| fermianyon (~lane@c-71-229-21-197.hsd1.al.comcast.net) left irc: Ping timeout: 240 seconds | 00:11 | |
| Ademan (~dan@adsl-71-141-224-79.dsl.snfc21.pacbell.net) joined #pypy. | 00:14 | |
| cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy. | 00:16 | |
| cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection | 00:16 | |
| papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) joined #pypy. | 00:21 | |
| papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep. | 00:30 | |
| justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Ping timeout: 252 seconds | 00:38 | |
| aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds | 00:55 | |
| aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy. | 01:05 | |
| cocoatomo (~cocoatomo@157.82.5.148) joined #pypy. | 01:06 | |
| cocoatomo (~cocoatomo@157.82.5.148) left irc: Remote host closed the connection | 01:07 | |
| ThomasWaldmann | ttps://bitbucket.org/mchaput/whoosh/issue/207/with-patch-pypy-and-whoosh-too-many-open solved. just in case someone else is having "out of files" issues: do not just look for open() and file(), but also for mmap()! | 01:11 |
|---|---|---|
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 01:12 | |
| aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds | 01:19 | |
| mfoord_ (~anonymous@python/psf/voidspace) joined #pypy. | 01:23 | |
| dolo | exit | 01:23 |
| dolo (~jfysh@ppp59-167-188-51.static.internode.on.net) left irc: Quit: ircII EPIC5-1.1.2 -- Are we there yet? | 01:23 | |
| mfoord (~anonymous@python/psf/voidspace) left irc: Ping timeout: 240 seconds | 01:25 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy. | 01:26 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host | 01:26 | |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 01:26 | |
| aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy. | 01:27 | |
| mfoord_ (~anonymous@python/psf/voidspace) left irc: Ping timeout: 240 seconds | 01:29 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/205 | 01:31 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/33 | 01:35 |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/777 | 01:52 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1926 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/679 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/339 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/329 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/677 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1196 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/860 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1513 | 02:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/790 | 02:00 |
| tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection | 02:12 | |
| mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord | 02:19 | |
| fermianyon (~lane@c-71-229-21-197.hsd1.al.comcast.net) joined #pypy. | 02:23 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Ping timeout: 244 seconds | 02:58 | |
| Kaskuka (~John@osbk-d9bf2242.pool.mediaWays.net) left irc: Ping timeout: 240 seconds | 03:04 | |
| Kaskuka (~John@osbk-4db163b1.pool.mediaWays.net) joined #pypy. | 03:04 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/1014 | 03:12 |
| dracman (~draco@212.255.21.252) left irc: Ping timeout: 252 seconds | 03:19 | |
| dracman (~draco@212.255.29.213) joined #pypy. | 03:25 | |
| justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) joined #pypy. | 03:31 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 03:40 | |
| justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Ping timeout: 252 seconds | 03:45 | |
| Nick change: forrestv -> forrestoshi | 03:49 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1926 | 04:13 |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 245 seconds | 04:27 | |
| justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) joined #pypy. | 04:31 | |
| tmu (~tmu@evil.oulu.nixu.fi) left irc: Ping timeout: 248 seconds | 04:33 | |
| tmu (~tmu@evil.oulu.nixu.fi) joined #pypy. | 04:34 | |
| mwhudson (~mwh@linaro/mwhudson) left irc: Ping timeout: 260 seconds | 04:37 | |
| nettok (~quassel@190.149.184.238) left irc: Ping timeout: 260 seconds | 04:55 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Computer has gone to sleep. | 04:59 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 05:10 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Client Quit | 05:12 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/860 | 05:18 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1513 | 05:19 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/677 | 05:30 |
| anish_ (anish@nat/redhat/x-fmpltnxpvuwwxlhc) joined #pypy. | 05:35 | |
| justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Ping timeout: 260 seconds | 05:35 | |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1196 | 05:36 |
| bbot2 | 3Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/679 | 05:46 |
| Justin_ (~Justin@114-34-52-12.HINET-IP.hinet.net) joined #pypy. | 05:50 | |
| Justin_ | hey guys, just wondering if there's any known issues running pypy with jinja2? I am able to get the setup running but performance is degraded significantly | 05:55 |
| Justin_ | I ran a few sampling profiles, numbers look like this | 05:56 |
| Justin_ | jinja2 (350ms using pypy, 35ms using python) | 05:57 |
| Justin_ | *cpython | 05:57 |
| Justin_ | django (700ms using pypy, 5s using cpython) | 05:58 |
| Justin_ | so using django's built in template engine, pypy has significant improvement | 05:58 |
| Justin_ | but using jinja2, performance is somehow degraded significantly | 05:58 |
| Justin_ | any ideas? | 05:59 |
| DasIch (~DasIch@p3E991405.dip.t-dialin.net) left irc: Quit: DasIch | 06:15 | |
| Arach (~arach@128-75-174-161.broadband.corbina.ru) left irc: Ping timeout: 260 seconds | 06:26 | |
| mikefc__ (9665bcc5@gateway/web/freenode/ip.150.101.188.197) joined #pypy. | 06:27 | |
| mikefc_ (9665bcc5@gateway/web/freenode/ip.150.101.188.197) left irc: Ping timeout: 258 seconds | 06:28 | |
| papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) joined #pypy. | 06:37 | |
| whyking (~quassel@p4FFB7B5C.dip.t-dialin.net) joined #pypy. | 06:45 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/790 | 06:46 |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/339 | 06:48 |
| mat^2 (~mathias@212.130.113.35) left irc: | 06:50 | |
| papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep. | 07:08 | |
| Justin_ | guess its too early of the day to be asking questions | 07:11 |
| Justin_ | :\ | 07:11 |
| JaRoel|4d (~jaroel|4d@sink.jaroel.nl) left irc: Remote host closed the connection | 07:30 | |
| mikefc__ | you can try | 07:35 |
| fermianyon (~lane@c-71-229-21-197.hsd1.al.comcast.net) left irc: Read error: Connection reset by peer | 07:42 | |
| bbot2 | 4Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/329 | 07:43 |
| Justin_ | i tried | 07:50 |
| Da_Blitz | Justin_: iirc jinja2 has cpython specifc hacks | 07:50 |
| Justin_ | don't want to repeat it | 07:51 |
| Da_Blitz | and uses c libs | 07:51 |
| Justin_ | but basically i have slower performance result mixing jinja2 with pypy | 07:51 |
| Justin_ | Da_Blitz: got a link? | 07:51 |
| Da_Blitz | i would assume it disables these hacks when not running on cpython | 07:52 |
| Da_Blitz | no, it comes up every now and again | 07:52 |
| Da_Blitz | didnt anwser earlier as i cant remeber the exact issues that cause it | 07:52 |
| Justin_ | oh i see | 07:52 |
| Da_Blitz | there may be a bug on the bug tracker | 07:52 |
| Da_Blitz | or somthing on the compatibility wiki if you want to take a closer look at it | 07:52 |
| Da_Blitz | AFAIK its not a pypy issue | 07:53 |
| Justin_ | i checked compatibility wiki already | 07:53 |
| Justin_ | under jinja2 it just says compatible | 07:54 |
| Justin_ | not much information | 07:54 |
| Da_Blitz | i would recomend idling, someone may clarify later | 07:54 |
| Justin_ | hmm ok i will hang around for a few more hours | 07:54 |
| Justin_ | i am gonna try and deploy this to my ubuntu box meanwhile | 07:54 |
| Justin_ | i think there's reports saying pypy being slow on osx | 07:55 |
| Justin_ | i did fine-tune some configs but performance didn't really change | 07:55 |
| Justin_ | but jinja2 (350ms using pypy, 35ms using cpython) and django (700ms using pypy, 5s using cpython) | 07:56 |
| Justin_ | these numbers really suggest something is not mixing well between jinja2 and pypy | 07:57 |
| Da_Blitz | most development is done on linux, so any help on other platforms is always apreciated | 07:57 |
| Justin_ | they both increased performance significantly alone | 07:57 |
| Da_Blitz | most of the devs are in the EU so they should be on later | 07:57 |
| mikefc__ | pypy runs fine on OSX. | 08:01 |
| mikefc__ | speedwise | 08:01 |
| Justin_ | it probably does if you have the right set of configurations | 08:04 |
| mikefc__ | i haven't translated it myself. I'm just using the nightly build | 08:05 |
| Da_Blitz | so you are basing pypy being slow purely based on jinja2 being slow? | 08:05 |
| Da_Blitz | or do you have other benchmarks that are also significantly slower? | 08:06 |
| Justin_ | no the issue is only when I mix the two | 08:06 |
| Justin_ | i created a sample test, generating 10x10000 tables | 08:06 |
| Justin_ | the result is like i said earlier | 08:07 |
| Justin_ | if i use django's built in template engine, it takes 5s to render, using pypy it takes 700ms | 08:07 |
| Justin_ | so improvement is quite significant | 08:07 |
| Justin_ | however if i run the same test using jinja2 template engine, it takes 350ms to render using pypy and 35ms to render using cpython | 08:08 |
| Da_Blitz | which is exactly what i would expect on all platforms | 08:08 |
| Justin_ | so jinja2 just works better with cpython? | 08:09 |
| Justin_ | i am not sure how it works behind the scenes | 08:09 |
| Da_Blitz | as i said earlier, it uses a whole bunch of cpython specifc hacks for speed | 08:09 |
| Da_Blitz | these hacks dont work on pypy | 08:09 |
| Da_Blitz | pypy also only speeds up python code | 08:09 |
| Justin_ | but 10x slower? | 08:09 |
| Justin_ | i would expect same or slightly better performance | 08:10 |
| Da_Blitz | cpython extensions are done via a compatability layer which may slow things down | 08:10 |
| Da_Blitz | the thing with expectations is they are not always correct | 08:11 |
| Da_Blitz | always better to mesure | 08:11 |
| Justin_ | hmm ok i think i am going to profile pypy + jinja2 a bit more in depth | 08:12 |
| Justin_ | maybe with cprofile | 08:12 |
| Justin_ | s | 08:12 |
| Justin_ | see where the actual bottlenecks are coming from | 08:12 |
| mikefc__ | cprofile isn't that great for pypy profiling | 08:12 |
| Justin_ | maybe not the numbers but actual queries | 08:12 |
| mikefc__ | fair enough. | 08:13 |
| mikefc__ | I don't know why cprofile is slower. http://stackoverflow.com/questions/7063508/pypy-significantly-slower-than-python | 08:13 |
| mikefc__ | but it's "known" | 08:13 |
| Justin_ | right i am aware of this actually | 08:14 |
| mikefc__ | OK. Just making sure :) | 08:14 |
| mikefc__ (9665bcc5@gateway/web/freenode/ip.150.101.188.197) left irc: Quit: Page closed | 08:14 | |
| fijal | hi | 08:26 |
| fijal | Justin_: can I have the template? | 08:29 |
| Justin_ | sure& its only a for loop though | 08:30 |
| Justin_ | hang on | 08:30 |
| Justin_ | this is the template | 08:32 |
| Justin_ | http://pastebin.com/meyX2L2k | 08:32 |
| Justin_ | this is the view | 08:33 |
| Justin_ | http://pastebin.com/XLa0ThaD | 08:33 |
| bivab (~david@fwstups.cs.uni-duesseldorf.de) joined #pypy. | 08:39 | |
| Justin_ | you're not a troll are you :D | 08:40 |
| fijal | me? | 08:42 |
| fijal | Justin_: where do I import stuff from? | 08:43 |
| fijal | ImportError: cannot import name 'render_to_response' | 08:45 |
| fijal | Justin_: ? | 08:45 |
| Justin_ | .... | 08:47 |
| Justin_ | this is built in django | 08:48 |
| kenaan | 12bivab arm-backend-2 116b5ffc5a9e85 15/pypy/jit/backend/arm/: add (set|get)_interiorfield_raw methods | 08:48 |
| kenaan | 12bivab arm-backend-2 11a317879ec998 15/pypy/jit/backend/arm/: print some information when hitting a missing operation in the backend | 08:48 |
| fijal | Justin_: yeah, ok, can you paste the entire thing and not leave me guessing? | 08:49 |
| fijal | Justin_: I would like to be able to run it | 08:50 |
| fijal | hakanardo: ping | 08:50 |
| Justin_ | from django.shortcuts import render_to_response | 08:50 |
| Justin_ | from django.template.context import RequestContext | 08:51 |
| Justin_ | i think thats about it | 08:51 |
| fijal | well | 08:52 |
| fijal | what do I pass as request? | 08:52 |
| fijal | Justin_: seriously this is not really a "runnable benchmark" | 08:53 |
| fijal | ideally I want a command line tool that I can run like this: | 08:53 |
| fijal | python x.py | 08:53 |
| fijal | and measure time | 08:53 |
| fijal | now I have to write main loop on my own and figure out how to pass django request | 08:53 |
| fijal | ImportError: Settings cannot be imported, because environment variable DJANGO_SETTINGS_MODULE is undefined. | 08:56 |
| fijal | Justin_: etc etc. | 08:56 |
| fijal | Justin_: why do you assume I know how to use django? | 08:56 |
| Justin_ | here | 08:58 |
| Justin_ | python -c "from struct import pack; print pack('5b', (41*len('99')), | 08:58 |
| Justin_ | pow(8,2)+20, 4900**0.5, range(78)[-1], 10)" | 08:58 |
| Justin_ | run that in command line you'll get your answer | 08:58 |
| fijal | Justin_: look, you come here complaining that something renders slowly | 09:00 |
| fijal | it's relatively expected that I can potentially help you, but seriously, not by reading django manual | 09:00 |
| fijal | I have much better stuff to do than read django manual now | 09:00 |
| fijal | either you can cooperate and provide a slow standalone program or you don't | 09:01 |
| fijal | "jinja2 rendering is slow" is a tad misleading | 09:01 |
| fijal | because you didn't even bother isolating jinja2 rendering | 09:01 |
| fijal | you still use pieces of django that I have no idea how work | 09:01 |
| Justin_ | ok sorry I might've' come off too strong there | 09:01 |
| Justin_ | but problem is, what you're asking is matter of rendering all the tables, it doesn't have anything to do with benchmarking | 09:02 |
| fijal | Justin_: seriously, if you come up with a command line tool that I can run, I can help you | 09:02 |
| fijal | how do I know? | 09:02 |
| fijal | and how do you know in the first place? | 09:02 |
| fijal | how do you know it's not add filter which is jinja non-default one is not at fault? | 09:02 |
| fijal | it requires quite some forensics to figure out what you really want | 09:02 |
| Justin_ | point is to install the right environments and just test from there | 09:02 |
| fijal | I have no idea | 09:03 |
| fijal | if you tell me how to achieve that, I can probably do that | 09:03 |
| fijal | commandline is better than on server btw | 09:03 |
| Justin_ | ok don't worry about it, thanks for offering to help anyway | 09:04 |
| mher (~Adium@tethras5.ucd.ie) joined #pypy. | 09:07 | |
| whyking (~quassel@p4FFB7B5C.dip.t-dialin.net) left irc: Ping timeout: 252 seconds | 09:09 | |
| fijal_ (~fijal@196-210-150-33.dynamic.isadsl.co.za) joined #pypy. | 09:25 | |
| fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) left irc: Ping timeout: 240 seconds | 09:25 | |
| Nick change: fijal_ -> fijal | 09:25 | |
| arigato (~arigo@89-40.5-85.cust.bluewin.ch) joined #pypy. | 09:31 | |
| rguillebert (~rguillebe@89-40.5-85.cust.bluewin.ch) joined #pypy. | 09:33 | |
| squiddy (~squiddy@g224197249.adsl.alicedsl.de) joined #pypy. | 09:36 | |
| matthp (c1a9466d@gateway/web/freenode/ip.193.169.70.109) joined #pypy. | 09:36 | |
| arigato | hi | 09:39 |
| fijal | hi armin | 09:39 |
| fijal | arigato: I have a crazy idea, want to listen? | 09:40 |
| fijal | :) | 09:40 |
| arigato | :-) | 09:40 |
| arigato | yes | 09:40 |
| matthp (c1a9466d@gateway/web/freenode/ip.193.169.70.109) left irc: Client Quit | 09:40 | |
| fijal | so, what would happen if we do the same thing as we're doing for bridges (short preamble inlining) for call_assembler? | 09:40 |
| fijal | would we be able to jump directly to the optimized loop and maybe not even allocate the frame? | 09:40 |
| Arach (~arach@128-75-182-202.broadband.corbina.ru) joined #pypy. | 09:41 | |
| arigato | I don't know | 09:42 |
| G2P (~G2P@fw-asn1.ornis.com) joined #pypy. | 09:43 | |
| arigato | sorry, that's about as much as I can say right now | 09:44 |
| fijal | ok | 09:44 |
| fijal | anyway, it's a food for thought for now :) | 09:44 |
| arigato | :-) | 09:44 |
| fijal | I don't know either, but it sounds like a good way to approach it | 09:44 |
| fijal | to inline the short preamble into call_assembler | 09:44 |
| arigato | what occurs if the guards from the short preamble fail? | 09:45 |
| fijal | we compile a bridge to the long preamble? | 09:46 |
| fijal | same as with bridges no? | 09:47 |
| antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) joined #pypy. | 09:47 | |
| fijal | arigato: note that as for numpy most of those guards are for example completely redundant but we don't have enough info yet to remove them | 09:47 |
| fijal | arigato: I guess we allocate the frame and then inline short preamble | 09:48 |
| fijal | which means the frame will be virtual | 09:48 |
| fijal | normal things occur in case of the bridge to the long preamble | 09:48 |
| arigato | fijal: I mean of course that the short preamble is inlined in the caller in what you said, so if a guard fails there, it will trigger recompilation in the caller, which is potentially bad | 09:48 |
| fijal | recompilation or a compilation of bridge? | 09:49 |
| Action: arigato doesn't get the question | 09:49 | |
| fijal | because it'll (ideally) start at the guard and compile just a bridge to the long preamble of the callee | 09:49 |
| fijal | no? | 09:50 |
| arigato | ok, I guess we're talking about a different solution | 09:50 |
| fijal | ok | 09:50 |
| fijal | sorry for the typical fijal-style change of topics, but is list(generator) supposed to be fast? | 09:50 |
| arigato | what did you mean above: " (short preamble inlining) for call_assembler" ? inlining the short preamble where? | 09:50 |
| fijal | so you trace until the point of the moment where you allocate frame and do call assembler | 09:51 |
| antocuni | hi | 09:51 |
| fijal | then instead, you allocate frame and inline the short preamble of a callee | 09:51 |
| fijal | and then call_assembler to the actual loop | 09:52 |
| arigato | yes, so what I'm saying is that you're adding potentially more guards to the *caller*, which may fail | 09:52 |
| fijal | yes, but they'll only start tracing for a bit | 09:53 |
| arigato | why? | 09:53 |
| fijal | and compile a bit that goes to the long preamble | 09:53 |
| arigato | that's what I don't get | 09:53 |
| arigato | if they fail, you compile a bridge that will also contain call_assembler and the whole rest of the caller | 09:53 |
| fijal | right | 09:54 |
| fijal | ok | 09:54 |
| fijal | arigato: yes, that's potentially bad | 09:58 |
| mikefc_ (~mikefc@60-242-240-196.static.tpgi.com.au) joined #pypy. | 10:06 | |
| dmarkey (~dmarkey@dmarkey.xen.prgmr.com) joined #pypy. | 10:06 | |
| nanonyme_ (nanonyme@unaffiliated/nanonyme) joined #pypy. | 10:07 | |
| mikefc (~mikefc@60-242-240-196.static.tpgi.com.au) got netsplit. | 10:07 | |
| PiotrSikora (~none@nginx/adept/piotrsikora) got netsplit. | 10:07 | |
| dmarkey_ (~dmarkey@dmarkey.xen.prgmr.com) got netsplit. | 10:07 | |
| _root_ (~Scalable@li252-14.members.linode.com) got netsplit. | 10:07 | |
| nanonyme (nanonyme@unaffiliated/nanonyme) got netsplit. | 10:07 | |
| __class__ (~class@99-105-56-217.lightspeed.sntcca.sbcglobal.net) got netsplit. | 10:07 | |
| jml (~jml@canonical/launchpad/jml) got netsplit. | 10:07 | |
| magcius (magcius@unaffiliated/magcius) got netsplit. | 10:07 | |
| Nick change: mikefc_ -> mikefc | 10:07 | |
| Possible future nick collision: mikefc | 10:07 | |
| __root__ (~Scalable@li252-14.members.linode.com) joined #pypy. | 10:07 | |
| _class_ (~class@99-105-56-217.lightspeed.sntcca.sbcglobal.net) joined #pypy. | 10:08 | |
| magcius (magcius@unaffiliated/magcius) returned to #pypy. | 10:08 | |
| PiotrSikora (~none@2001:41d0:2:936d::1) joined #pypy. | 10:08 | |
| Nick change: PiotrSikora -> Guest38408 | 10:09 | |
| Nick change: __root__ -> _root_ | 10:10 | |
| Possible future nick collision: _root_ | 10:10 | |
| Nick change: _class_ -> __class__ | 10:12 | |
| Possible future nick collision: __class__ | 10:12 | |
| jml (~jml@canonical/launchpad/jml) got lost in the net-split. | 10:18 | |
| dmarkey_ (~dmarkey@dmarkey.xen.prgmr.com) got lost in the net-split. | 10:18 | |
| nanonyme (nanonyme@unaffiliated/nanonyme) got lost in the net-split. | 10:18 | |
| kenaan | 12arigo extradoc 11c36311962789 15/sprintinfo/leysin-winter-2012/planning.txt: Planning session for today | 10:21 |
| kenaan | 12arigo extradoc 11588db0e8396b 15/: merge heads | 10:22 |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 10:33 | |
| antocuni | fijal: do you know if it's possible to merge a pull request inside a branch? | 10:34 |
| antocuni | e.g., I would like to merge the jvm-improvements inside its own branch, so we can run the tests before merging it to default | 10:35 |
| Action: antocuni would like not to have to play with transplant :-) | 10:35 | |
| ciphergoth (~paul@host238.lshift.net) joined #pypy. | 10:37 | |
| ciphergoth | With the sandbox, function calls like os.write are deferred to the controlling process | 10:40 |
| ciphergoth | Is there an easy way for me to add to the list of deferred functions? | 10:40 |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy. | 10:51 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host | 10:51 | |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 10:51 | |
| jml (~jml@mumak.net) joined #pypy. | 10:52 | |
| jml (~jml@mumak.net) left irc: Changing host | 10:52 | |
| jml (~jml@canonical/launchpad/jml) joined #pypy. | 10:52 | |
| arigato | ciphergoth: to add what? | 10:53 |
| ciphergoth | arigato: currently, if in my sandbox I try to call os.write(), that gets turned into a message for the controlling process. I'd like the same to be true of mymodule.foo() | 10:54 |
| arigato | where mymodule is a pure Python module, ideally? | 10:54 |
| matthp (c1a9466d@gateway/web/freenode/ip.193.169.70.109) joined #pypy. | 10:55 | |
| arigato | then no, not easily; you need to write some hack like def foo(string): os.write(42, string + '\n') | 10:55 |
| arigato | (and I really mean 42 literally) | 10:56 |
| matthp | from last night's logs: "<justinpeel> hmm.. np.array([[1,2],[3,4]]).prod(1) = array([2.0, 0.0]) in numpypy. should be array([2, 12])" | 10:56 |
| arigato | to improved the situation we could add "__pypy__.sandboxexternal(*args)" | 10:56 |
| arigato | for this very purpose | 10:56 |
| matthp | bug! _reduce_ufunc_impl() in interp_numarray.py should accept an identity argument! | 10:57 |
| arigato | (use -42 instead of 42 if you are concerned about conflicts) | 10:58 |
| matthp | arigato: sorry, bad lag here, didn't mean to cut you off. | 10:58 |
| arigato | matthp: no problem, feel free to cut me off. we're used to have two conversations in parallel | 10:59 |
| matthp | fijal: ping | 10:59 |
| matthp (c1a9466d@gateway/web/freenode/ip.193.169.70.109) left irc: Quit: Page closed | 11:00 | |
| ciphergoth | arigato: I note that you're respecting the types of arguments to os.write | 11:07 |
| ciphergoth | sadly it only returns an integer, which is rather limiting | 11:07 |
| arigato | yes | 11:07 |
| arigato | ciphergoth: you can use something else, like os.read() which returns a string | 11:08 |
| ciphergoth | but then only takes an integer :-) | 11:08 |
| arigato | yes :-) | 11:08 |
| arigato | it's all hacks | 11:08 |
| arigato | e.g. os.listdir() would take a string and return a list of strings | 11:08 |
| ciphergoth | I thought about trying to work by simulating a socket connection, but that gets pretty complex | 11:08 |
| arigato | basically __pypy__.sandboxexternal() could be made to accept any "simple-typed" arguments, and return another "simple-typed" argument | 11:10 |
| ciphergoth | I want to trace the sandboxed process outside of it | 11:10 |
| ciphergoth | one hit for "pypy sandboxexternal" | 11:11 |
| arigato | well sandboxexternal() is a name I just came up with | 11:11 |
| ciphergoth | ah OK | 11:11 |
| arigato | it's not implemented so far | 11:11 |
| arigato | the easiest is actually to have it take one string and return one string | 11:12 |
| ciphergoth | OK | 11:12 |
| arigato | then you can do any pickling on top of it | 11:12 |
| arigato | or marshal.dumps/loads | 11:13 |
| ciphergoth | right now it sounds like some sort of evil listdir hack might be the way forward | 11:13 |
| arigato | yes :-) | 11:13 |
| arigato | rather obscure but yes | 11:13 |
| ciphergoth | if there's another function that both takes and returns a string that I can override in a less hacky way that would be great | 11:14 |
| arigato | os.confstr() has the signature "str->str" | 11:14 |
| ciphergoth | aha! | 11:14 |
| arigato | ah, or just os.getenv(key)->str | 11:14 |
| ciphergoth | so how does pypy work out what functions to sandbox atm? I guess it's all the non-pure-Python ones | 11:14 |
| arigato | it's all the ones that include calls to library functions | 11:15 |
| arigato | and sandboxing actually occurs at the level of the calls to library functions | 11:15 |
| arigato | (as opposed to at the level of the Python wrapper os.listdir()) | 11:15 |
| ciphergoth | there's a Python wrapper? | 11:16 |
| mfoord | os.listdir *is* the python wrapper | 11:16 |
| ciphergoth | ah there isn't in CPython but maybe there is in pypy? | 11:17 |
| arigato | (ah, use confstr(), not getenv(), because there are other level of wrappers around the latter) | 11:17 |
| arigato | ciphergoth: no, there is too in CPython: os.listdir() is a small wrapper around the C libraries | 11:17 |
| arigato | (I also see os.readlink(str)->str) | 11:18 |
| ciphergoth | confused: searching /usr/lib/python2.7/os.py for "listdir" doesn't reveal a function of that name | 11:18 |
| arigato | ciphergoth: no, it's a wrapper implemented in C or RPython | 11:18 |
| arigato | it's still a wrapper because it mainly just calls the C library functions | 11:19 |
| arigato | (it's defined in the posix module, which is imported by "from posix import *" into os.py) | 11:19 |
| ciphergoth | ah, pypy/rpython/module/ll_os.py | 11:19 |
| ciphergoth | so it's where you call self.llexternal that we hit the wrapper | 11:20 |
| legogris (~legogris@c80-217-207-157.bredband.comhem.se) left irc: Ping timeout: 240 seconds | 11:20 | |
| arigato | in the sandbox case, yes, it's self.llexternal() that will ask the outer process what to do instead of doing it itself | 11:21 |
| arigato | but look at os.readlink() instead of listdir(), it's much simpler | 11:21 |
| legogris (~legogris@c80-217-207-157.bredband.comhem.se) joined #pypy. | 11:22 | |
| stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 11:23 | |
| stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy. | 11:24 | |
| Action: arigato tries to undo the mess in translator/stm in view of the new point of view | 11:27 | |
| ciphergoth | hmm the code for os_readlink_llimpl doesn't look right | 11:31 |
| ciphergoth | it looks like it assumes readlink will return 0 if bufsize is too small | 11:32 |
| ciphergoth | the manual page doesn't seem to say that | 11:32 |
| ciphergoth | sorry scratch that I'm talking rubbish | 11:33 |
| bivab (david@fwstups.cs.uni-duesseldorf.de) left #pypy. | 11:49 | |
| bivab_ (~david@134.99.36.142) joined #pypy. | 11:49 | |
| bivab_ (~david@134.99.36.142) left irc: Client Quit | 11:49 | |
| witulski (~stupsi@134.99.16.26) joined #pypy. | 11:59 | |
| witulski (stupsi@134.99.16.26) left #pypy. | 11:59 | |
| umgeher (~umgeher@unaffiliated/umgeher) left irc: Ping timeout: 255 seconds | 12:36 | |
| lizardo (~lizardo@189.2.128.130) joined #pypy. | 12:44 | |
| khs (~khs@2001:700:300:2120:725a:b6ff:fee5:a44) joined #pypy. | 12:49 | |
| kenaan | 12rguillebert py3k 115500c9127aab 15/pypy/interpreter/pyparser/test/test_pyparse.py: Add a failing test for extended attribute unpacking | 13:05 |
| aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds | 13:05 | |
| azanella (~azanella@189.6.80.131) joined #pypy. | 13:09 | |
| azanella (~azanella@189.6.80.131) left irc: Ping timeout: 260 seconds | 13:14 | |
| kenaan | 12rguillebert py3k 1134ff01570492 15/pypy/interpreter/pyparser/test/test_pyparse.py: Rename the test commited earlier, rewrite tests that should pass. | 13:18 |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 13:19 | |
| witulski (~stupsi@134.99.16.26) joined #pypy. | 13:21 | |
| Eventh (evenwiik@caracal.stud.ntnu.no) joined #pypy. | 13:24 | |
| kenaan | 12rguillebert py3k 11851536b685a6 15/pypy/interpreter/pyparser/test/test_pyparse.py: Parsing 0777 should fail on py3k | 13:25 |
| anish_ (anish@nat/redhat/x-fmpltnxpvuwwxlhc) left irc: Ping timeout: 252 seconds | 13:28 | |
| apanda (~apanda@59.178.210.229) joined #pypy. | 13:31 | |
| apanda (~apanda@59.178.210.229) left irc: Remote host closed the connection | 13:32 | |
| DasIch (~DasIch@p3E991405.dip.t-dialin.net) joined #pypy. | 13:35 | |
| kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy. | 13:35 | |
| bivab (~david@ip-62-143-59-5.unitymediagroup.de) joined #pypy. | 13:35 | |
| anish_ (anish@nat/redhat/x-utnpuiafquccehkh) joined #pypy. | 13:39 | |
| witulski (~stupsi@134.99.16.26) left irc: Ping timeout: 248 seconds | 13:49 | |
| kenaan | 12arigo stm 1191a0ee8fc4ab 15/pypy/translator/stm/test/: Add a (skipped) test about using the minimark GC. | 13:59 |
| kenaan | 12arigo stm 11382a5969beda 15/pypy/: Revert 0782958b144f. No longer needed. | 13:59 |
| kenaan | 12arigo stm 112965c13c2427 15/pypy/: (antocuni, arigo) Start to refactor the world. The idea is not to build on RPython threads any more. | 13:59 |
| anish_ (anish@nat/redhat/x-utnpuiafquccehkh) left irc: Ping timeout: 252 seconds | 14:00 | |
| aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy. | 14:01 | |
| whyking (~quassel@p4FFB7B5C.dip.t-dialin.net) joined #pypy. | 14:09 | |
| anish_ (anish@nat/redhat/x-koeusarsjwzkxuao) joined #pypy. | 14:13 | |
| nettok (~quassel@190.149.184.238) joined #pypy. | 14:18 | |
| nettok (~quassel@190.149.184.238) left irc: Remote host closed the connection | 14:21 | |
| anish_ (anish@nat/redhat/x-koeusarsjwzkxuao) left irc: Ping timeout: 240 seconds | 14:25 | |
| gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy. | 14:30 | |
| horieyui (horieyui@222.47.182.195) left irc: Quit: http://www.cnblogs.com/crazyhack | 14:32 | |
| Eventh (evenwiik@caracal.stud.ntnu.no) left irc: Quit: Lost terminal | 14:33 | |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection | 14:33 | |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 14:35 | |
| AlekSi (~AlekSi@85.235.191.82) joined #pypy. | 14:36 | |
| anish_ (anish@nat/redhat/x-jncolakpedvvseof) joined #pypy. | 14:38 | |
| kenaan | 12arigo stm 117dc69a93d5be 15/pypy/translator/stm/: (antocuni, arigo) Fix the llstminterp. Wondering a bit what is its purpose... | 14:43 |
| Rhy0lite (dje@nat/ibm/x-nqxtecyvgqkpgfrd) joined #pypy. | 14:49 | |
| mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord | 14:55 | |
| whitelynx|work (~whitelynx@63.241.75.144) joined #pypy. | 15:02 | |
| alexwillmer (~alexwillm@5ad0d588.bb.sky.com) joined #pypy. | 15:05 | |
| alexwillmer | Afternoon all, Is there such a thing as 2to3 but for a PyPy? A script that when run over a .py file warns/suggests replacements of CPython-isms that will be problematic on PyPy | 15:07 |
| alexwillmer | One example I'm thinking of: x is 0 vs x == 0 | 15:09 |
| gutworth | you should never write x == 0 in cpython either | 15:12 |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection | 15:12 | |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 15:13 | |
| alexwillmer | gutworth: I know that, but somebody didn't http://trac.edgewall.org/changeset/c8dceba41b5b23ec6dce1debfb8b828595790cda/jomae | 15:14 |
| JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 15:15 | |
| gutworth | essentially all cpython-isms are bad python coding | 15:15 |
| alexwillmer | (That's the link that corrects it. It wasn't me that spotted it) | 15:15 |
| alexwillmer | Wonder if pylint would report x is 0 | 15:15 |
| AlekSi (~AlekSi@85.235.191.82) left irc: Remote host closed the connection | 15:16 | |
| ciphergoth | gutworth: you know this, but you meant " you should never write x is 0 in cpython either" | 15:17 |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 260 seconds | 15:18 | |
| gutworth | indeed :) | 15:20 |
| mher (~Adium@tethras5.ucd.ie) left irc: Quit: Leaving. | 15:20 | |
| JaRoel (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection | 15:22 | |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy. | 15:23 | |
| witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy. | 15:25 | |
| witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy. | 15:27 | |
| mher (~Adium@tethras5.ucd.ie) joined #pypy. | 15:28 | |
| antocuni | we got tired of code reuse, so we started to reuse also commit messages :-) | 15:30 |
| antocuni | <spam> | 15:30 |
| kenaan | 12antocuni stm 11a81e51b21e06 15/pypy/translator/stm/: (arigo, antocuni): fix test_getfield_all_sizes, and make sure we test both the cases of stm_getfield inside and outs... | 15:30 |
| kenaan | 12antocuni stm 116b3914441db1 15/pypy/translator/stm/test/test_funcgen.py: (arigo, antocuni): fix test_setfield_all_sizes, and make sure we test both the cases of stm_setfield inside and outs... | 15:31 |
| kenaan | 12antocuni stm 1108f6815aba37 15/pypy/translator/stm/test/test_funcgen.py: (arigo, antocuni): fix test_getarrayitem_all_sizes, and make sure we test both the cases of stm_getarrayitem inside ... | 15:31 |
| kenaan | 12antocuni stm 11da7457ea38cd 15/pypy/translator/stm/test/test_funcgen.py: (arigo, antocuni): fix test_setarrayitem_all_sizes, and make sure we test both the cases of stm_setarrayitem inside ... | 15:31 |
| kenaan | 12antocuni stm 11e7847828c029 15/pypy/translator/stm/test/test_funcgen.py: (arigo, antocuni): fix test_getinteriorfield_all_sizes, and make sure we test both the cases of stm_getinteriorfield... | 15:31 |
| kenaan | 12antocuni stm 11a2b3f2c9ea45 15/pypy/translator/stm/test/test_funcgen.py: (arigo, antocuni): fix test_setinteriorfield_all_sizes, and make sure we test both the cases of stm_setinteriorfield... | 15:31 |
| antocuni | </spam> | 15:31 |
| gutworth | hehe, commit message parameterization? | 15:31 |
| timotimo | makes the hg repository compress better | 15:31 |
| gutworth | I think the classic pypy one is "test and fix" | 15:31 |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy. | 15:34 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host | 15:34 | |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 15:34 | |
| ltratt (~ltratt@198.45.112.87.dyn.plus.net) joined #pypy. | 15:34 | |
| stakkars_ (~tismer@82.113.98.207) joined #pypy. | 15:43 | |
| arigato | http://paste.pocoo.org/show/535680/ | 15:46 |
| rguillebert | is it rpython to call str with a W_Unicode object as the parameter ? | 15:51 |
| stakkars_ (~tismer@82.113.98.207) left irc: Quit: schnarch | 15:51 | |
| timotimo | rguillebert: if the W_Unicode object has a __str__ method i don't see why it shouldn't? | 15:52 |
| gutworth | no, it's not | 15:52 |
| stakkars_ (~tismer@82.113.98.207) joined #pypy. | 15:53 | |
| timotimo | oh, how come? | 15:53 |
| rguillebert | I should call space.something ? | 15:53 |
| gutworth | space.str | 15:53 |
| gutworth | what are you trying to do? | 15:53 |
| gutworth | timotimo: almost no special methods are rpython | 15:54 |
| timotimo | OK, my mistake then, carry on :) | 15:54 |
| rguillebert | I have a wrapped unicode string that I should unwrap and turn to unicode (in no particular order) | 15:54 |
| rguillebert | hum | 15:54 |
| rguillebert | turn to str | 15:54 |
| gutworth | well, you obviously cannot always do that | 15:55 |
| rguillebert | but I'm not sure that I should get a unicode string in the first place | 15:55 |
| rguillebert | so I'm curing the symptoms more than the disease | 15:55 |
| alexwillmer (alexwillm@5ad0d588.bb.sky.com) left #pypy. | 15:55 | |
| mher1 (~Adium@tethras5.ucd.ie) joined #pypy. | 15:55 | |
| mher (~Adium@tethras5.ucd.ie) left irc: Read error: Connection reset by peer | 15:55 | |
| rguillebert | (I'm working on py3k so there's stuff turned to unicode) | 15:56 |
| whyking (~quassel@p4FFB7B5C.dip.t-dialin.net) left irc: Ping timeout: 245 seconds | 16:00 | |
| squiddy (~squiddy@g224197249.adsl.alicedsl.de) left irc: Read error: Connection reset by peer | 16:02 | |
| anish_ (anish@nat/redhat/x-jncolakpedvvseof) left irc: Quit: Leaving | 16:02 | |
| jnoller (~jnoller@vpn.nasuni.com) joined #pypy. | 16:05 | |
| fijal | antocuni: are you still interested? | 16:07 |
| antocuni | fijal: in what? | 16:07 |
| fijal | antocuni: in how to deal with merging into branch | 16:09 |
| antocuni | I asked Michal to do it, but if you know an easy way it's always useful to know :-) | 16:10 |
| fijal | you just close the pull request and pull into a branch? | 16:14 |
| Alex_Gaynor | fijal: last night I was thinking about the same thing (inlining entry into call_assembler) | 16:14 |
| Alex_Gaynor | fijal: I'm not sure it lets you skip allocating frame, but it lets you avoid a bunch of unboxing | 16:14 |
| fijal | arigato: ping? | 16:16 |
| Alex_Gaynor | fijal, antocuni: Any opinion about letting JIT look into _codecs module? | 16:16 |
| Alex_Gaynor | ATM stuff like str.decode gets lost in that mess before you even run the encoder | 16:17 |
| arigato | fijal: pong? | 16:17 |
| fijal | arigato: can't we merge all traces after call_assembler anyway? | 16:17 |
| arigato | maybe | 16:17 |
| fijal | because call_assembler invalidates all possible caches anyway | 16:17 |
| fijal | Alex_Gaynor: try? | 16:18 |
| fijal | arigato: I agree it's twisting again my arguments, but this is what I (quite subconsciously) was thinking about | 16:20 |
| Alex_Gaynor | fijal: am :) | 16:20 |
| fijal | arigato: well, I won't work on it very soon | 16:24 |
| stakkars_ (~tismer@82.113.98.207) left irc: Quit: schnarch | 16:25 | |
| bivab_ (~david@ip-62-143-59-5.unitymediagroup.de) joined #pypy. | 16:25 | |
| bivab (~david@ip-62-143-59-5.unitymediagroup.de) left irc: Read error: Connection reset by peer | 16:25 | |
| Nick change: bivab_ -> bivab | 16:25 | |
| fijal | Alex_Gaynor: I'm looking at a jinja benchmark that's 3x slower than cpython | 16:26 |
| fijal | it's even very easy | 16:26 |
| Alex_Gaynor | fijal: oh yeah, that's the problem? | 16:26 |
| lesshaste | hi | 16:29 |
| lesshaste | I was just looking at the very poor http://attractivechaos.github.com/plb/ | 16:29 |
| whyking (~quassel@p4FFB7B5C.dip.t-dialin.net) joined #pypy. | 16:29 | |
| lesshaste | so I thought I would test some of them using pypy 1.7 | 16:29 |
| lesshaste | (don't ask why they used a 12 month old pypy) | 16:29 |
| lesshaste | their tests seem mostly broken | 16:31 |
| lesshaste | ./matmul_v1.py: line 7: syntax error near unexpected token `(' | 16:31 |
| lesshaste | ./matmul_v1.py: line 7: `def matmul(a, b): # FIXME: no error checking' | 16:31 |
| Alex_Gaynor | sounds like it's just missing a shebang line | 16:31 |
| rguillebert | so matrix multiplication doesn't use numpy ? | 16:31 |
| rguillebert | which can be counted as 'in the standard library' in the case of pypy :) | 16:32 |
| lesshaste | Alex_Gaynor: doh! | 16:33 |
| lesshaste | Alex_Gaynor: in any case.. I can't get any of the tests to have results anything like they quote | 16:33 |
| lesshaste | Alex_Gaynor: except for sudoko.. where it is clear that pypy just takes some time to warm up | 16:34 |
| Alex_Gaynor | as in the ratio between different impls or absolute? | 16:34 |
| Nick change: forrestoshi -> forrestv | 16:34 | |
| lesshaste | ratio | 16:34 |
| lesshaste | well also absolute as my pc must be much slower than theirs | 16:34 |
| lesshaste | and yet I get faster times | 16:34 |
| lesshaste | even for cpython | 16:34 |
| lesshaste | time ./matmul_v2.py | 16:35 |
| lesshaste | -9.3358333 | 16:35 |
| asmeurer_ (~asmeurer@129.138.218.98) joined #pypy. | 16:35 | |
| lesshaste | assuming that is what they really did | 16:35 |
| lesshaste | Alex_Gaynor: just tried dict.. it takes basically no time | 16:38 |
| lesshaste | very weird | 16:38 |
| bfirsh (u1308@gateway/web/irccloud.com/x-bimfhxsmopzfjqbz) left irc: Remote host closed the connection | 16:38 | |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-cfnpvjrrtplwnnkc) left irc: Remote host closed the connection | 16:38 | |
| oal (u4126@gateway/web/irccloud.com/x-jadakdcnvlfonink) left irc: Write error: Connection reset by peer | 16:38 | |
| lesshaste | oh I see what they did for dict.. involves running another app to get the input | 16:39 |
| lesshaste | oh.. well actually this is a nice example it seems :) dict_v1.py from https://github.com/attractivechaos/plb/tree/master/dict is very slow in pypy 1.7 | 16:42 |
| lesshaste | 7 seconds in cpython | 16:42 |
| lesshaste | 18 seconds in pypy | 16:42 |
| lesshaste | tested using file created by https://github.com/attractivechaos/plb/blob/master/ext/genint.c | 16:43 |
| bfirsh (u1308@gateway/web/irccloud.com/x-ghxvuahtgluskuol) joined #pypy. | 16:43 | |
| lesshaste | hi bfirsh | 16:45 |
| oal (u4126@gateway/web/irccloud.com/x-unpyohqhdbnlojoe) joined #pypy. | 16:45 | |
| fijal | lesshaste: a combination of slow read and a large dicts that are slow in pypy | 16:45 |
| lesshaste | fijal: ok | 16:46 |
| fijal | it's still interesting | 16:46 |
| lesshaste | and sudoko is all about the warm up time, right? | 16:47 |
| lesshaste | it appears to be much faster after a second or two | 16:47 |
| lesshaste | in general the tests seems really stupid as they take a second or two to complete | 16:50 |
| lesshaste | unless I am running them wrong | 16:50 |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-veirrgtasauxjhkw) joined #pypy. | 16:51 | |
| #pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services. | 16:51 | |
| kenaan | 12arigo default 11e244681d320e 15/pypy/rpython/: (antocuni, arigo) Add support for malloc'ing RPython instances non-movable, just by specifying "_alloc_nonmovable_... | 16:51 |
| kenaan | 12arigo stm 1130f56b4fb5ea 15/: hg merge default | 16:51 |
| kenaan | 12antocuni stm 11c3ebb4f01089 15/pypy/translator/stm/test/test_transform.py: (arigo, antocuni) fix test which failed after the renaming of stm_become_inevitable | 16:53 |
| kenaan | 12arigo stm 11a14e0f148aad 15/pypy/translator/stm/test/test_transform.py: hg merge | 16:53 |
| `fox` (~fox@host19-111-dynamic.20-79-r.retail.telecomitalia.it) joined #pypy. | 17:01 | |
| antocuni | <arigato>: "It looks safe enough" --> Segfault | 17:01 |
| fijal | :] | 17:01 |
| Action: fijal wants planning! | 17:02 | |
| lesshaste | fijal: hmm.. matmul_v1 and v2 take about 7 seconds on pypy and 1 second in shedskin .. from https://github.com/attractivechaos/plb/tree/master/matmul (with n = 500 in the source) | 17:02 |
| arigato | fijal: planning session? | 17:03 |
| arigato | or what do you mean? | 17:03 |
| Alex_Gaynor | arigato: planning.txt, people who aren't at the sprint like reading it :) | 17:04 |
| fijal | arigato: I wand planning file :) | 17:04 |
| fijal | lesshaste: make sure those arrays are both float then it'll go faster | 17:04 |
| fijal | right now it's a combination of int and float | 17:04 |
| arigato | fijal: we checked in one | 17:04 |
| fijal | uh missed that | 17:04 |
| fijal | lesshaste: 7s or 0.7s? | 17:06 |
| lesshaste | fijal: pypy seems 7 times slower | 17:07 |
| lesshaste | fijal: 7s for pypy, 1s for shedskin | 17:08 |
| fijal | lesshaste: how old is your pypy? | 17:08 |
| lesshaste | 1.7 | 17:08 |
| fijal | ah | 17:08 |
| fijal | we merged new cool stuff | 17:08 |
| fijal | it's much faster now | 17:08 |
| fijal | there is absolutely no point in running benchmarks with such an old version :) | 17:08 |
| lesshaste | http://buildbot.pypy.org/nightly/trunk/pypy-c-jit-latest-linux.tar.bz2 ? | 17:09 |
| fijal | yeah | 17:09 |
| lesshaste | fijal: 2 seconds :) | 17:10 |
| lesshaste | maybe the warm up time is the problem | 17:10 |
| lesshaste | I'll increase the matrix size | 17:11 |
| fijal | lesshaste: run it few times | 17:13 |
| fijal | lesshaste: seriously it's 0.7s on my laptop | 17:14 |
| fijal | it's not 2x faster | 17:14 |
| lesshaste | try matmul_v1.py with n = 800 | 17:14 |
| lesshaste | time ~/Desktop/pypy-c-jit-51333-eb0269c21eec-linux/bin/pypy ./matmul_v1.py | 17:14 |
| fijal | 2s | 17:14 |
| lesshaste | real0m6.620s | 17:14 |
| lesshaste | now try shedskin | 17:14 |
| fijal | what's your os? | 17:14 |
| lesshaste | time ./matmul_v1 | 17:15 |
| lesshaste | real0m3.186s | 17:15 |
| fijal | I don't have it and I'm leaving for dinner | 17:15 |
| lesshaste | ubuntu.. but my computer is slow | 17:15 |
| lesshaste | an oldish amd computer | 17:15 |
| kenaan | 12antocuni stm 111aa99f2f035f 15/pypy/: (arigo, antocuni): start to write the RPython level interface for using transactions in rlib/rstm.py. Move the comp... | 17:15 |
| kenaan | 12antocuni stm 11f19ffa0b7bb3 15/pypy/: (arigo, antocuni): rename _rffi_stm.* to _rffi_stm.stm_* and rstm.stm_* to rstm.* | 17:15 |
| lesshaste | there is a big speed difference between v1 and v2 | 17:15 |
| lesshaste | I'll retry dict with the latest pypy while you dine :) | 17:15 |
| kenaan | 12alex_gaynor default 11af6b237eaecf 15/pypy/module/pypyjit/policy.py: allow inlining into _codecs, makes simple decoding ~3x faster | 17:16 |
| kenaan | 12alex_gaynor default 118688ce42472e 15/pypy/rpython/: merged upstream | 17:16 |
| lesshaste | ime cat test.dict |~/Desktop/pypy-c-jit-51333-eb0269c21eec-linux/bin/pypy ./dict_v1.py | 17:16 |
| lesshaste | real0m10.590s | 17:16 |
| lesshaste | time cat test.dict |./dict_v1.py | 17:17 |
| lesshaste | real0m7.053s | 17:17 |
| lesshaste | so in the first case pypy is slower than shedskin and in the second it is slower than cpython | 17:18 |
| kenaan | 12arigo stm 11e2645247a9c2 15/pypy/: (antocuni, arigo) Move CompiledSTMTests to its own file in translator/stm/test/support.py. | 17:19 |
| lesshaste | but there is something fishy about the matmul example as v1 and v2 have very different performances | 17:19 |
| lesshaste | fijal: bon appetit | 17:20 |
| G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving. | 17:27 | |
| kenaan | 12arigo stm 11dbadc138e9b5 15/pypy/: (antocuni, arigo) Add an integration test to test_rstm.py: really call rstm.perform_transaction() in a C-compiled test. | 17:29 |
| mher1 (~Adium@tethras5.ucd.ie) left irc: Quit: Leaving. | 17:30 | |
| mattip (4fb1711d@gateway/web/freenode/ip.79.177.113.29) joined #pypy. | 17:35 | |
| antocuni | good, today I and armin added 642 lines of code, but removed 559 | 17:35 |
| antocuni | not too bad :-) | 17:35 |
| Action: mattip looking at axisops bug. | 17:36 | |
| oal (u4126@gateway/web/irccloud.com/x-unpyohqhdbnlojoe) left irc: Remote host closed the connection | 17:37 | |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-veirrgtasauxjhkw) left irc: Read error: Connection reset by peer | 17:37 | |
| bfirsh (u1308@gateway/web/irccloud.com/x-ghxvuahtgluskuol) left irc: Remote host closed the connection | 17:37 | |
| mattip | bug with a.prod(1) is not an issue with identity. | 17:37 |
| woyna (woyna@79.133.200.48) joined #pypy. | 17:38 | |
| bfirsh (u1308@gateway/web/irccloud.com/x-zthszjvtbgvwqwvt) joined #pypy. | 17:39 | |
| `fox` (~fox@host19-111-dynamic.20-79-r.retail.telecomitalia.it) left irc: Read error: Operation timed out | 17:40 | |
| oal (u4126@gateway/web/irccloud.com/x-vtujjnjpajthapsi) joined #pypy. | 17:43 | |
| ltratt (~ltratt@198.45.112.87.dyn.plus.net) left irc: Quit: using sirc version 2.211+KSIRC/1.3.12 | 17:43 | |
| amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy. | 17:44 | |
| bivab (~david@ip-62-143-59-5.unitymediagroup.de) left irc: Quit: bivab | 17:45 | |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-ywjjhaqqlviyrqlw) joined #pypy. | 17:45 | |
| #pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services. | 17:45 | |
| khs (~khs@2001:700:300:2120:725a:b6ff:fee5:a44) left irc: Quit: Leaving | 17:46 | |
| mattip | fijal: Your use of first_line is even more clever than my first analysis led me to believe. very cool. | 17:53 |
| kenaan | 12arigo stm 111e52821ea7bc 15/pypy/rlib/test/test_rstm.py: (antocuni, arigo) Test (maybe) for inevitable transactions. | 17:56 |
| kenaan | 12arigo stm 11f565b7971e14 15/pypy/: (antocuni, arigo) A way to get and test the current transaction mode, for debugging. | 17:56 |
| jnoller (~jnoller@vpn.nasuni.com) left irc: Quit: Leaving... | 17:58 | |
| squiddy (~squiddy@g224197249.adsl.alicedsl.de) joined #pypy. | 17:59 | |
| oal (u4126@gateway/web/irccloud.com/x-vtujjnjpajthapsi) left irc: Remote host closed the connection | 17:59 | |
| rguillebert | hi amaury_ | 18:00 |
| oal (u4126@gateway/web/irccloud.com/x-itkkrhjtcqjwpxbe) joined #pypy. | 18:03 | |
| DasIch_ (~DasIch@p3E991405.dip.t-dialin.net) joined #pypy. | 18:06 | |
| dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) joined #pypy. | 18:06 | |
| DasIch (~DasIch@p3E991405.dip.t-dialin.net) left irc: Ping timeout: 252 seconds | 18:09 | |
| Nick change: DasIch_ -> DasIch | 18:09 | |
| mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord | 18:13 | |
| danchr (~danchr@cl-848.chi-02.us.sixxs.net) left irc: Ping timeout: 248 seconds | 18:18 | |
| danchr (~danchr@cl-848.chi-02.us.sixxs.net) joined #pypy. | 18:19 | |
| antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) left irc: Ping timeout: 252 seconds | 18:22 | |
| arigato (~arigo@89-40.5-85.cust.bluewin.ch) left irc: Ping timeout: 252 seconds | 18:22 | |
| davisagli (~davisagli@davisagli.com) left irc: Excess Flood | 18:25 | |
| davisagli (~davisagli@davisagli.com) joined #pypy. | 18:25 | |
| exnihilo (~nathan@descartes.its.unc.edu) joined #pypy. | 18:26 | |
| nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds | 18:28 | |
| mattip (4fb1711d@gateway/web/freenode/ip.79.177.113.29) left irc: Quit: Page closed | 18:31 | |
| oal (u4126@gateway/web/irccloud.com/x-itkkrhjtcqjwpxbe) left irc: Remote host closed the connection | 18:34 | |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-ywjjhaqqlviyrqlw) left irc: Remote host closed the connection | 18:34 | |
| bfirsh (u1308@gateway/web/irccloud.com/x-zthszjvtbgvwqwvt) left irc: Remote host closed the connection | 18:34 | |
| amaury_ | rguillebert: hi | 18:37 |
| amaury_ | how is the sprint? | 18:37 |
| bfirsh (u1308@gateway/web/irccloud.com/x-zeyuomfbcehjdgda) joined #pypy. | 18:38 | |
| bfirsh (u1308@gateway/web/irccloud.com/x-zeyuomfbcehjdgda) left irc: Remote host closed the connection | 18:39 | |
| bfirsh (u1308@gateway/web/irccloud.com/x-xcpfqjeimrkfzbxe) joined #pypy. | 18:42 | |
| Alex_Gaynor (u1246@gateway/web/irccloud.com/x-dlgnnjmsxhmyelkc) joined #pypy. | 18:46 | |
| #pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services. | 18:46 | |
| whyking (~quassel@p4FFB7B5C.dip.t-dialin.net) left irc: Ping timeout: 244 seconds | 18:48 | |
| `fox` (~fox@host19-111-dynamic.20-79-r.retail.telecomitalia.it) joined #pypy. | 18:54 | |
| int3__ (~int3__@bb116-14-110-92.singnet.com.sg) joined #pypy. | 19:00 | |
| mattip (4fb1711d@gateway/web/freenode/ip.79.177.113.29) joined #pypy. | 19:00 | |
| mwhudson (~mwh@121.73.77.183) joined #pypy. | 19:01 | |
| mwhudson (~mwh@121.73.77.183) left irc: Changing host | 19:01 | |
| mwhudson (~mwh@linaro/mwhudson) joined #pypy. | 19:01 | |
| hakanardo | fijal: pong | 19:01 |
| icrazyhack (~horieyui@183.60.100.166) joined #pypy. | 19:03 | |
| ojii (~ojii@50-57-189-190.static.cloud-ips.com) left irc: Ping timeout: 260 seconds | 19:03 | |
| `fox` (~fox@host19-111-dynamic.20-79-r.retail.telecomitalia.it) left irc: Quit: Lost terminal | 19:05 | |
| ojii (~ojii@50.57.189.190) joined #pypy. | 19:06 | |
| nickik (~nick@217.150.234.222) joined #pypy. | 19:07 | |
| oal (u4126@gateway/web/irccloud.com/x-lhdzgqgbidfplosv) joined #pypy. | 19:10 | |
| kenaan | 12mattip default 11ee1b4ea67ed9 15/pypy/module/micronumpy/: add failing test, add fix | 19:11 |
| mattip | fijal: can you take a look at that ^^^ just to make sure? | 19:11 |
| mattip | we needed a test where: | 19:12 |
| mattip | - identity is neither None nor 0 and | 19:13 |
| mattip | - we iterate across something other than the 0 axis | 19:13 |
| mattip | justinpeel: thanks for pointing out the bug. | 19:14 |
| timotimo | hm, i haven't really kept track; how far is multi-dimensional numpy arrays? | 19:15 |
| mattip | slices and reduce functions work. Hopefully we will have matrix multiply in a week or so. | 19:18 |
| timotimo | so, obviously, creating multidim arrays works? nice. | 19:19 |
| mattip | well, no hstack, concatenate, etc but yes reshape | 19:20 |
| mattip | and you can do array([[1,2], [3,4]]) | 19:20 |
| mattip | what's your pet feature? | 19:21 |
| timotimo | uh | 19:22 |
| timotimo | not much beyond simple slicing really | 19:22 |
| mattip | I sometimes wonder how long the tail is for numpy, | 19:23 |
| timotimo | oh, i do have one occurence of the histogram function i believe | 19:23 |
| mattip | There's so many features but I commonly use very few. | 19:23 |
| fijal | hakanardo: hey | 19:23 |
| hakanardo | fijal: hi :) | 19:23 |
| fijal | hakanardo: so I had a weird idea for call_assembler | 19:24 |
| fijal | to just inline the short preamble and maybe avoid allocating the frame | 19:25 |
| fijal | but we need trace merging before we can do that | 19:25 |
| hakanardo | cool | 19:26 |
| hakanardo | trace merging should sort of work by now... | 19:26 |
| hakanardo | if the virtual state at the end of a bridge matches the virtual state of a loop | 19:27 |
| fijal | ok | 19:27 |
| hakanardo | you can jump directly to the loop by inlining the short preamble | 19:27 |
| fijal | hakanardo: so, can we merge traces at each call_assembler? | 19:27 |
| fijal | since call_assembler is always invalidating all possible caches | 19:27 |
| fijal | it's not a jump, it's a call btw | 19:27 |
| hakanardo | hmm | 19:28 |
| fijal | call_assembler | 19:28 |
| moreati (~alex@5ad0d588.bb.sky.com) joined #pypy. | 19:28 | |
| hakanardo | so you would like to place a label infront of ever call_assembler | 19:28 |
| fijal | no | 19:28 |
| fijal | I want to replace call_assembler with another call_assembler | 19:28 |
| fijal | essentially | 19:28 |
| fijal | and possibly merge traces after call_assembler | 19:28 |
| hakanardo | we need a label at the point where you merge traces | 19:29 |
| fijal | yes, that's fine | 19:29 |
| timotimo | mattip: is there C vs F memory layout? | 19:29 |
| fijal | timotimo: no | 19:29 |
| timotimo | also, does pypy 1.7 have multidim ndarray? | 19:29 |
| hakanardo | so it's right after the call_assembler then that we want the label I suppose | 19:29 |
| fijal | timotimo: don't use 1.7 | 19:29 |
| fijal | :) | 19:29 |
| fijal | hakanardo: so how does it work? what sort of structures have to match? | 19:29 |
| JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection | 19:29 | |
| timotimo | oh, is 1.7 especially broken in some way? | 19:29 |
| fijal | timotimo: no, just nightly is much better, especially when it comes to numpy | 19:30 |
| timotimo | of course | 19:30 |
| timotimo | the question is when i'll be able to claim "stable pypy can run my stuff" | 19:30 |
| hakanardo | fijal: the virtualstate has to match | 19:30 |
| hakanardo | check out jump_to_already_compiled_trace in unroll.py | 19:30 |
| fijal | hakanardo: ok | 19:31 |
| timotimo | i personally am quite fond of the nightlies | 19:31 |
| fijal | timotimo: you mean numpy? | 19:31 |
| hakanardo | it loops over all specialized version of a loop and tries to find on that maches | 19:31 |
| hakanardo | and if it does i procudes the jump there | 19:31 |
| fijal | ok | 19:31 |
| fijal | I guess right now what I have is: | 19:31 |
| fijal | allocate the frame, do call_assembler that goes to preamble | 19:31 |
| hakanardo | target.virtual_state.generalization_of is the check | 19:31 |
| fijal | what I want is: | 19:31 |
| fijal | avoid allocating the frame and call directly the loop | 19:32 |
| fijal | and then merge traces after call_assembler | 19:32 |
| Alex_Gaynor | how can you avoid allocaint the frame? | 19:32 |
| Alex_Gaynor | the loop assumes it exists | 19:32 |
| timotimo | where can the code that is used to generate the listing at http://buildbot.pypy.org/nightly/ be found? | 19:32 |
| timotimo | i'd like to contribute "sorting by date of last change" | 19:33 |
| fijal | Alex_Gaynor: maybe we can compile the loop slightly differently? | 19:33 |
| timotimo | hm, or maybe alphabetical sorting is OK, but it should still display some kind of "date of last change" or something | 19:33 |
| fijal | timotimo: it's a normal directory view of twisted.web | 19:33 |
| Alex_Gaynor | fijal: could be | 19:33 |
| fijal | Alex_Gaynor: in general we *know* the frame does not escape | 19:34 |
| fijal | if all we do is create it and then do call_assembler | 19:34 |
| Alex_Gaynor | fijal: even just inlining entry and calling to preamble would help since you avoid unboxing | 19:34 |
| int3__ (~int3__@bb116-14-110-92.singnet.com.sg) left irc: Remote host closed the connection | 19:35 | |
| fijal | Alex_Gaynor: and potentially tons of guards and stuff | 19:35 |
| fijal | but yes | 19:35 |
| Alex_Gaynor | fijal: there are never guards in the entry | 19:35 |
| fijal | eh | 19:35 |
| fijal | I said calling to the loop, not to preamble | 19:36 |
| Alex_Gaynor | that's harder :) | 19:36 |
| timotimo | hrm. | 19:36 |
| timotimo | that's a bit big to just jump in | 19:37 |
| timotimo | well, it's not that important anyway | 19:37 |
| lesshaste | how do they measure memory usage (in linux) for benchmarks like http://attractivechaos.github.com/plb/? | 19:43 |
| timotimo | lesshaste: use "time" | 19:45 |
| timotimo | (the one your OS offers, not the one your shell offers) | 19:46 |
| lesshaste | timotimo: how do I get the OS one not the shell one? | 19:46 |
| timotimo | in zsh, just write =time | 19:46 |
| timotimo | otherwise use which to find out the absolute path and run that | 19:46 |
| fijal | lesshaste: there is no reasonable way to measure memory in linux | 19:46 |
| fijal | lesshaste: and those benchmarks are very shitty | 19:46 |
| fijal | /usr/bin/time is the one OS offers | 19:47 |
| lesshaste | timotimo: I get 0.00user 0.00system 0:00.01elapsed 23%CPU (0avgtext+0avgdata 2512maxresident)k | 19:47 |
| lesshaste | 8inputs+0outputs (0major+204minor)pagefaults 0swaps | 19:47 |
| timotimo | yeah, the maxresident size is probably the most interesting one | 19:47 |
| timotimo | but fijal is correct | 19:47 |
| mwhudson (~mwh@linaro/mwhudson) left irc: Ping timeout: 245 seconds | 19:47 | |
| lesshaste | sure I know the problems in linux | 19:47 |
| lesshaste | timotimo: that figure is printed *before* the output to the program! | 19:48 |
| timotimo | you have a more complicated commandline, do you not? | 19:48 |
| lesshaste | fijal: they are indeed.. I was trying to make some sense of them | 19:48 |
| lesshaste | /usr/bin/time cat sudoku.txt|~/Desktop/pypy-c-jit-51333-eb0269c21eec-linux/bin/pypy sudoku_v1.py | 19:48 |
| timotimo | yes, that measures how long cat takes to output the text file | 19:49 |
| lesshaste | right | 19:49 |
| timotimo | put the time after the pipe and you'll be fine | 19:49 |
| lesshaste | good point | 19:49 |
| lesshaste | thanks | 19:49 |
| lesshaste | it's a stupid benchmark.. I need to work out they repeated in 50 times | 19:50 |
| lesshaste | the code the point to isn't exactly the benchmarks they did | 19:50 |
| lesshaste | which is annoying | 19:50 |
| lesshaste | work out how they... | 19:50 |
| fijal | lesshaste: those are really silly benchmarks, I would not spend time there | 19:51 |
| lesshaste | fijal: you make a good point of course. The shedskin asked me to run them though so I would like to give pypy a reasonable chance | 19:52 |
| lesshaste | which means not having them run for 1 second | 19:52 |
| JaRoel|4d (~jaroel|4d@2001:980:7133:1:b812:5826:d3e:b3e1) joined #pypy. | 19:53 | |
| mwhudson (~mwh@linaro/mwhudson) joined #pypy. | 19:54 | |
| lesshaste | The shedskin guy | 19:55 |
| fijal | lesshaste: no, 1s is fine for such benchmarks | 19:55 |
| lesshaste | of course I could just ignore him :) | 19:55 |
| fijal | but they're seriously dumb | 19:55 |
| lesshaste | fijal: isn't there a warm up cost for pypy? | 19:55 |
| fijal | yeah, but it depends on code size | 19:55 |
| fijal | and this code is tiny | 19:55 |
| lesshaste | I mean shedskin does run a compiler first which isn't counted in the time :) | 19:55 |
| fijal | lesshaste: if he wants to brag how shedskin is faster on those examples, let him | 19:56 |
| lesshaste | fijal: in the sudoko code it does look like the first iteration is much slower | 19:56 |
| lesshaste | fijal: the thing is that I suspect it isn't if the tests are made sensible :) | 19:56 |
| fijal | well | 19:56 |
| fijal | I guess whatever, really | 19:56 |
| nedbat (~nedbat@python/psf/nedbat) joined #pypy. | 19:56 | |
| fijal | shedskin would do stuff that will be faster *because it's not python* | 19:56 |
| lesshaste | maybe I should just add the compiler time :) | 19:56 |
| fijal | I don't really care | 19:56 |
| woyna (woyna@79.133.200.48) left irc: Excess Flood | 19:56 | |
| tmu (~tmu@evil.oulu.nixu.fi) got netsplit. | 19:57 | |
| EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) got netsplit. | 19:57 | |
| setmeaway (~setmeaway@118.45.149.247) got netsplit. | 19:57 | |
| Arnar (~arnarb@li187-148.members.linode.com) got netsplit. | 19:57 | |
| panni_ (smuxi@descent.fragstore.net) got netsplit. | 19:57 | |
| bogner` (~bogner@2600:3c03::f03c:91ff:fedf:7ef4) got netsplit. | 19:57 | |
| ericholscher (~eric@173.203.223.63) got netsplit. | 19:57 | |
| hakanardo (~hakan@h59ec009e.seluldx.dyn.perspektivbredband.net) got netsplit. | 19:57 | |
| tito (~tito@pdpc/supporter/active/tito) got netsplit. | 19:57 | |
| metvik (~metvik@unaffiliated/metvik) got netsplit. | 19:57 | |
| smuxi (smuxi@descent.fragstore.net) joined #pypy. | 19:57 | |
| Nick change: smuxi -> panni_ | 19:57 | |
| Possible future nick collision: panni_ | 19:57 | |
| bogner (bogner@2600:3c03::f03c:91ff:fedf:7ef4) joined #pypy. | 19:57 | |
| tmu (~tmu@evil.oulu.nixu.fi) returned to #pypy. | 19:57 | |
| lesshaste | fijal: talking of which... are the any rpython benchmarks? | 19:57 |
| lesshaste | are there | 19:57 |
| tito (~tito@178.73.216.85) joined #pypy. | 19:57 | |
| tito (~tito@178.73.216.85) left irc: Changing host | 19:57 | |
| tito (~tito@pdpc/supporter/active/tito) joined #pypy. | 19:57 | |
| EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) returned to #pypy. | 19:57 | |
| hakanardo (~hakan@h59ec009e.seluldx.dyn.perspektivbredband.net) returned to #pypy. | 19:57 | |
| setmeaway (stemearay@118.45.149.247) joined #pypy. | 19:57 | |
| lesshaste | fijal: you do compare pypy to C and C certainly isn't python :) | 19:57 |
| timotimo | rpython is seriously not meant to be used for production apps that are not interpreters | 19:58 |
| timotimo | well, that wasn't worded very well | 19:58 |
| fijal | lesshaste: yes, but I mean people spend time optimizing those benchmarks for shedskin | 19:58 |
| ericholscher (~eric@173.203.223.63) returned to #pypy. | 19:59 | |
| fijal | lesshaste: but noone looked into optimizing them for pypy | 19:59 |
| lesshaste | fijal: a good point | 19:59 |
| fijal | I know how Mark goes about it, I've seen it "if you tweak this tiny detail, shedskin goes much faster" | 19:59 |
| lesshaste | fijal: oh yes I know all about that | 19:59 |
| fijal | and some of those are enforced because otherwise it won't run at all | 19:59 |
| lesshaste | fijal: I am not blind to these things :) | 19:59 |
| lesshaste | which is why I thought at least if I ran the tests I wouldn't cheat | 19:59 |
| fijal | what I'm saying is that warmup time is probably not a big deal, but I don't care in optimizing those ones | 19:59 |
| lesshaste | sure | 20:00 |
| timotimo | yeah, pypy2 will have a static compiler-like preprocessor in front of the jitted interpreter that statically pre-optimises code that is sufficiently shedskin-like!!!!! not | 20:00 |
| fijal | timotimo: more like a thing that complains :) | 20:01 |
| Arnar (~arnarb@li187-148.members.linode.com) returned to #pypy. | 20:02 | |
| timotimo | that sounds sane | 20:02 |
| mattip_ (4fb1711d@gateway/web/freenode/ip.79.177.113.29) joined #pypy. | 20:02 | |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 20:03 | |
| mattip (4fb1711d@gateway/web/freenode/ip.79.177.113.29) left irc: Ping timeout: 258 seconds | 20:04 | |
| Nick change: mattip_ -> mattip | 20:05 | |
| mfoord (~anonymous@python/psf/voidspace) left irc: Client Quit | 20:06 | |
| Guest38408 (~none@2001:41d0:2:936d::1) left irc: Excess Flood | 20:06 | |
| bogner` (~bogner@2600:3c03::f03c:91ff:fedf:7ef4) got lost in the net-split. | 20:08 | |
| metvik (~metvik@unaffiliated/metvik) got lost in the net-split. | 20:08 | |
| PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy. | 20:09 | |
| kenaan | 12fijal default 11f6b8525d8a10 15/pypy/objspace/std/unicodeobject.py: A bit experimental - try to preallocate the size of unicode join and remove a pointless performance hack (the gener... | 20:16 |
| kenaan | 12fijal default 115e43d79c76a7 15/pypy/: mere | 20:16 |
| Alex_Gaynor | fijal: ugh, this should reall- share with strs version :/ | 20:17 |
| fijal | yes | 20:17 |
| Alex_Gaynor | same for the whole file really :/ | 20:18 |
| fijal | possibly yes | 20:19 |
| fijal | this is one of the very few things that might not share code actually | 20:19 |
| Alex_Gaynor | why not? | 20:19 |
| lesshaste | fijal: you have persuaded me not to reply to Mark | 20:22 |
| fijal | because the string join has a fallback to unicode path | 20:22 |
| fijal | lesshaste: why? | 20:22 |
| lesshaste | fijal: because of the points you made above | 20:22 |
| lesshaste | fijal: about the benchmarks being tailored for a particular purpose | 20:23 |
| danchr (~danchr@cl-848.chi-02.us.sixxs.net) left irc: Ping timeout: 272 seconds | 20:23 | |
| dukebilgewater (~dukebilge@cpe-74-78-69-183.maine.res.rr.com) joined #pypy. | 20:23 | |
| lesshaste | fijal: you didn't realise you were so persuasive? :) | 20:24 |
| fijal | lesshaste: I also said "let him brag about it" | 20:26 |
| fijal | seriously though | 20:26 |
| fijal | I think the space that mark is aiming for is already occupied by cython | 20:27 |
| fijal | and he has not enough power to beat cython PR | 20:27 |
| timotimo | especially when cython gets the pypy backend | 20:27 |
| fijal | I don't think it's relevant | 20:27 |
| DasIch_ (~DasIch@p4FFDED18.dip.t-dialin.net) joined #pypy. | 20:32 | |
| DasIch (~DasIch@p3E991405.dip.t-dialin.net) left irc: Ping timeout: 252 seconds | 20:34 | |
| Nick change: DasIch_ -> DasIch | 20:34 | |
| fijal | lesshaste: so seriously, he's a bit barking at the wrong tree. | 20:35 |
| fijal | if you want to write those benchmarks in pypy, use numpy | 20:35 |
| aboudreault (~alanb@osgeo/member/aboudreault) left irc: Quit: Leaving | 20:42 | |
| lambacck (~chris@d24-150-124-118.home.cgocable.net) left irc: Quit: lambacck | 20:42 | |
| fijal | mattip: your fix is wrong IMO | 20:42 |
| fijal | first_line is supposed to mean "we never touched this field" | 20:43 |
| fijal | first_line should be true for a while and then false imo | 20:44 |
| fijal | it should never change back to true | 20:44 |
| fijal | oh maybe not | 20:44 |
| fijal | right | 20:45 |
| fijal | ok :) | 20:45 |
| mattip | :) | 20:50 |
| mattip | even cooler than you thought | 20:51 |
| danchr_ (~danchr@cl-848.chi-02.us.sixxs.net) joined #pypy. | 20:53 | |
| Rhy0lite (dje@nat/ibm/x-nqxtecyvgqkpgfrd) left irc: Quit: Leaving | 20:54 | |
| fijal | ya | 20:55 |
| fijal | mattip: note it's not as fast as it should be though | 20:55 |
| fijal | but that's unrelated to numpy itself | 20:56 |
| mattip | right. but the slowdown is really the test that was there in the original code. | 20:56 |
| mattip | anyhow I like compact and correct better than fast :) | 20:57 |
| danchr_ (~danchr@cl-848.chi-02.us.sixxs.net) left irc: Ping timeout: 248 seconds | 20:57 | |
| mattip | well, with the usual caveats (not too much slower) | 20:57 |
| lambacck (~chris@d24-150-124-118.home.cgocable.net) joined #pypy. | 21:00 | |
| lizardo (~lizardo@189.2.128.130) left irc: Quit: Leaving | 21:01 | |
| Trundle (~andy@89.204.139.101) joined #pypy. | 21:02 | |
| Trundle (~andy@89.204.139.101) left irc: Changing host | 21:02 | |
| Trundle (~andy@python/site-packages/trundle) joined #pypy. | 21:02 | |
| papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) joined #pypy. | 21:12 | |
| tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy. | 21:24 | |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy. | 21:39 | |
| rguillebert | amaury_, good, I worked a bit on py3k | 21:59 |
| rguillebert | the parser now refuses the old octal syntax | 21:59 |
| amaury_ | did you change it? I have seen only changes in tests | 22:00 |
| rguillebert | I did but I didn't push it, can you take a look before ? | 22:02 |
| amaury_ | did you change genpytokenize.py? | 22:04 |
| rguillebert | yes | 22:04 |
| rguillebert | I took what you did for the long syntax as an example | 22:05 |
| mwhudson (~mwh@linaro/mwhudson) left irc: Ping timeout: 248 seconds | 22:05 | |
| amaury_ | yes, probably similar | 22:05 |
| rguillebert | it's just a bit different because 0 was parsed as an octal number | 22:06 |
| stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy. | 22:06 | |
| rguillebert | it's there https://bitbucket.org/rguillebert/pypy | 22:06 |
| amaury_ | if your test pass, then you did it correctly :) | 22:06 |
| rguillebert | it does (at least for the interpreter tests) except for one that uses the old syntax | 22:08 |
| amaury_ | :) | 22:08 |
| amaury_ | why did you need this "zero" target? | 22:08 |
| amaury_ | could you accept all digits in decNumber instead? | 22:09 |
| rguillebert | no because python3 rejects numbers that starts with 0 | 22:09 |
| amaury_ | oh | 22:10 |
| amaury_ | true | 22:10 |
| rguillebert | and I tried to "only 0" inside decNumber, but it looked bad and didn't work | 22:11 |
| amaury_ | What happens with "0x"? | 22:13 |
| amaury_ | is it parsed as 0 followed by an identifier? | 22:13 |
| rguillebert | it works fine so I guess it's parsed correctly | 22:15 |
| antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) joined #pypy. | 22:16 | |
| amaury_ | "works fine" | 22:16 |
| amaury_ | should return "Syntax Error: invalid token" | 22:16 |
| rguillebert | >>>> 0x12 | 22:16 |
| rguillebert | 18 | 22:16 |
| amaury_ | 0x alone | 22:16 |
| rguillebert | ah | 22:16 |
| rguillebert | SyntaxError | 22:16 |
| amaury_ | and the error string? | 22:17 |
| rguillebert | invalid syntax | 22:17 |
| rguillebert | it's the same if I do 012 or 12L | 22:17 |
| amaury_ | CPython has a different message | 22:17 |
| amaury_ | for 0x and 1x | 22:18 |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Read error: Connection timed out | 22:18 | |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 22:18 | |
| rguillebert | Is there a case where PyPy returns that message ? | 22:20 |
| nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy. | 22:20 | |
| Nick change: nanonyme_ -> nanonyme | 22:22 | |
| mfoord (~anonymous@python/psf/voidspace) left irc: Ping timeout: 240 seconds | 22:23 | |
| amaury_ | rguillebert: no | 22:24 |
| amaury_ | so ignore my remark | 22:24 |
| rguillebert | can I push it ? | 22:26 |
| mwhudson (~mwh@linaro/mwhudson) joined #pypy. | 22:27 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy. | 22:27 | |
| mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host | 22:27 | |
| mfoord (~anonymous@python/psf/voidspace) joined #pypy. | 22:27 | |
| amaury_ | yes | 22:28 |
| amaury_ | btw, did you try a translation? | 22:28 |
| rguillebert | yes it translates | 22:29 |
| amaury_ | ./pypy-c crashes (in interactive mode only), I think I found why | 22:29 |
| nickik (~nick@217.150.234.222) left irc: Quit: Verlassend | 22:29 | |
| amaury_ | something related to the termio module | 22:29 |
| rguillebert | do you get 'NoneType' object is not iterable ? | 22:29 |
| nickik (~nick@217.150.234.222) joined #pypy. | 22:29 | |
| amaury_ | I don't remember | 22:29 |
| rguillebert | that's what I get even if I try to run something | 22:31 |
| rguillebert | py.py works though | 22:31 |
| amaury_ | my pypy-c works when I run a command | 22:33 |
| amaury_ | ./pypy-c -c "print(1)" | 22:33 |
| amaury_ | or more complex | 22:33 |
| amaury_ | ./pypy-c -m test test_list | 22:33 |
| rguillebert | it doesn't :s | 22:35 |
| amaury_ | traceback? | 22:36 |
| rguillebert | it's an OperationError so I guess it's at interp-level | 22:36 |
| amaury_ | do you happen to have non-ascii7 characters in your environment? | 22:37 |
| rguillebert | I was running it on tannit but my localisation variables where set to french | 22:39 |
| stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Quit: schnarch | 22:39 | |
| rguillebert | indeed it works | 22:39 |
| rguillebert | if I set LC_ALL to C | 22:40 |
| amaury_ | ah, I get the error if I set "LANG=fr" | 22:40 |
| rguillebert | I'll take a look at it tomorrow | 22:45 |
| rguillebert | if I do make lldebug in the source directory I should get a proper traceback right ? | 22:46 |
| tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection | 22:47 | |
| amaury_ | hum, not sure | 22:50 |
| amaury_ | I'm already using lldebug | 22:50 |
| mattip (4fb1711d@gateway/web/freenode/ip.79.177.113.29) left irc: Quit: Page closed | 22:53 | |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/778 | 23:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/206 | 23:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/1015 | 23:00 |
| bbot2 | Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/34 | 23:00 |
| rguillebert | so it should be ok now ? | 23:00 |
| mikefc_ (9665bcc5@gateway/web/freenode/ip.150.101.188.197) joined #pypy. | 23:01 | |
| dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) left irc: Quit: Leaving | 23:05 | |
| papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep. | 23:12 | |
| amaury_ | rguillebert: please submit your code, I'll try to find this issue with LANG=fr | 23:16 |
| kenaan | 12rguillebert py3k 11d745dbc657b4 15/pypy/interpreter/: Reject old octal literals | 23:17 |
| kenaan | 12rguillebert py3k 114714f110a1ae 15/pypy/translator/goal/nanos.py: Use new octal literals | 23:17 |
| amaury_ | yeah, now I get a "normal" crash in interactive mode | 23:17 |
| amaury_ | some SyntaxError in pyrepl module | 23:17 |
| amaury_ | which is not at all ready for py3k | 23:17 |
| rguillebert | I don't know why banning the old syntax broke the translation | 23:21 |
| squiddy (~squiddy@g224197249.adsl.alicedsl.de) left irc: Ping timeout: 252 seconds | 23:21 | |
| danchr (~danchr@cl-848.chi-02.us.sixxs.net) joined #pypy. | 23:21 | |
| amaury_ | did it? | 23:21 |
| amaury_ | maybe some applevel code | 23:22 |
| rguillebert | yes, that's why I changed nanos.py | 23:22 |
| amaury_ | where does it break? | 23:22 |
| rguillebert | I don't remember but there's code at applevel | 23:23 |
| rguillebert | so it's probably in there | 23:23 |
| amaury_ | rguillebert: the issue with "LANG=fr" comes from sys.getfilesystemencoding() which returns None and breaks initstdio() in app_main.py | 23:24 |
| rguillebert | ok | 23:24 |
| amaury_ | I won't fix it now, but so you know for tomorrow :) | 23:24 |
| rguillebert | yes, I'll look at it | 23:25 |
| mutux (~muratux@139.179.102.153) joined #pypy. | 23:26 | |
| kenaan | 12amauryfa py3k 11a20df1bb1bb8 15/pypy/module/imp/: PEP3147: .pyc files are now named ./__pycache__/foo.pypy-17.pyc (Note: the "nolonepycfile" option makes less sense... | 23:26 |
| kenaan | 12amauryfa py3k 11708194fb76ef 15/lib_pypy/binascii.py: Remove long() and L suffix from binascii.py | 23:26 |
| kenaan | 12amauryfa py3k 11dcae009093d8 15/pypy/objspace/std/longtype.py: Fix a test in test_longobject.py | 23:26 |
| kenaan | 12amauryfa py3k 114a6fb1817feb 15/pypy/module/termios/: Fixes in termios module | 23:26 |
| kenaan | 12amauryfa py3k 113843c6835175 15/pypy/objspace/fake/objspace.py: bin/checkmodule.py: Add support for space.wrabytes | 23:26 |
| squiddy (~squiddy@g229037172.adsl.alicedsl.de) joined #pypy. | 23:29 | |
| voidspace (~voidspace@python/psf/voidspace) left irc: Ping timeout: 240 seconds | 23:35 | |
| mfoord (~anonymous@python/psf/voidspace) left irc: Ping timeout: 260 seconds | 23:35 | |
| gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated! | 23:36 | |
| rguillebert (~rguillebe@89-40.5-85.cust.bluewin.ch) left irc: Remote host closed the connection | 23:38 | |
| Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Quit: Ex+re | 23:40 | |
| antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) left irc: Ping timeout: 252 seconds | 23:43 | |
| JaRoel (~jaroel|4d@sink.jaroel.nl) joined #pypy. | 23:47 | |
| JaRoel|4d (~jaroel|4d@2001:980:7133:1:b812:5826:d3e:b3e1) left irc: Ping timeout: 260 seconds | 23:50 | |
| stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer | 23:56 | |
| stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy. | 23:56 | |
| Trundle (~andy@python/site-packages/trundle) left irc: Ping timeout: 240 seconds | 23:58 | |
| zain (~textual@c-67-160-201-63.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep. | 23:59 | |
| --- Tue Jan 17 2012 | 00:00 | |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!