --jit was passed, and to pypy-c-jit if not). It might be correct, but it looks wrong. | 09:56 |
| kvda (~kvda@124-169-134-221.dyn.iinet.net.au) left irc: Quit: x___x | 09:56 | |
| fijal | it's correct for the current hardcoded way we run it | 09:57 |
| fijal | which is pypy-c-jit vs pypy-c-nojit | 09:57 |
| fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) left irc: Ping timeout: 240 seconds | 10:01 | |
| antocuni | rguillebert: 99083d2bfef1 comes with no tests :-( | 10:02 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 10:08 | |
| fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) joined #pypy. | 10:09 | |
| arigato | fijal: is it really correct? | 10:10 |
| fijal | arigato: in what sense correct? | 10:11 |
| fijal | "works for our current setting in the buildbot" kind of correct | 10:11 |
| arigato | "so obscurely upside-down correct that I wouldn't mind a comment" kind of correct | 10:12 |
| fijal | well | 10:12 |
| fijal | the entire file is like this | 10:12 |
| fijal | not that it's a very good excuse obviously | 10:12 |
| arigato | ideally, what would help is a comment written by someone that has the slightest clue about why this is correct | 10:14 |
| arigato | (or even needed, as "--jit" alone appears not to be a valid transation option) | 10:17 |
| witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy. | 10:18 | |
| kenaan | 12mattip numppy-flatitter 11dd2ec11ec817 15/pypy/module/micronumpy/: correct faulty tests | 10:18 |
| witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy. | 10:18 | |
| apoirier_away (~apoirier@sakura.nagare.org) joined #pypy. | 10:19 | |
| mattip (c1a9466d@gateway/web/freenode/ip.193.169.70.109) joined #pypy. | 10:19 | |
| mattip | I think I finished with flatiter. | 10:20 |
| Nick change: apoirier_away -> apoirier | 10:20 | |
| mikefc_ | python -c "import numpy as np; print np.empty_like( np.array([1,2,3]) )" | 10:21 |
| mikefc_ | [4298185672 4298185648 4298185624] | 10:21 |
| mikefc_ | ^^ wtf are they? | 10:21 |
| mattip | I raise an exception on arange(10).flat[ arange(3) ] but we don't support indexing by matrices yet anyway. | 10:22 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Read error: Connection reset by peer | 10:22 | |
| mattip | mikefc: -1 | 10:22 |
| mikefc_ | mattip: ta. didn't recognise it | 10:23 |
| inhahe (~inhahe@c-50-128-212-190.hsd1.fl.comcast.net) joined #pypy. | 10:23 | |
| arigato | not really -1 | 10:24 |
| mikefc_ | somehow ANDed/shifted with the original array | 10:24 |
| arigato | that's just random garbage, as documented | 10:24 |
| mikefc_ | it seem deterministic. | 10:24 |
| mikefc_ | seems | 10:24 |
| mattip | now on to logical_and / logical_or as ufuncs, ok? | 10:26 |
| fijal | mikefc_: it's not deterministic | 10:27 |
| fijal | mattip: cool | 10:27 |
| fijal | mattip: I'll look at flatiter | 10:27 |
| fijal | mattip: your fixes fixed my problem with ravel though, cool :) | 10:27 |
| dmitrey (~quassel@77.121.195.200) joined #pypy. | 10:28 | |
| mattip | this week I'm defining success as reducing your workload :) | 10:28 |
| fijal | haha | 10:28 |
| mattip | next week, the opposite. | 10:29 |
| fijal | :( | 10:29 |
| mattip | ( dot is almost ready ) | 10:29 |
| mattip | ttfn | 10:29 |
| mattip (c1a9466d@gateway/web/freenode/ip.193.169.70.109) left irc: Quit: Page closed | 10:29 | |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 10:31 | |
| fijal | arigato: ok, I guess | 10:32 |
| ltratt (~ltratt@87.114.126.76) joined #pypy. | 10:37 | |
| fijal | ltratt: hi | 10:47 |
| fijal | arigato: the branch is trying to clean this up | 10:48 |
| fijal | so we'll pass executable_name as an explicit parameter to runner | 10:48 |
| ltratt | fijal: good morning | 10:50 |
| stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 10:51 | |
| Nick change: stakkars_ -> stakkars | 10:51 | |
| stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy. | 10:51 | |
| fijal | ltratt: we still did not fix your bug | 10:51 |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy. | 10:51 | |
| fijal | chances are you're using a virtualizable incorrectly and it's not caught correctly | 10:52 |
| whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) joined #pypy. | 10:57 | |
| ltratt | ah | 10:57 |
| ltratt | how might one use a virtualizable incorrectly? | 10:57 |
| Kaskuka (~John@osbk-4db14ccb.pool.mediaWays.net) left irc: Remote host closed the connection | 10:59 | |
| csenger` | fijal: I don't see how the data of both benchmarks (changed and baseline) can be stored in codespeed in one go. http://paste.pocoo.org/show/539169/ has the runs both with changed == True and changed == False and the data transfered + saved in result.json. | 11:00 |
| dmitrey | I see in CPython numpy asscalar(a) returns a.item(), while it's absent in numpypy. What there should be instead? | 11:00 |
| csenger` | It doesn't transfer what 'baseline' is. Codespeed has no information that results for an executable (e.g. pypy-c) are always results against pypy-c-jit so it could calculate the result for pypy-c-jit. | 11:00 |
| csenger` | The only way to not run baseline each and every time is to cache the raw result once and make unladen_swallow.perf use it instead of running baseline again. | 11:01 |
| Kaskuka (~John@osbk-4db14ccb.pool.mediaWays.net) joined #pypy. | 11:02 | |
| csenger` | (but I wouldn't invest time for speed_python as it is idle 23 hours a day ;) | 11:02 |
| AlekSi (~AlekSi@83.149.8.143) left irc: Remote host closed the connection | 11:06 | |
| azanella (~azanella@189.6.80.131) joined #pypy. | 11:06 | |
| CIA-17 | 03Dmitrey 07roundup * 10#1008/numpypy: commit the code for empty_like, zeros_like, ones_like: | 11:18 |
| CIA-17 | That time I proposed the implementation method ndarray.fill() was absent. Now | 11:18 |
| CIA-17 | it's present and I could commit the copy-pasted code from numpy. Co ... * 14https://bugs.pypy.org/issue1008 | 11:18 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Quit: pnguyen | 11:20 | |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 11:21 | |
| voidspace (~voidspace@python/psf/voidspace) joined #pypy. | 11:24 | |
| fijal | csenger`: the transfer goes in two goes | 11:30 |
| fijal | csenger`: you can see two calls to perform_upload, no? | 11:31 |
| fijal | dmitrey: .item should be implemented | 11:31 |
| fijal | dmitrey: it's more important to put stuff in correct places than having it early | 11:32 |
| fijal | so say if a method is missing, don't propose a patch that works around it | 11:32 |
| dmitrey | unfortunately I have no nough experience to work with so low level numpy funcs, I would better go for something more high-level instead. What about those empty_like etc, could I perform a test commit for them right now? | 11:33 |
| dmitrey | nough = enough | 11:34 |
| dmitrey | I don't know inner structure of numpy ndarray well enough | 11:34 |
| dmitrey | to implement .item method | 11:35 |
| fijal | empty_like is implemented in C no? | 11:36 |
| fijal | what's your proposed implementation? | 11:36 |
| fijal | pure-python goes to lib_pypy/numpypy/... and tests to module/test_lib_pypy/numpypy/... | 11:36 |
| nshepperd_ (~asdfg@ppp118-209-88-197.lns20.mel4.internode.on.net) joined #pypy. | 11:36 | |
| Nick change: ojii -> ojii|lunch | 11:36 | |
| kenaan | 12fijal buildbot 111c946f853e6b 15/numready.py: require an executable | 11:37 |
| kenaan | 12fijal buildbot 11914834b4058a 15/numready.py: minor adjustments | 11:37 |
| CIA-17 | 03arigo 07roundup * 10#995/jit bug with threads: [chatting] You forgot to attach x.py. * 14https://bugs.pypy.org/issue995 | 11:39 |
| CIA-17 | 03arigo 07roundup * 10#1016/RPython AssertionError while running async io greenlet test: | 11:44 |
| CIA-17 | papercrane: can you give us one of the version numbers that you tried? As a | 11:44 |
| CIA-17 | mercurial identifier, i.e. 12 hexadecimal digits. Thanks! * 14https://bugs.pypy.org/issue1016 | 11:44 |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 11:50 | |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Quit: pnguyen | 11:55 | |
| voidspace (~voidspace@python/psf/voidspace) left irc: Quit: Bye... | 11:59 | |
| whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) left irc: Read error: Connection reset by peer | 12:01 | |
| Nick change: ojii|lunch -> ojii | 12:08 | |
| voidspace (~voidspace@python/psf/voidspace) joined #pypy. | 12:15 | |
| CIA-17 | 03Dmitrey 07roundup * 10#1008/numpypy: commit the code for empty_like, zeros_like, ones_like: | 12:15 |
| CIA-17 | here's numpy zeros_like func latest version with 3 new parameters (in Linux | 12:15 |
| CIA-17 | channel it's still old-style 1-argument version): | 12:15 |
| CIA-17 | def zeros_like(a ... * 14https://bugs.pypy.org/issue1008 | 12:15 |
| lizardo (lizardo@nat/indt/x-xkyttychahfdyknj) joined #pypy. | 12:32 | |
| ltratt (~ltratt@87.114.126.76) left irc: Quit: using sirc version 2.211+KSIRC/1.3.12 | 12:33 | |
| khs (~khs@151.43.202.84.customer.cdi.no) joined #pypy. | 12:35 | |
| lizardo (lizardo@nat/indt/x-xkyttychahfdyknj) left irc: Ping timeout: 240 seconds | 12:39 | |
| voidspace (~voidspace@python/psf/voidspace) left irc: Quit: Goodbye... | 12:44 | |
| christophler (~web49_an_@84.45.87.204) left irc: Remote host closed the connection | 12:48 | |
| lizardo (lizardo@nat/indt/x-oyudojmhpvbyljkz) joined #pypy. | 12:50 | |
| lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) joined #pypy. | 12:53 | |
| lesshaste | hi | 12:54 |
| witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy. | 12:55 | |
| arigato | hi | 12:57 |
| witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy. | 12:57 | |
| dmitrey (~quassel@77.121.195.200) left irc: Remote host closed the connection | 13:00 | |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 13:00 | |
| DasIch (~DasIch@p3E990BC9.dip.t-dialin.net) joined #pypy. | 13:01 | |
| CIA-17 | 03arigo 07roundup * 10#1016/RPython AssertionError while running async io greenlet test: | 13:03 |
| CIA-17 | I also get "Resource temporarily unavailable" systematically, on Linux 64-bit. | 13:03 |
| CIA-17 | I can't test the original problem :-/ * 14https://bugs.pypy.org/issue1016 | 13:03 |
| nshepperd_ (~asdfg@ppp118-209-88-197.lns20.mel4.internode.on.net) left irc: Quit: WeeChat 0.3.6 | 13:04 | |
| anish_ (anish@nat/redhat/x-kcpqxtjeyognfria) left irc: Ping timeout: 255 seconds | 13:05 | |
| jaffachief (~textual@host86-170-15-228.range86-170.btcentralplus.com) joined #pypy. | 13:08 | |
| anish_ (anish@nat/redhat/x-nsexopspoqpwtsfb) joined #pypy. | 13:12 | |
| arigato | fijal: https://bugs.pypy.org/issue995: You forgot to attach x.py. | 13:16 |
| anish_ (anish@nat/redhat/x-nsexopspoqpwtsfb) left irc: Ping timeout: 260 seconds | 13:20 | |
| anish_ (anish@nat/redhat/x-vpgockvpcwaidqro) joined #pypy. | 13:21 | |
| voidspace (~voidspace@python/psf/voidspace) joined #pypy. | 13:34 | |
| antocuni (~antocuni@host155-125-dynamic.25-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds | 13:39 | |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Quit: pnguyen | 13:41 | |
| tumbleweed | fijal: you thought this was worth upstreaming: http://anonscm.debian.org/gitweb/?p=collab-maint/pypy.git;a=blob;f=debian/patches/compiler-flags | 13:56 |
| tumbleweed | acceptable as is? Should they rather only come from translation cmdline options? | 13:56 |
| tumbleweed | Should I touch every posixish platforrm? | 13:56 |
| tumbleweed | I suppose testcases are wanted | 13:58 |
| _root_ (~Scalable@li252-14.members.linode.com) left irc: Ping timeout: 252 seconds | 14:05 | |
| dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) left irc: Quit: Leaving | 14:05 | |
| Nick change: csenger` -> csenger|away | 14:17 | |
| ltratt (~ltratt@87.114.126.76) joined #pypy. | 14:17 | |
| fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) joined #pypy. | 14:24 | |
| AlekSi (~AlekSi@176.57.74.133) joined #pypy. | 14:28 | |
| kenaan | 12arigo default 11c32fbc60b64c 15/pypy/: Move these three functions at interp-level: potential threading issues if they are written at app-level. | 14:28 |
| kenaan | 12arigo default 11598e02bd0eb8 15/pypy/module/_hashlib/interp_hashlib.py: Try to be more careful about free()ing this in case something fails. | 14:34 |
| CIA-17 | 03arigo 07roundup * 10#934/subprocess hangs in 1.7 (but not in 1.6 or 1.5): | 14:39 |
| CIA-17 | A Linux 64 version from today, 09532c746387, does not seem to have any problem | 14:39 |
| CIA-17 | any more, neither built normally nor with lldebug. I just got onc ... * 14https://bugs.pypy.org/issue934 | 14:39 |
| whitelynx (~whitelynx@63.241.75.144) joined #pypy. | 14:43 | |
| aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy. | 14:47 | |
| dmitrey (~quassel@77.121.195.200) joined #pypy. | 14:55 | |
| mat^2 (~mathias@212.130.113.35) joined #pypy. | 14:58 | |
| JaredWigmore (~jared@n112119177158.netvigator.com) joined #pypy. | 15:06 | |
| stakkars (~tismer@i59F77E49.versanet.de) left irc: Quit: schnarch | 15:07 | |
| Nick change: stakkars_ -> stakkars | 15:07 | |
| dmitrey (~quassel@77.121.195.200) left irc: Ping timeout: 260 seconds | 15:08 | |
| jaffachief (~textual@host86-170-15-228.range86-170.btcentralplus.com) left irc: Ping timeout: 244 seconds | 15:21 | |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection | 15:26 | |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 15:27 | |
| JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 15:29 | |
| gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy. | 15:30 | |
| Rhy0lite (dje@nat/ibm/x-ppcsaykmlnbhbldb) joined #pypy. | 15:31 | |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 244 seconds | 15:32 | |
| stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy. | 15:34 | |
| stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 15:34 | |
| Nick change: stakkars_ -> stakkars | 15:34 | |
| witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy. | 15:36 | |
| witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy. | 15:39 | |
| dmalcolm (david@nat/redhat/x-utvbryvimixqpokk) joined #pypy. | 15:41 | |
| anish_ (anish@nat/redhat/x-vpgockvpcwaidqro) left irc: Ping timeout: 248 seconds | 15:41 | |
| JaredWigmore (~jared@n112119177158.netvigator.com) left irc: Remote host closed the connection | 15:44 | |
| voidspace (~voidspace@python/psf/voidspace) left irc: Quit: Goodbye... | 15:47 | |
| dmitrey (~quassel@77.121.195.200) joined #pypy. | 15:47 | |
| mattip (~chatzilla@bzq-79-183-105-189.red.bezeqint.net) joined #pypy. | 15:49 | |
| mattip | hi. | 15:51 |
| arigato | hi | 15:51 |
| cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy. | 15:55 | |
| kenaan | 12mattip matrixmath-dot 113431bdf3f8a1 15/pypy/module/micronumpy/test/test_numarray.py: added test from numpy docstring | 15:55 |
| kenaan | 12mattip numpypy-reshape 111c041f97db71 15/: close useless branch | 15:55 |
| lesshaste | hi | 15:57 |
| mattip | fijal: ping | 15:57 |
| ltratt | hi all. is the only defined way to hold a weakref in RPython through RWeakValueDictionary and friends? | 15:57 |
| dmalcolm (david@nat/redhat/x-utvbryvimixqpokk) left irc: Quit: Leaving | 16:02 | |
| Alex_Gaynor | ltratt: no, import weakref; weakref.ref() works fine | 16:05 |
| Alex_Gaynor | (IIRC) | 16:05 |
| EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) joined #pypy. | 16:07 | |
| apoirier_away (~apoirier@sakura.nagare.org) joined #pypy. | 16:08 | |
| apoirier (~apoirier@sakura.nagare.org) left irc: Ping timeout: 260 seconds | 16:08 | |
| ltratt | ah, that's rpython friendly? excellent! | 16:09 |
| arigato | ltratt: "Supports ref() without callbacks" says the docstring of pypy.rlib.rweakref | 16:17 |
| mvt (~mvantelli@87.213.45.85) left irc: Quit: This computer has gone to sleep | 16:24 | |
| ltratt | arigato: i wasn't sure if that referred to rweakref or weakref | 16:25 |
| sacred_coder (~rambaho@202.3.77.231) left irc: Quit: Ex-Chat | 16:25 | |
| whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) joined #pypy. | 16:26 | |
| arigato | right | 16:27 |
| Action: arigato improves the docstring a bit | 16:27 | |
| ltratt | :) | 16:28 |
| ltratt | in my private repository, i have a patch which allows unlimited inlining of converge-level code (ok, it tends to crash because of bug #978, but it works for some very simple examples). | 16:30 |
| ltratt | it's not clear to me that unlimited inlining is always a great idea | 16:30 |
| kenaan | 12arigo stm 116eb15b35a17d 15/pypy/translator/stm/: add the stm_thread_id() helper | 16:30 |
| kenaan | 12arigo stm 113c5c86ba384f 15/pypy/module/transaction/__init__.py: More hacking... | 16:30 |
| kenaan | 12arigo stm 11db48523657f2 15/pypy/: Use thread_id() from stm instead of the one from threads. | 16:30 |
| kenaan | 12arigo stm 11e5329461f8ca 15/pypy/module/transaction/: Make the ExecutionContext again a thread-local. See comments. | 16:30 |
| kenaan | 12arigo default 11207092a8b272 15/pypy/rlib/rweakref.py: Improve the docstring and re-export weakref.ref() under the name pypy.rlib.rweakref.ref(). | 16:30 |
| voidspace (~voidspace@python/psf/voidspace) joined #pypy. | 16:30 | |
| ltratt | since a) one spends a long time tracing b) there's a greater chance of getting stuck in the trace interpreter in a very slow bytecode | 16:30 |
| arigato | yes, so far we solve this by carefully tuning the trace_limit parameter | 16:31 |
| arigato | if the trace becomes longer than this, then we retry by inlining less | 16:31 |
| ltratt | how is the less inlining done? | 16:32 |
| oal (u4126@gateway/web/irccloud.com/x-crkmvjyjqnsbdkgj) left irc: Remote host closed the connection | 16:33 | |
| bfirsh (u1308@gateway/web/irccloud.com/x-mtetdvqwlsuhbgdu) left irc: Remote host closed the connection | 16:33 | |
| ltratt | i'm not sure how my VM should know when a trace has been too long, and therefore how to inline less | 16:33 |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-yorxsgsorvedmskc) left irc: Remote host closed the connection | 16:34 | |
| rguillebert | hi | 16:34 |
| arigato | (yay, richards.py finally runs in parallel! running 1 iteration takes 6.96s, running 2 takes 7.61s) | 16:35 |
| arigato | (and 4 takes 8.18s) | 16:35 |
| rguillebert | with a GC ? | 16:35 |
| rguillebert | :) | 16:35 |
| arigato | of course not :-) | 16:35 |
| arigato | ltratt: that's automatic | 16:35 |
| ltratt | arigato: aha! i didn't realise that. that's very nice | 16:36 |
| arigato | when the jit metainterp records operation, it notices it gets too many of them, and retries by inlining less | 16:36 |
| ltratt | i now think i'm getting close to the JIT I want | 16:36 |
| arigato | "inlining" here is defined as a recursive call from the main jit loop to the main jit loop | 16:36 |
| ltratt | can that recursive call be indirect? | 16:37 |
| arigato | yes | 16:37 |
| ltratt | ok, then i guess it's doing the right thing already | 16:37 |
| arigato | :-) | 16:37 |
| ltratt | up until now, some short-running programs have been made worse by the JIT | 16:37 |
| ltratt | but i think that's partly due to crappy slot lookup code | 16:37 |
| mattip (chatzilla@bzq-79-183-105-189.red.bezeqint.net) left #pypy. | 16:37 | |
| ltratt | i've now pretty much fixed that | 16:37 |
| ltratt | on cvstone, for example, i'm now only 50% slower than pypy (rather than 100% before) | 16:38 |
| arigato | and richards.cv? :-) | 16:38 |
| ltratt | i don't really know what the effect will be on bigger programs until issue 978 is resolved, but i'm hopeful | 16:38 |
| ltratt | richards bumps into #978... | 16:38 |
| arigato | ok | 16:38 |
| ltratt | don't worry, though, pypy will still beat converge ;) | 16:40 |
| ltratt | but at least it will then be a fairer contest | 16:41 |
| arigato | (...even with 4 cores pypy-stm is still 3x slower at a regular "pypy --jit off"... right now it sounds like you'd need a 16-core machine to benefit) | 16:41 |
| arigato | s/slower at/slower than | 16:41 |
| arigato | ltratt: :-) | 16:42 |
| arigato | (hum, we still get a large number of aborts...) | 16:45 |
| arigato | or well, 212739, for a 10-seconds program... is that huge? or merely large? | 16:46 |
| arigato | ah | 16:54 |
| arigato | all RPython strings are read character-by-character via stm instead of directly | 16:55 |
| arigato | probably missing the fact that 'immutable' is true on a part of the STR lltype | 16:55 |
| Kaskuka (~John@osbk-4db14ccb.pool.mediaWays.net) left irc: Remote host closed the connection | 17:01 | |
| bgola (~bgola@c951826e.virtua.com.br) left irc: Ping timeout: 252 seconds | 17:01 | |
| Kaskuka (~John@osbk-4db14ccb.pool.mediaWays.net) joined #pypy. | 17:04 | |
| cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection | 17:06 | |
| bfirsh (u1308@gateway/web/irccloud.com/x-nsyfrckocafmnpsb) joined #pypy. | 17:06 | |
| Action: Da_Blitz thinks arigato deserves some kind of award for implementing STM quicker than everyone thought posible | 17:06 | |
| Taggnostr (~quassel@dyn57-362.yok.fi) left irc: Read error: Connection reset by peer | 17:07 | |
| Taggnostr (~quassel@dyn57-362.yok.fi) joined #pypy. | 17:11 | |
| rguillebert | Da_Blitz, "STM won't work" | 17:14 |
| voidspace | arigato: hey, you coming to pycon? | 17:17 |
| arigato | voidspace: hi! "likely" | 17:21 |
| voidspace | arigato: cool, will you be able to make the language summit? | 17:22 |
| arigato | yes, likely | 17:22 |
| voidspace | hehe, great | 17:22 |
| arigato | sorry for not being more precise right now :-) | 17:22 |
| voidspace | arigato: I'll put you down as a yes and if it turns out you can't come you can let me know | 17:22 |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-benrfcmzgaejjsps) joined #pypy. | 17:22 | |
| #pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services. | 17:22 | |
| arigato | voidspace: ok, thanks | 17:23 |
| AlekSi (~AlekSi@176.57.74.133) left irc: Read error: Connection reset by peer | 17:23 | |
| AlekSi (~AlekSi@176.57.74.133) joined #pypy. | 17:24 | |
| JaRoel (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection | 17:25 | |
| lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Read error: Operation timed out | 17:28 | |
| CIA-17 | 03Dmitrey 07roundup * 10#1008/numpypy: commit the code for empty_like, zeros_like, ones_like: | 17:32 |
| CIA-17 | I have pasted latest version of my empty_like to http://pastebin.com/yUznKBGV , | 17:32 |
| CIA-17 | as for ones_like and zeros_like - they were taken from numpy sour ... * 14https://bugs.pypy.org/issue1008 | 17:32 |
| apoirier_away (apoirier@sakura.nagare.org) left #pypy ("Leaving..."). | 17:34 | |
| G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving. | 17:34 | |
| tito (~tito@pdpc/supporter/active/tito) left irc: Ping timeout: 255 seconds | 17:34 | |
| tito (~tito@178.73.216.222) joined #pypy. | 17:36 | |
| tito (~tito@178.73.216.222) left irc: Changing host | 17:36 | |
| tito (~tito@pdpc/supporter/active/tito) joined #pypy. | 17:36 | |
| amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy. | 17:41 | |
| arigato | ok... the next target is to remove stm handling of all frame accesses in the main loop | 17:42 |
| arigato | this requires some care | 17:42 |
| arigato | in fact it may only require some care for sys._currentframes() | 17:42 |
| arigato | ah no, there is also an issue with generator frames | 17:43 |
| etrepum (~bob@75-101-96-144.dsl.static.sonic.net) left irc: Ping timeout: 240 seconds | 17:47 | |
| oal (u4126@gateway/web/irccloud.com/x-qcpgnudvfefbglmo) joined #pypy. | 17:50 | |
| arigato (~arigo@adsl-89-217-233-144.adslplus.ch) left irc: Ping timeout: 252 seconds | 17:53 | |
| etrepum (~bob@75-101-96-144.dsl.static.sonic.net) joined #pypy. | 17:59 | |
| etrepum (~bob@75-101-96-144.dsl.static.sonic.net) left irc: Ping timeout: 252 seconds | 18:09 | |
| JaRoel|4d (~jaroel|4d@sink.jaroel.nl) joined #pypy. | 18:13 | |
| etrepum (~bob@75.101.96.144) joined #pypy. | 18:14 | |
| Telugodu (~rambaho@202.3.77.231) joined #pypy. | 18:17 | |
| etrepum (~bob@75.101.96.144) left irc: Ping timeout: 252 seconds | 18:19 | |
| kenaan | 12amauryfa merge-2.7.2 119c27a0100918 15/pypy/module/_multibytecodec/: Fix multibytecodec incremental tests: ensure that .encode(final=True) is always called at the end. Also have... | 18:20 |
| Alex_Gaynor | amaury: So we now have: error messages, mmap, io, and site tests to deal with? | 18:21 |
| amaury | yes | 18:22 |
| amaury | I'll try to fix the visible failure in test_io | 18:22 |
| dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) joined #pypy. | 18:22 | |
| Alex_Gaynor | ok, I tried copying the fix from cpython for io, but it's another case of "that patch is insufficient" | 18:22 |
| amaury | test_write_rewind_write() | 18:22 |
| amaury | same here :) | 18:23 |
| asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy. | 18:23 | |
| Alex_Gaynor | hehe | 18:23 |
| amaury | time for debugging | 18:23 |
| amaury | fortunately pypy implementation mimics CPython's | 18:23 |
| amaury | so gdb here, pdb there | 18:23 |
| amaury | and step side by side | 18:23 |
| Alex_Gaynor | the pdb version is much nicer, IMO :) | 18:23 |
| Alex_Gaynor | fijal: ping | 18:28 |
| zain (~textual@bzq-79-182-196-50.red.bezeqint.net) left irc: Quit: Computer has gone to sleep. | 18:28 | |
| Telugodu (~rambaho@202.3.77.231) left irc: Remote host closed the connection | 18:30 | |
| Telugodu (~rambaho@202.3.77.231) joined #pypy. | 18:30 | |
| etrepum (~bob@75-101-96-144.dsl.static.sonic.net) joined #pypy. | 18:32 | |
| ericflo (~ericflo@75.103.8.110) joined #pypy. | 18:37 | |
| etrepum (~bob@75-101-96-144.dsl.static.sonic.net) left irc: Ping timeout: 245 seconds | 18:44 | |
| arigato (~arigo@adsl-89-217-233-144.adslplus.ch) joined #pypy. | 18:46 | |
| AlekSi (~AlekSi@176.57.74.133) left irc: Remote host closed the connection | 18:51 | |
| Alex_Gaynor | arigato: Did you fix the issue with immutable reads going through stm? | 18:53 |
| Alex_Gaynor | [unrelated] is it just me or should Metainterp.in_recursion really be call_depth or so | 18:53 |
| arigato | Alex_Gaynor: yes, it was just getinteriorfield on partially-immutable structures like STR | 18:55 |
| arigato | yes, in_recursion is really the app_level_recursion_depth | 18:55 |
| harrisonpartch (~sr@69.209.201.172) joined #pypy. | 18:56 | |
| voidspace (~voidspace@python/psf/voidspace) left irc: Quit: Goodbye... | 18:59 | |
| papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) joined #pypy. | 19:05 | |
| amaury | Alex_Gaynor: found the bug in test_write_rewind_write() | 19:08 |
| lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy. | 19:09 | |
| kenaan | 12amauryfa merge-2.7.2 11bf9a09ac155b 15/pypy/module/_io/: Test and fix in mixed read/write/seek/write operations | 19:11 |
| durin42 (~durin@adium/durin42) left irc: Ping timeout: 252 seconds | 19:12 | |
| durin42 (~durin@173-15-14-174-Illinois.hfc.comcastbusiness.net) joined #pypy. | 19:14 | |
| durin42 (~durin@173-15-14-174-Illinois.hfc.comcastbusiness.net) left irc: Changing host | 19:14 | |
| durin42 (~durin@adium/durin42) joined #pypy. | 19:14 | |
| Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 252 seconds | 19:18 | |
| CIA-17 | 03papercrane 07roundup * 10#1016/RPython AssertionError while running async io greenlet test: | 19:19 |
| CIA-17 | Yep, that error comes up on Linux as the errno's are different on Linux than on | 19:19 |
| CIA-17 | OSX. I got a version working on Linux but was unable to get the ... * 14https://bugs.pypy.org/issue1016 | 19:19 |
| Tobu (~Tobu@unaffiliated/tobu) joined #pypy. | 19:22 | |
| jokey (~jokey@gentoo/developer/jokey) joined #pypy. | 19:24 | |
| antocuni (~antocuni@host155-125-dynamic.25-79-r.retail.telecomitalia.it) joined #pypy. | 19:43 | |
| mattip (~chatzilla@bzq-79-183-105-189.red.bezeqint.net) joined #pypy. | 19:44 | |
| mattip | hi | 19:46 |
| mattip | anyone want to discuss logical ops on numpypy arrays? | 19:46 |
| mattip | http://paste.pocoo.org/show/539432/ | 19:46 |
| mattip | this is how numpy calculates logical_and | 19:47 |
| mattip | I'm wondering where it fits into the pypy infrastructure. | 19:47 |
| whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) left irc: Read error: Operation timed out | 19:49 | |
| mattip | is there already this functionality? | 19:50 |
| mattip | and I just need to import it | 19:50 |
| mattip | or do I need to create some new functions? | 19:50 |
| dmitrey | AFAIK some progress toward logical operations have been done | 19:51 |
| dmitrey | ask fijal for more details | 19:51 |
| mattip | dmitrey: thanks. | 19:51 |
| zain (~textual@bzq-79-179-136-94.red.bezeqint.net) joined #pypy. | 19:58 | |
| CIA-17 | 03arigo 07roundup * 10#1004/Sandbox only works for single-threaded code: | 20:01 |
| CIA-17 | [wontfix] There are other issues with threads, including the fact that we are slightly | 20:01 |
| CIA-17 | less sure that there are no attackable bugs in threaded co ... * 14https://bugs.pypy.org/issue1004 | 20:01 |
| lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 240 seconds | 20:02 | |
| leshaste (~lesshaste@87-194-206-189.bethere.co.uk) joined #pypy. | 20:11 | |
| leshaste | hi | 20:11 |
| mattip (chatzilla@bzq-79-183-105-189.red.bezeqint.net) left #pypy. | 20:15 | |
| zain (~textual@bzq-79-179-136-94.red.bezeqint.net) left irc: Quit: Computer has gone to sleep. | 20:22 | |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 20:23 | |
| amaury | How would you call a Lock object which raises RuntimeError when acquire() is called again from the owning thread? | 20:27 |
| amaury | NonReentrantLock? | 20:27 |
| exarkun | isn't that the default kind of lock? | 20:27 |
| amaury | (context: the _io module should be signal-safe) | 20:27 |
| amaury | no, the default lock deadlocks in this case | 20:27 |
| Alex_Gaynor | what does it do when acquired from another thread? | 20:28 |
| amaury | it waits | 20:28 |
| Alex_Gaynor | hmm, I'm not sure what to call it | 20:28 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Quit: pnguyen | 20:36 | |
| dmitrey (~quassel@77.121.195.200) left irc: Remote host closed the connection | 20:44 | |
| Rhy0lite (dje@nat/ibm/x-ppcsaykmlnbhbldb) left irc: Quit: Leaving | 20:46 | |
| squiddy (~squiddy@f053084057.adsl.alicedsl.de) left irc: Quit: Leaving | 20:47 | |
| fijal | Alex_Gaynor: pong | 20:48 |
| antocuni (~antocuni@host155-125-dynamic.25-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds | 20:48 | |
| fijal | arigato: ping | 20:49 |
| Alex_Gaynor | fijal: hey, do you have any opinions about exposing current JIT parameters? | 20:49 |
| Alex_Gaynor | ISTM the portal would allow this to be donesomehow | 20:49 |
| fijal | yeah | 20:49 |
| fijal | it's doable :) | 20:49 |
| fijal | a bit annoying though | 20:49 |
| fijal | I would not worry personally too much | 20:49 |
| fijal | for now | 20:49 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 20:50 | |
| Alex_Gaynor | I guess I can skip it, annoying though | 20:50 |
| CIA-17 | 03fijal 07roundup * 10#995/jit bug with threads: Fortunately still in /tmp/x.py on tannit ;-) http://paste.pocoo.org/show/539462/ * 14https://bugs.pypy.org/issue995 | 20:51 |
| Alex_Gaynor | fijal: I think I had anoyher thing, but I forgot | 20:51 |
| Alex_Gaynor | what are you working on ATM? | 20:52 |
| fijal | right now nothing | 20:53 |
| fijal | I was thinking about pushing numready and writing a short blog post that we did that | 20:53 |
| Alex_Gaynor | great, I caN HELP/REVIEW | 20:54 |
| Alex_Gaynor | oops | 20:54 |
| fijal | I don't think I need that | 20:54 |
| fijal | I'm also working on figuring out how financial aid works with registering for pycon | 20:54 |
| dmitrey (~quassel@77.121.195.200) joined #pypy. | 20:59 | |
| amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Quit: Quitte | 21:00 | |
| amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy. | 21:00 | |
| dmitrey (~quassel@77.121.195.200) left irc: Remote host closed the connection | 21:01 | |
| pjenvey | amaury - I'd call it 'a better Lock'.. =] how about TryLock | 21:01 |
| amaury | ok | 21:02 |
| amaury | I thought there was some official name for this | 21:02 |
| lizardo (lizardo@nat/indt/x-oyudojmhpvbyljkz) left irc: Quit: Leaving | 21:03 | |
| pjenvey | maybe but not that I can remember | 21:03 |
| amaury | In any case RPython code looks nicer than C | 21:05 |
| Alex_Gaynor | :D | 21:06 |
| amaury | instead of the ENTER_BUFFERED(self)/LEAVE_BUFFERED(self) macro pais | 21:06 |
| amaury | pairs | 21:06 |
| amaury | I have a nice "with" block | 21:06 |
| Alex_Gaynor | RPython is very nice somtimes | 21:10 |
| kenaan | 12amauryfa merge-2.7.2 11624143b36c8d 15/pypy/module/_io/: The underlying stream of a BufferedIO can call arbitrary code, or yield to some signal handler; protect agai... | 21:14 |
| amaury | here you are! | 21:14 |
| fijal | Alex_Gaynor: so yeah, numready | 21:14 |
| Alex_Gaynor | fijal: want me to make it less terrible? | 21:14 |
| fijal | Alex_Gaynor: mattip said flatiter branch is done (?) | 21:14 |
| fijal | want to make it work maybe first | 21:15 |
| Alex_Gaynor | ? | 21:15 |
| fijal | that was about which part? | 21:15 |
| Alex_Gaynor | the last two lines | 21:15 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1528 [12amaury, merge-2.7.2] | 21:15 |
| fijal | ah they're unrelated :) | 21:15 |
| fijal | I want to make numready work | 21:16 |
| fijal | and the flatiter is done if you want to have a look | 21:16 |
| fijal | Alex_Gaynor, amaury: thanks for helping with the release btw | 21:16 |
| Alex_Gaynor | sure, I can take a look | 21:16 |
| mvt (~mvt@53530442.cm-6-4a.dynamic.ziggo.nl) joined #pypy. | 21:17 | |
| Alex_Gaynor | fijal: to_coords in O(n) this doesn't seem right | 21:19 |
| Alex_Gaynor | fijal: do slices of a flatiter create a view or copy? | 21:22 |
| fijal | no idea | 21:25 |
| fijal | check? | 21:25 |
| Alex_Gaynor | sigh, stupid numpy | 21:26 |
| Alex_Gaynor | ok, looks like slices of flat iters do in fact copy | 21:26 |
| Alex_Gaynor | who the hell knows why | 21:27 |
| mattip (4fb769bd@gateway/web/freenode/ip.79.183.105.189) joined #pypy. | 21:27 | |
| mattip | hi | 21:29 |
| fijal | mattip: hi | 21:32 |
| mattip | I got bored and had a look at your reorganization of numpypy. I like. | 21:33 |
| mvt (~mvt@53530442.cm-6-4a.dynamic.ziggo.nl) left irc: Quit: ZNC - http://znc.in | 21:33 | |
| mattip | or would you rather have a whole new set of logical ops? | 21:34 |
| mattip | I noticed that there are some failing tests. Would you like help? | 21:34 |
| mvt (~mvt@53530442.cm-6-4a.dynamic.ziggo.nl) joined #pypy. | 21:35 | |
| fijal | mattip: what failing tests | 21:36 |
| mattip | I'm now away from that box, but I "hg up"ped to numyp-back-to-applevel, and played around. | 21:38 |
| mattip | as expected when I ran tests, some failed since it's not done yet, right? | 21:38 |
| mattip | (32 bit linux ubuntu) | 21:38 |
| fijal | ah right | 21:39 |
| fijal | back-to-applevel is horribly broken | 21:39 |
| fijal | thats' ok :) | 21:39 |
| mattip | so I can do one of four things now: | 21:41 |
| mattip | - try to add new logical functions | 21:42 |
| mattip | - try to work on numpy-back-to-applevel | 21:42 |
| mattip | - try to clean u the windows nightly build so it passes more tests | 21:42 |
| mattip | - get a good night'ssleep | 21:43 |
| mattip | anyone care to vote? | 21:43 |
| fijal | hahaha | 21:44 |
| fijal | I bet the last item is the best | 21:44 |
| fijal | as far as I'm concerned it makes most sense if you do 4) than 1) | 21:44 |
| mattip | cool. I'll take a look tomorrow. | 21:46 |
| mattip | Alex_Gaynor: I couldn't figure out a better way to calculate coords from indices since they do not map nicely to slices. | 21:48 |
| Alex_Gaynor | mattip: that also makes __getitem__ on flatiter O(n) though, right? | 21:49 |
| mattip | yes | 21:49 |
| mattip | :( | 21:49 |
| kenaan | 12fijal numpy-back-to-applevel 119c6ebf53fca0 15/: minor improvements | 21:49 |
| Alex_Gaynor | well, maybe not, if I'm reading correctly there's only a single iteration | 21:50 |
| mattip | no, there are two. | 21:50 |
| mattip | one in the calculation of the coords | 21:50 |
| mattip | and one to get the actual values | 21:50 |
| kenaan | 12fijal numpy-back-to-applevel 11f35e55900c6e 15/pypy/module/micronumpy/interp_boxes.py: a test commit | 21:51 |
| Alex_Gaynor | right | 21:51 |
| mattip | Can't we claim using an iterator for getitem is a non-recommended API in the docs somewhere? | 21:52 |
| fijal | no | 21:52 |
| fijal | getitem should do calculations | 21:52 |
| fijal | like item % shape[0], ... | 21:52 |
| andrewfr (~andrew@modemcable211.101-70-69.static.videotron.ca) joined #pypy. | 21:53 | |
| mattip | I created a separate function to do that: _to_coords | 21:53 |
| fijal | so what;'s wrong? | 21:54 |
| mattip | but then flatiter's getitem calls _to_coords - looping through the sequence - then call the underlying array's getitem with the list of coords - looping again. | 21:54 |
| Action: fijal is confused | 21:55 | |
| Alex_Gaynor | fijal: read the code :) | 21:55 |
| fijal | ok | 21:55 |
| mattip | _to_coords is in flatiter's getitem, not in BaseArray's getitem | 21:55 |
| aboudreault (~alanb@osgeo/member/aboudreault) left irc: Read error: Operation timed out | 21:55 | |
| fijal | can you paste the name of the branch including all typos? | 21:55 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Read error: Connection reset by peer | 21:55 | |
| hruske (~Gasper@188-230-156-183.dynamic.t-2.net) joined #pypy. | 21:56 | |
| Action: mattip blush | 21:56 | |
| mattip | numppy-flatitter | 21:57 |
| kenaan | 12fijal numpy-back-to-applevel 1127230fad5112 15/pypy/module/micronumpy/interp_boxes.py: add another test commit | 21:59 |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1528 [12amaury, merge-2.7.2] | 22:02 |
| stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 22:02 | |
| stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy. | 22:02 | |
| ltratt (~ltratt@87.114.126.76) left irc: Quit: using sirc version 2.211+KSIRC/1.3.12 | 22:05 | |
| fijal | ok read | 22:06 |
| fijal | this is ugly, but not wrong complexity no? | 22:06 |
| fijal | mattip: it's enough to convert start and stop | 22:06 |
| fijal | and then you can skip steps | 22:06 |
| fijal | just use an iterator | 22:06 |
| fijal | uh no | 22:07 |
| fijal | this is super ugly | 22:07 |
| mattip | what happens if start is (0,5) and stop is (1,3) ? | 22:07 |
| andrewfr (andrew@modemcable211.101-70-69.static.videotron.ca) left #pypy. | 22:07 | |
| fijal | mattip: you know how many numbers you want, don't you? | 22:08 |
| mattip | yes. | 22:08 |
| fijal | you create an iterator at position "start" | 22:08 |
| fijal | and you iterate it as many times as needed to get N elements | 22:08 |
| fijal | skipping steps | 22:08 |
| mattip | good. | 22:08 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 22:09 | |
| fijal | I think I don't care that much about iterating over steps | 22:09 |
| fijal | let's say | 22:09 |
| mattip | now I have something to do. There were 5 possibilities. | 22:09 |
| tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy. | 22:10 | |
| amaury | hugh | 22:10 |
| amaury | >>>> os.path.exists('/tmp\x00unlikely') | 22:10 |
| amaury | True | 22:10 |
| tumbleweed | fwiw, I've requested a sync of pypy into Ubuntu: https://launchpad.net/ubuntu/precise/+queue | 22:10 |
| fijal | hm | 22:11 |
| fijal | Alex_Gaynor: can you think about running numpy docstrings btw? | 22:11 |
| fijal | tumbleweed: cool! | 22:11 |
| Alex_Gaynor | fijal: I'm sure I can think about it, can you be more precise thoguh? | 22:11 |
| Alex_Gaynor | tumbleweed: FWIW I know Barry Warsaw has it in a ppa | 22:11 |
| fijal | tumbleweed: amaury and Alex_Gaynor are merging 2.7.2 and armin fixed some bugs | 22:11 |
| fijal | Alex_Gaynor: as in "can we run them as tests" | 22:11 |
| fijal | or can we export them as tests | 22:11 |
| fijal | either way | 22:11 |
| Alex_Gaynor | fijal: somehow, with doctest, probably | 22:12 |
| tumbleweed | Alex_Gaynor: yeah, I was wanting him to try and do an arm build. He works at canonical and so can get arm in his ppa | 22:12 |
| fijal | I would be maybe slightly happier with a file export | 22:12 |
| Alex_Gaynor | amaury: due to a bug, or you happen to have that file? | 22:13 |
| Alex_Gaynor | I guess if you use null padded str that'd be a bug | 22:13 |
| mattip | fijal: I think I lost your train of thought: "I think I don't care that much about iterating over steps; let's say" ? | 22:13 |
| amaury | try it yourself: | 22:13 |
| amaury | I have /tmp of course | 22:13 |
| amaury | and when /tmp\x00abcd is passed to open() or access() | 22:14 |
| Alex_Gaynor | amaury: pypy or cpython? | 22:14 |
| amaury | there is a zero in the middle of char* | 22:14 |
| amaury | four angles >>>> | 22:14 |
| fijal | mattip: I mean if you call iter.next() step times | 22:14 |
| amaury | pypy! | 22:14 |
| Alex_Gaynor | yay, pypy bug | 22:15 |
| amaury | CPython has a check for NUL bytes | 22:15 |
| fijal | Alex_Gaynor: there is another thing which I'm increasingly worried about | 22:15 |
| Alex_Gaynor | fijal: WRT what? | 22:15 |
| Alex_Gaynor | amaury: CPython should write a test, and we should fix it :) | 22:15 |
| amaury | it has one, in test_site | 22:15 |
| Alex_Gaynor | that's a silly place for it | 22:16 |
| fijal | Alex_Gaynor: we have iterators that are very costly to use if they're not jitted | 22:16 |
| amaury | A bit indirect I confess | 22:16 |
| Alex_Gaynor | fijal: is there any point in using them without the JIT? | 22:16 |
| fijal | to be precise we do immutable reallocation precisely because when they get jitted it's efficient | 22:16 |
| fijal | well, yes sometimes | 22:16 |
| Alex_Gaynor | amaury: AH, is this why this test fails? | 22:16 |
| fijal | or maybe | 22:16 |
| fijal | but that means sticking jitdrivers to places like flatiter.getitem | 22:16 |
| fijal | but also I have cases where it's inconvinient | 22:16 |
| Alex_Gaynor | fijal: well, I claim let's take it one step at a time | 22:17 |
| amaury | Alex_Gaynor: yes | 22:17 |
| fijal | so what I would need to do is to reallocate them without changing data | 22:17 |
| Alex_Gaynor | fijal: what do you mean? | 22:17 |
| Alex_Gaynor | you mean make .next() inplace? | 22:17 |
| fijal | I have in a loop: | 22:17 |
| fijal | if something: | 22:17 |
| fijal | iter = iter.next() | 22:17 |
| fijal | else: | 22:17 |
| fijal | do_nothing | 22:17 |
| fijal | so right now I would have to write iter = iter.next_do_nothing() | 22:18 |
| fijal | or something dumb like this | 22:18 |
| Alex_Gaynor | right, so they're both virtual | 22:18 |
| fijal | right | 22:18 |
| fijal | but this is super silly | 22:18 |
| Alex_Gaynor | a bit, yes | 22:18 |
| fijal | because I have to care about reallocating them | 22:18 |
| fijal | right now I just didn't do it | 22:18 |
| fijal | because it's code explosion | 22:19 |
| Alex_Gaynor | ok, we can always make changes later | 22:19 |
| fijal | so I stuck with the simples possible version for now | 22:19 |
| Alex_Gaynor | it's not like it isn't pretty easy to spot in traces | 22:19 |
| fijal | no | 22:19 |
| fijal | also bridges | 22:19 |
| fijal | and whatnot :) | 22:19 |
| fijal | but yes, later | 22:19 |
| Alex_Gaynor | later | 22:20 |
| Alex_Gaynor | for now, let's get stuff working, and have core stuff fast | 22:20 |
| fijal | I already have a few jit optimizations in mind | 22:20 |
| fijal | well | 22:20 |
| fijal | it's not fast as of now | 22:20 |
| Alex_Gaynor | give them to hakanardo :) | 22:20 |
| azanella (~azanella@189.6.80.131) left irc: Ping timeout: 244 seconds | 22:20 | |
| fijal | some axis stuff or something like that increases iterators sometimes | 22:20 |
| Alex_Gaynor | ? | 22:25 |
| Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 272 seconds | 22:25 | |
| mattip | arggh. numpy can do this: a.flat[2,3,-7, 0] | 22:26 |
| fijal | what does it do? | 22:26 |
| Alex_Gaynor | really? | 22:26 |
| Alex_Gaynor | for me it raises an Exception | 22:26 |
| mattip | numpy.version.version =='1.3.0' | 22:27 |
| Alex_Gaynor | that's old :) | 22:27 |
| mattip | a=numpy.arange(12).reshape(3,4).T | 22:28 |
| Alex_Gaynor | I'm on 1.6 | 22:28 |
| mattip | centos 6.1 | 22:28 |
| mikefc__ (3cf2f0c4@gateway/web/freenode/ip.60.242.240.196) joined #pypy. | 22:28 | |
| fijal | Alex_Gaynor: that's old :) | 22:29 |
| Alex_Gaynor | heh | 22:29 |
| mattip | well good then, can we just not support uneven steps within the sequence? | 22:30 |
| fijal | "uneven steps"? | 22:31 |
| mattip | yes: a.flat[2,3,7,8] | 22:31 |
| fijal | it does not work on a recent numpy, what's the problem? | 22:31 |
| Alex_Gaynor | flat doesn't support tuple indexing in getitem at all in 1.6 | 22:32 |
| Alex_Gaynor | (well, excpet 0 or 1-item tuples) | 22:32 |
| Action: mattip has to upgrade :( | 22:32 | |
| mattip | someone today asked what numpy API is our target, do we have an answer we can stick in a WIKI page? | 22:33 |
| hruske (~Gasper@188-230-156-183.dynamic.t-2.net) left irc: Quit: Leaving | 22:36 | |
| Tobu (~Tobu@unaffiliated/tobu) joined #pypy. | 22:37 | |
| tumbleweed | poke again: Committable? http://anonscm.debian.org/gitweb/?p=collab-maint/pypy.git;a=blob;f=debian/patches/compiler-flags | 22:39 |
| harrisonpartch (~sr@69.209.201.172) left irc: Remote host closed the connection | 22:39 | |
| Alex_Gaynor | tumbleweed: It looks ok to me, but run it by fijal or someone | 22:40 |
| fijal | +1 | 22:40 |
| fijal | freebsd buildbot never worked so I really don' | 22:40 |
| fijal | t care | 22:40 |
| tumbleweed | thanks | 22:41 |
| tumbleweed | there are still the dead cflags and ldflags translation command line arguments, which should probably be removed or used | 22:41 |
| kenaan | 12fijal default 11a56e47f860e4 15/pypy/module/micronumpy/tool/numready.py: This file belongs here for now | 22:41 |
| kenaan | 12fijal buildbot 11fdaf7c879572 15/: remove numready from here | 22:42 |
| Alex_Gaynor | fijal: ok, I'm going to try to make numready be less awful now | 22:42 |
| Alex_Gaynor | well, tonight | 22:42 |
| aboSamoor (~rmyeid@nbl.cewit.stonybrook.edu) left irc: Read error: Operation timed out | 22:42 | |
| kenaan | 12stefanor default 11173aa3e5cde0 15/pypy/translator/platform/: Get additional compiler and linker flags from the environment (rather than ignoring them on linux, and replacing... | 22:43 |
| amaury | pypy should reject filenames with NUL bytes | 22:43 |
| amaury | this could be done in ll_os.py and others, but it may have performance impact | 22:44 |
| amaury | non jittable functions for example | 22:44 |
| mattip (4fb769bd@gateway/web/freenode/ip.79.183.105.189) left irc: Quit: Page closed | 22:44 | |
| Alex_Gaynor | amaury: shrug, so it'll have a performance impact, that's life :) | 22:44 |
| aboSamoor (~rmyeid@nbl.cewit.stonybrook.edu) joined #pypy. | 22:45 | |
| arigato | shouldn't it be done in str2charp() directly? | 22:45 |
| Alex_Gaynor | the problem is str2charp cuts things off, that's the point | 22:46 |
| amaury | not always | 22:46 |
| amaury | os.write() probably uses str2charp as well | 22:46 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Quit: pnguyen | 22:46 | |
| Alex_Gaynor | no, it probably uses str2charpn | 22:47 |
| Alex_Gaynor | or so | 22:47 |
| Alex_Gaynor | since it's valid to have a NUL in a write clal | 22:47 |
| amaury | actually, it uses rffi.get_nonmovingbuffer | 22:47 |
| amaury | so you may be right | 22:47 |
| Alex_Gaynor | fijal: uhhh, your commit to builds.py is a syntaxError | 22:48 |
| bja (~bja@69.11.216.98) joined #pypy. | 22:48 | |
| tumbleweed | while we're talking about not caring about freebsd, here's another pair that I needed to get pypy behaving on Debian GNU/kFreeBSD http://anonscm.debian.org/gitweb/?p=collab-maint/pypy.git;a=blob;f=debian/patches/kfreebsd-glibc http://anonscm.debian.org/gitweb/?p=collab-maint/pypy.git;a=blob;f=debian/patches/kfreebsd-test-hangs | 22:51 |
| aboSamoor (~rmyeid@nbl.cewit.stonybrook.edu) left irc: Ping timeout: 240 seconds | 22:51 | |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 22:51 | |
| tumbleweed | sys.platform = gnkfreebsdX so one wants startswith freebsd when checking for FreeBSD's libc, and 'freebsd' in for checking for a FreeBSD kernel | 22:52 |
| Alex_Gaynor | tumbleweed: in those expressions, can you use any() | 22:52 |
| Alex_Gaynor | it reads much more cleanly to me that way | 22:52 |
| Alex_Gaynor | (first patch) | 22:52 |
| Alex_Gaynor | second patch is fine | 22:53 |
| tumbleweed | sure | 22:53 |
| aboSamoor (~rmyeid@nbl.cewit.stonybrook.edu) joined #pypy. | 22:54 | |
| tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection | 22:54 | |
| fijal | Alex_Gaynor: uhhhh | 22:54 |
| Alex_Gaynor | fijal: ? | 22:54 |
| fijal | Alex_Gaynor: I did not mean to commit there :/ | 22:54 |
| Alex_Gaynor | fijal: uncommit it! | 22:54 |
| kenaan | 12fijal buildbot 11166c710d4406 15/bot2/pypybuildbot/builds.py: this was not supposed to be commited | 22:55 |
| fijal | done | 22:55 |
| fijal | arigato: seen my bug report update? | 22:55 |
| amaury | bah | 22:55 |
| amaury | raise OperationError(space.w_TypeError) from rffi.str2charp is not easy | 22:56 |
| fijal | no | 22:56 |
| Alex_Gaynor | amaury: I don't think it's correct there anyways | 22:56 |
| fijal | amaury: what we do is we raise TypeError and reraise it as OperationError at the call site | 22:56 |
| Alex_Gaynor | It's not correct to do it in rffi.str2charp AFAICT | 22:57 |
| amaury | or have a specific RPython exception for this | 22:57 |
| amaury | Alex_Gaynor: where would you do it | 22:57 |
| fijal | for example | 22:57 |
| amaury | in all posix functions? | 22:57 |
| fijal | hm | 22:57 |
| Alex_Gaynor | amaury: probably at each specific callsite where it should be checked | 22:57 |
| Alex_Gaynor | :/ | 22:57 |
| Alex_Gaynor | maybe not | 22:57 |
| fijal | shouldn't we just use rffi.str2charpn always? | 22:57 |
| Alex_Gaynor | fijal: there is no sr2charpn I'm stupid | 22:58 |
| amaury | CPython checks this in the "s" argument code | 22:58 |
| Alex_Gaynor | str2charp already has the length (of course) | 22:58 |
| fijal | uh right | 22:58 |
| fijal | amaury: maybe we can check it in "str" argument then? | 22:58 |
| aboSamoor (~rmyeid@nbl.cewit.stonybrook.edu) left irc: Ping timeout: 245 seconds | 22:58 | |
| dripton (~dripton@pool-72-86-45-238.clppva.fios.verizon.net) joined #pypy. | 22:58 | |
| fijal | it sounds a bit obscure if you ask me that strp2charp can raise something else than MemoryError | 22:58 |
| Alex_Gaynor | it's not correct in str argument, and we have no char* argument type | 22:59 |
| amaury | and use "bufferstr" when NUL is allowed | 22:59 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/785 | 23:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/214 | 23:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/1024 | 23:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/41 | 23:00 |
| aboSamoor (~rmyeid@130.245.191.56) joined #pypy. | 23:00 | |
| fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) left irc: Quit: happier days, hopefully! | 23:01 | |
| kenaan | 12amauryfa default 1136fefb3d271e 15/pypy/: a, b = space.unpackiterable(w_obj, 2) is not supposed to raise ValueError | 23:01 |
| kenaan | 12stefanor default 1126b2fb7939ee 15/pypy/: Debian kFreeBSD uses glibc Check for sys.platform.startswith("freebsd") when the check is for FreeBSD's userlan... | 23:02 |
| kenaan | 12stefanor default 1167ae2044a17f 15/pypy/module/: Test hangs on kfreebsd | 23:02 |
| fijal | http://paste.pocoo.org/show/539524/ | 23:04 |
| fijal | Alex_Gaynor: ? | 23:04 |
| Alex_Gaynor | fijal: looks like buildbot nonsense to me :) | 23:05 |
| Alex_Gaynor | I suppose it works though | 23:05 |
| asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Ping timeout: 255 seconds | 23:05 | |
| fijal | http://paste.pocoo.org/show/539526/ | 23:05 |
| kenaan | 12fijal buildbot 11fbcf930f4155 15/bot2/pypybuildbot/builds.py: try to upload numpy compatibility website when doing benchmarks | 23:06 |
| amaury | In CPython, PyArg_ParseTuple(args, "s", ...) reject NUL bytes | 23:07 |
| amaury | PyArg_ParseTuple(args, "s#", ...) accepts NUL bytes | 23:07 |
| amaury | but you must pass a variable to hold the length | 23:07 |
| fijal | that's a very comprehensive API description | 23:07 |
| fijal | almost like immutable_fields | 23:08 |
| amaury | in pypy, we use unwrap_spec(x=str) in both cases | 23:08 |
| amaury | for example, inet_aton() and inet_ntoa() | 23:09 |
| fijal | how about bufferstr? | 23:09 |
| amaury | yes | 23:09 |
| Alex_Gaynor | bufferstr is a 3rd thing | 23:09 |
| amaury | but every function must be checked | 23:09 |
| Alex_Gaynor | isn't it? | 23:09 |
| amaury | in CPython, s# also works for buffer-compatible objects | 23:10 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Quit: pnguyen | 23:11 | |
| Alex_Gaynor | perhaps we should just add a new one | 23:11 |
| amaury | which is? | 23:12 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 23:12 | |
| fijal | nonnullstr | 23:12 |
| fijal | ? | 23:12 |
| amaury | what for? | 23:12 |
| fijal | not sure | 23:12 |
| fijal | no idea | 23:12 |
| fijal | how many cpython has? | 23:12 |
| amaury | "s": string with no NUL -> char* | 23:13 |
| amaury | "s#": any -> char*, size_t | 23:13 |
| amaury | "s*": buffer-like -> struct Py_Buffer | 23:13 |
| amaury | "s#": buffer-like -> struct Py_Buffer | 23:14 |
| amaury | (the 2nd was incorrect) | 23:14 |
| Alex_Gaynor | CPython is scary | 23:14 |
| amaury | err, sorry again | 23:14 |
| fijal | what's the difference between s# and s*? | 23:14 |
| fijal | yes please :) | 23:14 |
| amaury | "s#": buffer-like -> char*, ssize_t | 23:14 |
| amaury | "s*": buffer-like -> struct Py_Buffer | 23:15 |
| amaury | ok | 23:15 |
| tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy. | 23:22 | |
| amaury | we don't really care of Py_buffer, since RPython strings have a length | 23:22 |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Quit: pnguyen | 23:22 | |
| pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy. | 23:24 | |
| zain (~textual@bzq-79-179-136-94.red.bezeqint.net) joined #pypy. | 23:26 | |
| bja (~bja@69.11.216.98) left irc: Read error: Connection reset by peer | 23:28 | |
| bja (~bja@69.11.216.98) joined #pypy. | 23:31 | |
| whitelynx (~whitelynx@63.241.75.144) left irc: Quit: Ex-Chat | 23:34 | |
| voidspace (~voidspace@87-194-212-65.bethere.co.uk) joined #pypy. | 23:47 | |
| voidspace (~voidspace@87-194-212-65.bethere.co.uk) left irc: Changing host | 23:47 | |
| voidspace (~voidspace@python/psf/voidspace) joined #pypy. | 23:47 | |
| arigato (~arigo@adsl-89-217-233-144.adslplus.ch) left irc: Ping timeout: 252 seconds | 23:53 | |
| amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 252 seconds | 23:59 | |
| --- Tue Jan 24 2012 | 00:00 | |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!