#pypy IRC log for Tuesday, 2011-12-06

nedbat (~nedbat@python/psf/nedbat) joined #pypy.00:07
lac (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.00:13
lac_ (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Ping timeout: 252 seconds00:14
dmalcolm (david@nat/redhat/x-fvsigngeczovtbqf) left irc: Quit: Leaving00:15
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 252 seconds00:18
tlynn_ (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) left irc: Ping timeout: 240 seconds00:25
mat^2 (~mathias@212.130.113.35) left irc: 00:27
mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.00:34
mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host00:34
mfoord (~anonymous@python/psf/voidspace) joined #pypy.00:34
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/73500:37
ousado_ (~ousado@hmbg-5f76557a.pool.mediaWays.net) joined #pypy.00:37
ousado (~ousado@hmbg-5f762bbb.pool.mediaWays.net) left irc: Ping timeout: 244 seconds00:41
davisagli (davisagli@davisagli.com) left #pypy.00:50
derdon (~derdon@p5DE894EA.dip.t-dialin.net) left irc: Remote host closed the connection01:05
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) joined #pypy.01:07
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/16101:11
etrepum (~bob@accessnat4.mochimedia.net) left irc: Ping timeout: 244 seconds01:11
aurojit (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Remote host closed the connection01:31
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.01:32
stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) joined #pypy.01:35
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Ping timeout: 248 seconds01:36
stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) left irc: Client Quit01:37
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/187801:37
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/61501:37
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/63301:37
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/114301:37
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/29601:37
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/81801:37
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/146001:37
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/73901:37
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/28401:37
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) joined #pypy.01:41
mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord01:43
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.01:45
ericflo (~ericflo@75.103.8.110) joined #pypy.01:50
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Remote host closed the connection01:58
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.01:59
stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) joined #pypy.02:01
stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) left irc: Client Quit02:01
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Ping timeout: 240 seconds02:03
stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) joined #pypy.02:03
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/28402:05
brutal_chaoshow do I benchmark pypy?02:08
neunon (~snoonan@unaffiliated/neunon) left irc: Read error: Operation timed out02:08
stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) left irc: Quit: schnarch02:22
brutal_chaosabout how long does it take to run the benchmarks?02:33
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.02:37
Shanita (~John@osbk-4db15ac2.pool.mediaWays.net) joined #pypy.02:41
Kaskuka (~John@osbk-4db17372.pool.mediaWays.net) left irc: Ping timeout: 240 seconds02:41
PqSe (~joe@cafe.rspkg.com) left irc: Remote host closed the connection02:53
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/96502:56
ericflo (~ericflo@75.103.8.110) left irc: Quit: ericflo02:56
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!03:11
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 258 seconds03:18
asmeurer__ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Quit: asmeurer__03:30
harrison (~quassel@adsl-69-209-194-12.dsl.chcgil.sbcglobal.net) left irc: Remote host closed the connection03:34
davisagli (~davisagli@davisagli.com) joined #pypy.03:56
kennethreitz (~kennethre@adsl-99-191-194-225.dsl.pltn13.sbcglobal.net) joined #pypy.03:57
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 252 seconds03:57
mattip (~mp@bzq-79-177-116-180.red.bezeqint.net) joined #pypy.03:59
mattip (mp@bzq-79-177-116-180.red.bezeqint.net) left #pypy.04:01
mattip (~mp@bzq-79-177-116-180.red.bezeqint.net) joined #pypy.04:04
mattiphi04:05
mattip (mp@bzq-79-177-116-180.red.bezeqint.net) left #pypy.04:05
bbot23Success: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/187804:06
kushal (~kdas@114.143.162.64) joined #pypy.05:19
kushal (~kdas@114.143.162.64) left irc: Changing host05:19
kushal (~kdas@fedora/kushal) joined #pypy.05:19
kennethreitz (~kennethre@adsl-99-191-194-225.dsl.pltn13.sbcglobal.net) left irc: Quit: Computer has gone to sleep.05:19
kushal (~kdas@fedora/kushal) left irc: Client Quit05:22
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/81805:23
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/63305:32
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/146005:33
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/114305:38
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/61505:44
kennethreitz (~kennethre@ip65-46-252-62.z252-46-65.customer.algx.net) joined #pypy.05:49
kennethreitz (~kennethre@ip65-46-252-62.z252-46-65.customer.algx.net) left irc: Quit: Computer has gone to sleep.06:02
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.06:04
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Remote host closed the connection06:05
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.06:05
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Ping timeout: 255 seconds06:09
Nick change: Gulopine -> Gulaway06:19
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.06:23
fijal (~fijal@197.171.140.170) joined #pypy.06:24
mattip (c1a94669@gateway/web/freenode/ip.193.169.70.105) joined #pypy.06:35
mattiphi06:35
mattipfijal: ping06:36
fijalpong06:37
mattipI committed a version of dot.06:37
mattipIt seems a bit tangential to your efforts to unify iterators.06:37
mattipIt seems to work for scalar, 2d, and multiple dimensional arrays, I even copied the numpy docstrings.06:38
mattipI also checked slices.06:38
mattipIt also does not stuff up the zjit tests.06:39
mattipI would be interested in your review, not necessarily for merging into default but more for the approach taken.06:40
mattipWhen you have a chance.06:42
mattipthat's all for now.06:44
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 240 seconds06:46
overminder (overminder@wf124-152.ust.hk) joined #pypy.06:46
fijalmattip: ok, cool06:48
fijalI'm a bit behind on reviews, I'll try to merge reshape today06:48
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.06:48
gutworth (~benjamin@dhcp-128-36-84-91.central.yale.edu) left irc: Ping timeout: 244 seconds06:49
mattipok. It sure would make writing tests easier.06:50
Action: mattip nose-> grindstone06:50
mattip (c1a94669@gateway/web/freenode/ip.193.169.70.105) left irc: Quit: Page closed06:51
Alex_Gaynorfijal: FWIW jterrace looked at reshape and said it seemed to work for him06:52
overminder (overminder@wf124-152.ust.hk) left irc: Quit: Leaving06:53
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy06:53
bbot23Success: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/73906:53
overminder (overminder@wf124-152.ust.hk) joined #pypy.06:55
overminder (overminder@wf124-152.ust.hk) left irc: Client Quit06:56
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 244 seconds06:57
overminder (overminder@wf124-152.ust.hk) joined #pypy.06:58
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.06:58
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/29607:05
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 240 seconds07:10
kenaan12hakanardo jit-targets 114c76bc680c2f 15/pypy/jit/metainterp/test/test_memmgr.py: suport for setup_class and teardown_class07:12
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.07:16
Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Quit: Ex+re07:18
mattip (c1a94669@gateway/web/freenode/ip.193.169.70.105) joined #pypy.07:28
mattiphi07:28
fijalwb07:28
mattipforgot something: if you do look at reshape today, I had a problem I didn't know how to solve s I left it as a skipped test.07:29
mattipb=a.reshape((2,3,4))07:29
mattipworks, but07:29
mattipb=a.reshape(2, 3, 4)07:29
mattipis skipped cuz I couldn't work out how to handle *args 07:30
fijalah07:30
fijalthat's relatively easy07:31
fijalyou pass __args__07:31
fijaland then do parsing on them07:31
fijalI can do it if you don't feel like it07:31
fijalbut you can just look up for __args__07:31
mattipOK, great, if you don't get around to it I will fix it later (after european work hours)07:31
fijalok, cool07:32
fijalI'll try to get to that07:32
mattipbye now.07:32
fijalthere is however a cool wave today ;-)07:32
mattip;)07:32
Alex_Gaynorpff, the docstring for reshape isnt' accurte07:32
mattiphuh?07:33
Alex_Gaynormattip: numpy's docstring doesn't mention you can pas *args07:33
squiddy (~squiddy@f053081179.adsl.alicedsl.de) joined #pypy.07:34
fijalindeed07:34
`fox` (~fox@95.232.111.114) joined #pypy.07:34
mattipok, so maybe it's a non-documented feature. I personally use it though, so I want it :).07:38
fijalmattip: I'm fine saying it's an unsupported mess07:38
fijalah! :)07:38
mattipBesides, I learned something new: __args__07:38
fijalis it documented anywhere?07:38
fijalyou can also use args_w07:39
mattipcool. Gotta run.07:40
mattip (c1a94669@gateway/web/freenode/ip.193.169.70.105) left irc: Quit: Page closed07:40
overminder (overminder@wf124-152.ust.hk) left irc: Ping timeout: 252 seconds07:40
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.07:47
cpets (c358b60a@gateway/web/freenode/ip.195.88.182.10) joined #pypy.07:47
JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy.07:48
roelbruggink (~jaroel|4d@office.fourdigits.nl) joined #pypy.07:48
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 255 seconds07:52
JaRoel (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 268 seconds07:53
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Remote host closed the connection08:01
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.08:01
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Ping timeout: 252 seconds08:06
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 240 seconds08:11
Alex_Gaynorfijal: less articles, more reshape :)08:14
fijalI'm on holiday ;-)08:14
Action: Alex_Gaynor can't wait for a holiday so he can have some pypy time08:15
fijalhow mattip manages branches so they create bogus conflicts?08:16
fijalI honestly doubt it's his issue, rather mercurials08:16
Alex_Gaynormerge conflict?  blame hg :)08:17
fijalwho renamed NDimArray to W_NDimArray?08:17
Alex_GaynorMe08:17
fijalokey08:20
fijalwhy not W_NdimSlice?08:20
fijalexport float32 :)08:20
Alex_Gaynorfijal: I exported flaot3208:20
Alex_Gaynorfijal: fair question, I don't have a good answer08:20
fijalanyway, don't do it *now*08:22
fijaluntil I merge08:22
Alex_GaynorI'm not, I'm writing a paper08:22
`fox` (~fox@95.232.111.114) left irc: Ping timeout: 248 seconds08:25
roelbruggink (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection08:26
kenaan12fijal default 11ea547b8be18f 15/pypy/module/micronumpy/: merge matrix-reshape-merge branch. Thanks mattip for doing that.  Adds settable shape of an array as well as reshap...08:26
fijalgood08:26
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.08:26
Alex_Gaynorclose the branch?08:27
fijalyes yes, pushing :)08:27
kenaan12fijal default 11e36a33ee5710 15/: close merged branch08:27
kenaan12fijal matrixmath-reshape 11c8b9ca8d5c25 15/: close merged branch08:27
Alex_Gaynoreh, you closed default?08:27
fijalnice08:29
fijaldo anything to it and it'll unclose08:30
kenaan12fijal matrixmath-reshape-merge 1183b5695339e8 15/: close merged branch08:30
Alex_Gaynorfijal: feel like unclosing it now :)08:30
fijalwhen I have something to commit08:31
fijalit's not harmful or anything08:31
Alex_Gaynorok08:31
kenaan12fijal default 11ea4102860b98 15/pypy/module/micronumpy/: rename NDimSlice to W_NDimSlice, kill some unused imports08:33
fijaldone :)08:36
Alex_Gaynor:)08:36
Alex_Gaynorcool08:36
fijalso08:37
fijalmy idea about signatures08:38
fijalis so we can consider sharing arrays if we want to as well08:38
fijalessentially you provide numbering08:38
fijalfor iterators in the first place08:38
fijalso say (a + b) * (c + d)08:38
fijalnote that a and b and c and d have to be the same size for it to work08:38
fijalwould become08:39
fijalMultiply(Add(0, 0), Add(0, 0))08:39
fijaland you have a list of iterators08:39
Alex_GaynorSo you basically build a list of iterators, and then for each op you use the iterator that's there, and let the optimizer remove the dead iterators?08:40
Alex_Gaynorthat lets you remove the duplicate counters, how do you show teh JIT that the arrays are the same obj?08:40
fijalyou can provide another numbering about the same arrays08:40
Alex_Gaynorok, seems sane-ish to me08:41
Alex_Gaynorwill require a refactoring of the Signature08:41
fijalthe thing is this way you also get rid of bridges with tons of guards08:41
fijalbecause you walk signatures and not iterators08:41
Alex_Gaynorindeed08:41
fijalyou don't have a Call2 iterator anymore08:41
mvt (~mvantelli@87.213.45.85) joined #pypy.08:41
Alex_GaynorI'm not totally sure the optimizer can handle all of this TBH, but if we build a prototype we can always improve the optimizer08:42
Alex_Gaynorparticularly I'm concerned about what exactly you make the greenkey, and if enough is immutable on a signature08:42
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.08:45
fijalyes08:45
fijalno, the optimizer has nothing to do with it08:46
Alex_Gaynorwell you know what I meant08:46
fijalbecause all we need is to have an elaborate enough greenkey08:46
fijalthen everything gets nicely handled at tracing08:46
Alex_Gaynorlist of iterators needs to be virtualizable I think08:47
fijalor you create a new one each iteration08:47
fijalit's easier than current recursive structure08:47
Alex_Gaynorsomething like that08:47
fijalIMO08:47
Alex_Gaynorfine with me, makes sense IMO08:47
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.08:51
bivab (~david@fwstups.cs.uni-duesseldorf.de) joined #pypy.08:52
fijalnow we need some sort of bytecode that we can hash on08:52
fijalprobably08:52
fijalhm08:52
fijalright08:52
fijalwhy don't we just create a bytecode?08:52
Alex_Gaynorfijal: so basically we're reverting to pre-my AST refactor08:52
Alex_Gaynor:)08:52
cpets_ (c358b60a@gateway/web/freenode/ip.195.88.182.10) joined #pypy.08:53
fijalnot quite I guess08:53
fijalbut does not it make sense?08:53
Alex_Gaynorwell, how do you pass around call1/call208:53
Alex_GaynorI guess an alternate fix is to be able to do green key by some eq func, and not id08:53
fijalright08:54
fijalbut isn't it if you have strings?08:54
cpets (c358b60a@gateway/web/freenode/ip.195.88.182.10) left irc: Ping timeout: 265 seconds08:55
Alex_Gaynorno, it'd id() for strings AFAIK, that's why in previous bytecode we cached it after we compiled it08:55
fijalit's equality not id08:55
fijalah08:55
Alex_Gaynorno, I'm pretty sure it'd id08:55
fijalokey08:55
fijalwell, it's easy you intern the string08:55
Alex_Gaynorsure, interning the stinrg isn't the complex part of this08:56
fijalbut yeah, essentially08:57
fijalit seems your solution is not flying any more :)08:57
Alex_Gaynormaybe it's not a 747, but it's a nice zeppellin08:57
fijal:]09:03
Alex_Gaynorfijal: actually, if you cache each individual instance in the tree maybe it works, or maybe it's a mess, bit unclear.  maybe try to write a version without a greenkey first09:05
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood09:08
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Remote host closed the connection09:08
fijalok09:08
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.09:09
Alex_Gaynorfijal: so you're volunteering to do this? :)09:11
fijalyes, I guess so09:11
fijalI have another horrible worry09:11
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.09:12
fijalthat it's simply waiting for cache09:12
fijalbecause we create a new array each tmie09:12
Alex_Gaynoryou mean CPU cache09:12
fijalthat's why I would like to have a[:] = b09:12
fijalusing memcopy for now09:12
Alex_Gaynorgive me a way to check if an array is contigious, and I'll do it09:13
fijalif array.strides[-1] == 109:13
fijalthen array is contiguous09:13
fijalno09:13
fijalnot necesarilly09:13
Alex_Gaynorsee, this is why I didn't write this09:13
fijalarray.strides[-1] == 1 and array.backstrides[-1] == shape[0] or so09:14
fijalor shape[-1] :)09:14
fijalok09:14
fijalI can write the function09:14
fijalis_contiguous()09:14
Alex_Gaynoryes :)09:14
fijalbut you fix the spelling09:14
Alex_Gaynoris_contiguous09:14
antocuni (~antocuni@host129-122-dynamic.16-79-r.retail.telecomitalia.it) joined #pypy.09:16
mkerrin (~mkerrin@wg1-nat.fwgal01.deri.ie) joined #pypy.09:18
mkerrin (mkerrin@wg1-nat.fwgal01.deri.ie) left #pypy.09:18
mkerrin (~mkerrin@wg1-nat.fwgal01.deri.ie) joined #pypy.09:23
fijalAlex_Gaynor: do you want some review on the paper btw?09:26
fijalI'm doing numpy on pypy talk at a local PUG on Sat09:26
Alex_Gaynorfijal: the paper I was writing tonight?  not really it's totally nontechnical09:27
fijalokey09:27
Alex_Gaynorfijal: I'll show you my slides on testing in pypy tomorrowish though09:27
fijalcool09:27
fijalI wonder what to talk about09:27
Alex_Gaynorshow our benchmarks :)09:28
apanda (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) joined #pypy.09:29
antocuni (~antocuni@host129-122-dynamic.16-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds09:34
kenaan12ctismer default 11f6023488fc52 15/pypy/: fixed _socket int/uint matters09:41
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.09:42
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.09:46
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) joined #pypy.09:48
ojii (~ojii@40-34.60-188.cust.bluewin.ch) joined #pypy.09:50
hruske (~Gasper@2001:15c0:66e9:14:806e:69ff:fe1e:105f) joined #pypy.10:02
hruske (~Gasper@2001:15c0:66e9:14:806e:69ff:fe1e:105f) left irc: Client Quit10:02
Fluxidafter running benchmark runner i got results like: Min: 0.168075 -> 0.214997: 1.2792x slower � whis value is which? it run pypy firsty and cpython second, so the slower one is cpython?10:23
stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) joined #pypy.10:35
stakkars (~tismer@p5DDB7E68.dip.t-dialin.net) left irc: Read error: Connection reset by peer10:35
Nick change: stakkars_ -> stakkars10:35
rokujyouhitoma (~rokujyouh@122x220x64x254.ap122.ftth.ucom.ne.jp) joined #pypy.10:35
cpets_ (c358b60a@gateway/web/freenode/ip.195.88.182.10) left irc: Quit: Page closed10:36
Fluxiduh, as far i understand the code, second one is pypy, so it's slower?10:39
mkerrin (~mkerrin@wg1-nat.fwgal01.deri.ie) left irc: Quit: mkerrin10:41
ronnyFluxid: what benchmark are you running, and how exactly do you run ti?10:45
ronny*it10:45
Fluxidone moment10:45
bivab (~david@fwstups.cs.uni-duesseldorf.de) left irc: Ping timeout: 244 seconds10:46
Fluxidronny: http://paste.pocoo.org/show/516993/ first one is without jit in pypy. i modified unladen_swallow/perf.py to run 100 iterations instead of 50, but slowdown is about the same10:48
Fluxidresults are rather different from http://speed.pypy.org/comparison/?exe=1%2BL%2Bdefault%2C3%2BL%2Bdefault&ben=27&env=1&hor=false&bas=2%2B472&chart=relative+bars10:50
Fluxidwhat i did wrong?10:50
neunon (~snoonan@unaffiliated/neunon) joined #pypy.10:50
bivab (~david@fwstups.cs.uni-duesseldorf.de) joined #pypy.10:54
antocuni (~antocuni@host129-122-dynamic.16-79-r.retail.telecomitalia.it) joined #pypy.10:55
Fluxidafter running 500 iters it's still ~1.2x slower... why?10:56
lac (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Remote host closed the connection11:00
rokujyouhitoma (~rokujyouh@122x220x64x254.ap122.ftth.ucom.ne.jp) left irc: Remote host closed the connection11:00
ronnyFluxid: i get the same result with nightly11:02
nedbat (~nedbat@python/psf/nedbat) joined #pypy.11:02
Fluxidbut I use 1.7... then why speed.pypy.org shows it's faster? are there more iterations or something? but i don't see a way to set how many iterations a test should do, so... how, why?11:04
ronnynot sure11:06
bivab (~david@fwstups.cs.uni-duesseldorf.de) left irc: Read error: Connection reset by peer11:06
ronnyon the django benchmark its 8x faster tho11:06
bivab (~david@fwstups.cs.uni-duesseldorf.de) joined #pypy.11:06
Fluxidfijal: any ideas?11:08
derdon (~derdon@pD9E1DC47.dip.t-dialin.net) joined #pypy.11:11
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.11:12
nedbat (~nedbat@python/psf/nedbat) left irc: Read error: Connection reset by peer11:13
nedbat (~nedbat@python/psf/nedbat) joined #pypy.11:15
arigato (~arigo@fwstups.cs.uni-duesseldorf.de) joined #pypy.11:15
arigatohi11:16
nedbat (~nedbat@python/psf/nedbat) left irc: Disconnected by services11:16
nedbat_ (~nedbat@python/psf/nedbat) joined #pypy.11:16
squiddy (~squiddy@f053081179.adsl.alicedsl.de) left irc: Remote host closed the connection11:22
arigatofijal: I heard you talked about DAGs11:32
arigatofor numpy compute()11:32
arigatoit looks like a good idea to do something about it11:33
arigatohttp://paste.pocoo.org/show/517004/11:33
arigatothis example takes forever on pypy11:33
arigatoI can explain why after lunch if it's unclear .-)11:33
brutal_chaosfijal: hey, I have test results, should I run saveresults.pt?11:41
brutal_chaossaveresults.py*11:42
Nick change: nedbat_ -> nedbat11:48
brutal_chaosarigato, fijal I have benchmark results from my pypy-1.7 shadowstack clang build. What do I need to send to the mailing list? I have the printed results and the results.json.11:51
espes (~espes@123-243-221-248.tpgi.com.au) left irc: Read error: Connection reset by peer11:53
espes (~espes@123-243-221-248.tpgi.com.au) joined #pypy.11:54
cpets (c358b60a@gateway/web/freenode/ip.195.88.182.10) joined #pypy.12:03
derdon (~derdon@pD9E1DC47.dip.t-dialin.net) left irc: Remote host closed the connection12:07
antocuni (~antocuni@host129-122-dynamic.16-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds12:08
santagada (~leonardo@177.18.63.214) joined #pypy.12:21
antocuni (~antocuni@host129-122-dynamic.16-79-r.retail.telecomitalia.it) joined #pypy.12:56
harrison (~quassel@adsl-69-209-222-31.dsl.chcgil.sbcglobal.net) joined #pypy.13:04
lizardo (~lizardo@189.2.128.130) joined #pypy.13:05
Nick change: DasIch_ -> DasIch13:07
kenaan12l.diekmann type-specialized-instances 11940ed396c2d9 15/pypy/objspace/std/mapdict.py: some fixes for type-specialized-attributes13:11
Rhy0lite (dje@nat/ibm/x-ymiaplfkvqbopvxt) joined #pypy.13:13
kenaan12l.diekmann type-specialized-instances 1168bb4e39d162 15/: merge with default13:15
kenaan12l.diekmann type-specialized-instances 110c894ad8e9b5 15/pypy/objspace/std/mapdict.py: oups13:25
k_bx (~k_bx@195.20.130.1) joined #pypy.13:25
dgl (~dgl@109.86.165.231) joined #pypy.13:48
squiddy (~squiddy@w4089.dip.tu-dresden.de) joined #pypy.13:50
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) joined #pypy.13:56
fijal (~fijal@197.171.140.170) left irc: Read error: Connection reset by peer14:09
kdas_ (~kdas@114.143.162.64) joined #pypy.14:12
cheater (~cheater@178-26-42-68-dynip.superkabel.de) joined #pypy.14:12
cheaterhi14:12
cheaterwhat do i use if i want my pypy based app to read or write xml?14:12
Nick change: Gulaway -> Gulopine14:13
kdas_ (~kdas@114.143.162.64) left irc: Client Quit14:16
ronnycheater: there is xml.etree in the stdlib, what do you intend to do?14:17
cheaterronny: just read and write some xml based config files14:17
cheaterronny: and possibly access some properties using xpath (so i'd need some basic xslt too)14:18
ronnycheater: xml.etree has ElementPath for xpath14:21
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.14:21
ronnycheater: i dont see how xslt is needed for reading values14:21
ronnybtw, why xml config?14:21
cheaterright, it does14:21
cheaterronny: ask the 100 or so people who worked on this Enterprise Megapplication before me :))14:22
ronnyoh, enterprise, i see14:22
ronnycheater: what do you currently use to read the xml?14:23
cheaterlibxml214:23
squiddy (~squiddy@w4089.dip.tu-dresden.de) left irc: Quit: Leaving14:24
ronnybbl14:25
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) joined #pypy.14:25
cheaterronny: so does xml.etree work well under pypy? no problems?14:27
Action: exarkun bets there are problems14:29
gutworth (~benjamin@128.36.84.91) joined #pypy.14:41
Nick change: Gulopine -> Gulaway14:42
whitelynx|work (~whitelynx@63.241.75.144) joined #pypy.14:44
hruske (~Gasper@188-230-156-183.dynamic.t-2.net) joined #pypy.14:47
cwillu_at_work (~cwillu@cwillu-1-pt.tunnel.tserv13.ash1.ipv6.he.net) joined #pypy.14:53
antocuni (~antocuni@host129-122-dynamic.16-79-r.retail.telecomitalia.it) left irc: Quit: Leaving14:56
cpets (c358b60a@gateway/web/freenode/ip.195.88.182.10) left irc: Quit: Page closed14:59
mkerrin (~mkerrin@wg1-nat.fwgal01.deri.ie) joined #pypy.15:10
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 248 seconds15:11
kkris (~kris@80-123-36-77.adsl.highway.telekom.at) joined #pypy.15:21
overminder (~overminde@wf116-092.ust.hk) joined #pypy.15:24
stakkarshi all,15:29
stakkarsgot an issue with micronumpy15:29
stakkarsis micronumpy working otherwise?15:30
stakkarsthis is win32, but I see no reason why win32 should be special?15:30
anders_AU (~dunder@130.225.20.54) joined #pypy.15:31
mattip (~chatzilla@bzq-79-179-96-122.red.bezeqint.net) joined #pypy.15:33
mattiphi15:33
anders_AUHi15:34
fprimex (~fprimex@brent-macbook.sc.fsu.edu) joined #pypy.15:34
anders_AUDoes anyone know why I can't include a specific module with pypy that I can include with cPython?15:34
exarkunanders_AU: Do you mean "import", not "include"?15:35
anders_AUI'm doing some weather data analysis, and would like to use the pygrib package15:35
anders_AUyes, import15:35
anders_AUI do "import pygrib", and i get an error15:35
exarkunDid you install pygrib for PyPy?15:35
anders_AUNot specifically15:35
anders_AUdo you know the module?15:35
exarkunPyPy generally won't find modules that you only installed for CPython.15:35
anders_AUokay15:35
exarkunI don't really know it, no.  From its website, I see it's a wrapper for a C library.15:36
anders_AUyes15:36
exarkunPyPy supports much of the CPython C API, but not all of it.15:36
anders_AUCould I test it by copying the cPython pygrib module into the folder I downloaded from the PyPy homepage?15:36
exarkunYou can try building/installing pygrib for PyPy: pypy setup.py install - just as you might expect - and it may work, or you may find it uses some part of the CPython C API that PyPy does not yet support.15:36
squiddy (~squiddy@f053081179.adsl.alicedsl.de) joined #pypy.15:36
anders_AUI'll try15:37
brutal_chaosanders_AU: did you run 'pypy setup.py' for the module 15:37
exarkunPyPy is not binary (ABI) compatible with CPython, so you have to rebuild it for PyPy, you can't re-use the CPython build.15:37
anders_AUNo. I'll try. I'm quite new to pypy, just discovered it last week15:37
brutal_chaosanders_AU: are you usimng windows?15:38
mattip (chatzilla@bzq-79-179-96-122.red.bezeqint.net) left #pypy.15:38
anders_AUno, I'm on gentoo linux. And I downloaded the binary file from the homepage because I didn't have enough RAM to translate it15:38
brutal_chaosok. (yay gentoo!)15:39
anders_AUpypy-1.7 % find ./ -name "setup.py" returns nothing though15:39
brutal_chaoswin32 buidling (of anythin) is a majnor pain, thaqts why i ask.15:39
brutal_chaosanders_AU: zsh as well?15:39
anders_AUyes15:40
brutal_chaos:)15:40
anders_AU:-)15:40
exarkunanders_AU: I see a setup.py here, https://code.google.com/p/pygrib/source/browse/#svn%2Ftrunk15:40
exarkunanders_AU: Are you looking in the pygrib source?15:41
brutal_chaosanders_AU: you dont need to translate pypy. the setup.py was in reference to the module you are trying to use15:41
exarkunI also see pyx files, though.  Those are Pyrex/Cython, which does not work with PyPy.15:41
anders_AUoh, that setup, I thought setup.py should come from pypy :-)15:41
dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) joined #pypy.15:42
anders_AUI downloaded pygrib through the Gentoo package system, so I'm downloading the tar file now15:42
brutal_chaosanders_AU: it uses pyx files which use code that will not work with pypy15:43
anders_AUTraceback (most recent call last):15:44
anders_AU  File "app_main.py", line 51, in run_toplevel15:44
anders_AU  File "./setup.py", line 2, in <module>15:44
anders_AU    import os, glob, numpy, sys15:44
anders_AUImportError: No module named numpy15:44
anders_AUThis might also explain a few things15:44
anders_AUI need numpypy instead15:45
kenaan12bivab arm-backend-2 1163e21292fab4 15/pypy/jit/backend/: (arigo, bivab): aahhh. Fix an issue with floats that are spilled in a loop and later read in a brigde. Due to...15:46
kenaan12bivab arm-backend-2 115140a209b50a 15/pypy/jit/backend/arm/opassembler.py: remove some unused imports15:46
kenaan12bivab arm-backend-2 119e1351914eb6 15/pypy/jit/backend/arm/codebuilder.py: fix an issue with the private attribute renaming going on when reading a field that starts with '__'  and doe...15:46
anders_AUTraceback (most recent call last):15:46
anders_AU  File "app_main.py", line 51, in run_toplevel15:46
anders_AU  File "./setup.py", line 22, in <module>15:46
anders_AU    incdirs=[numpy.get_include()]15:46
anders_AUAttributeError: 'module' object has no attribute 'get_include'..15:46
anders_AUOkay, I think I'll just dump the data in a binary file and load it from pypy for now 15:46
anders_AUThanks for the help!15:47
linluxiang (~linluxian@183.5.134.180) joined #pypy.15:55
stakkarsarigato: hi15:56
arigatostakkars: re micronumpy: you can ignore it right now, there are also reported issues15:56
arigatoon linux15:56
stakkarsarigato: has micronumpy been changed?15:56
arigatoit's being actively worked on, ye15:56
arigatoyes15:56
stakkarsthanks.15:56
stakkarsarI would like to create a stable win32 and put it into release15:58
arigatowin32-64?15:58
stakkarsno, just the default. still hoping to get it to build, my last crash was mith micronumpy15:59
stakkarsI want a version which does not change, and where I can compare my branch against16:00
stakkarsand which is known to have no issues.16:00
arigatoany particular reason you cannot use the officially released pypy-win32-1.7 ?16:02
stakkarsbecause that is broken, I thought?16:03
stakkarsoh, it is not? ok16:03
linluxiang (~linluxian@183.5.134.180) left irc: Remote host closed the connection16:03
mvt (~mvantelli@87.213.45.85) left irc: Ping timeout: 240 seconds16:04
linluxiang (~linluxian@183.5.134.180) joined #pypy.16:04
kdas_ (~kdas@114.143.162.64) joined #pypy.16:07
Shanita (~John@osbk-4db15ac2.pool.mediaWays.net) left irc: Ping timeout: 240 seconds16:08
EnCuKou (~encukou@a91-156-243-246.elisa-laajakaista.fi) joined #pypy.16:08
arigatobah, a mostly useless but cool feature we can implement nowadays16:10
stakkarsarigato: my only intent was to see if I fixed all issues from the signed/unsigned checkin.16:10
arigatowhen gdb'ing through jit-generated frames16:10
Shanita (~John@osbk-4db15ac2.pool.mediaWays.net) joined #pypy.16:10
arigatowe could have the frames show the correct source file and line number in the .py files...16:10
stakkarswhow16:11
arigatostakkars: ah, ok.  this was done after 1.716:11
anders_AUJust out of curiosity, has anyone gotten parallel python to work? (import pp). Or any other threading package for the matter.16:11
anders_AUAt the moment I'm just dividing my problem between completely separate python processes.16:11
stakkarsarigato: so I should test without micronumpy and hope, right?16:12
arigatostakkars: yes16:12
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection16:17
anders_AU (dunder@130.225.20.54) left #pypy.16:17
kenaan12ctismer win64-stage1 11b853ad20830d 15/: merge default16:18
bivab (~david@fwstups.cs.uni-duesseldorf.de) left irc: Ping timeout: 240 seconds16:18
stakkarsarigato: my goal is to compare win32 default and my branch, to make sure that I don't break anything, because I want to merge16:19
stakkarsso I'm building 32bit in default and win64-stage1, with 32bit only. When they both translate, I can merge for the first time!16:21
arigato:-)16:22
kenaan12arigo default 11b0cec017f8a9 15/: Test and fix.16:22
arigatoindeed, micronumpy does not translate right now16:23
cfbolzno cookies16:24
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.16:24
antocuni (~antocuni@host129-122-dynamic.16-79-r.retail.telecomitalia.it) joined #pypy.16:24
arigatoand it fails its own non-translated tests, too16:25
Action: arigato backs it out16:26
jterrace (~jterrace@aegis.CS.Princeton.EDU) joined #pypy.16:27
linluxiang (~linluxian@183.5.134.180) left irc: Quit: linluxiang16:29
arigatoah, it's only with python2.5, so fine16:31
kenaan12arigo default 11a88a15a5ccb7 15/pypy/module/micronumpy/interp_numarray.py: Translation fix.  No cookie16:44
arigato (~arigo@fwstups.cs.uni-duesseldorf.de) left irc: Quit: See you16:46
leshaste (~lesshaste@87-194-206-189.bethere.co.uk) joined #pypy.16:47
leshasteis there an easy way to install pypy 1.7 in lucid?16:47
exarkunleshaste: download the binary package, unpack it16:48
exarkunrun pypy from it16:48
leshasteok :)16:49
leshastethanks16:49
strtok (~strtok@thales.strtok.net) left irc: Ping timeout: 240 seconds16:49
jterraceleshaste: there is a PPA: https://launchpad.net/~pypy/+archive/ppa might not have lucid for 1.7 though16:51
jterracebut you might just be able to install the oneiric package on lucid anyway16:51
leshasteyes that seems to be 1.316:51
jterraceleshaste: the 1.7 oneiric package should install fine in lucid though, all it has are /usr/bin/pypy and some man pages, nothing oneiric specific16:53
leshasteok thanks.. I just downloaded the package and tried it out16:55
leshasteI only for a 45 to 26 second speed up :)16:55
leshastewhich I am surprisingly disappointed by16:55
leshastegiven how great pypy is :)16:55
jterraceyou mean cpython 45, pypy 26?16:55
K-os (~gabriel@cl-3210.ham-01.de.sixxs.net) joined #pypy.16:56
leshasteI am on 2.6.516:57
fzzzy_ (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.16:57
leshasteand ran the pypy I downloaded16:57
aurojit (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) joined #pypy.16:57
leshaste1.7 I think16:57
leshasteoh.. yes :)16:57
leshastesorry16:57
leshasteI misread what you wrote16:57
jterrace50% is still pretty good16:57
leshasteyes.. it just loops though16:58
leshastewhich is why I thought it might be faster16:58
fzzzy_ (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Client Quit16:58
leshastehttp://pastebin.com/vxWEfz3Y  very simple code16:58
stakkars_ (~tismer@dslb-088-074-053-099.pools.arcor-ip.net) joined #pypy.16:59
leshastemaybe most of the time is spent in the set intersection17:00
leshastewhich is not jit'able17:00
apanda (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) left irc: Ping timeout: 240 seconds17:00
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Ping timeout: 240 seconds17:00
jterraceim guessing the function calls17:00
jterraceyou could profile it17:00
leshasteremind me how please :)17:00
jterracehttp://docs.python.org/library/profile.html17:01
leshasteoh sorry.. I meant whether pypy had profiling17:01
fijal (~fijal@41.55.46.72) joined #pypy.17:03
kdas_ (~kdas@114.143.162.64) left irc: Quit: This computer has gone to sleep17:03
fijalhi17:04
fijalFluxid: what benchmark?17:04
exarkunleshaste: What are you running?17:04
leshastea small piece of code I wrote... http://pastebin.com/vxWEfz3Y17:04
fijalarigato: yes I know :)17:04
kenaan12ctismer win64-stage1 117c2b76526385 15/: Merge with default17:04
dgl (~dgl@109.86.165.231) left irc: Read error: Operation timed out17:05
Fluxidfijal: as in paste, bm_mako17:07
leshastejterrace, so curiously it took 110 seconds when profiled and spends 5625398   29.714    0.000   49.485    0.000 hamming.py:22(hamdist)17:08
fijaldifferent mako?17:08
fijalFluxid: what mako do you run?17:08
leshastewhich looks like half the time in the hamdist function17:08
leshastewhich I would have thought could be sped up a lot17:08
Fluxidi think it was 0.5.0 but i;m not at work anymore so i can't check17:08
leshasteassuming I read the profiling correctly17:08
fijalFluxid: did you just checkout pypy/benchmarks and run runner.py -b mako?17:09
fijalor so17:09
fijalor what did you do?17:09
G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving.17:09
jterraceleshaste: yeah, that's what i thought17:09
leshastejterrace, http://pastebin.com/n1NjzTFa17:09
Fluxidfijal: http://paste.pocoo.org/show/516993/17:10
leshastejterrace, so it spends almost no time doing the intersections?17:10
jterraceset intersection is fast17:10
Fluxidfirst one i tried without jit, and second with no additional options17:10
fijalFluxid: yes, but the second one is a different python17:11
fijallook what it run17:11
fijalI think you're comparing pypy --jit off with cpython17:11
fijal?17:11
leshastejterrace, so the punchline is that I really need to improve my python code :)17:11
fijalif you want to specify both pythons it needs to be specified twice17:11
Fluxidfijal: yes, in the first one i'm indeed comparing pypy --jit off with cpython17:11
leshastejterrace, is it possible to profile the pypy version to see what was sped up?17:11
jterracepretty sure cProfile works in pypy17:12
Fluxidin second one i'm comparing pypy with no addtional options with cpython17:12
jterraceleshaste: might want to try changing that zip to itertools.izip17:12
Fluxidthe path is is almost the same because both pythons are in virtualenvs, cpython 2.7.2 is in py-venv and pypy 1.7 is in pypy-env17:13
fijalhttp://paste.pocoo.org/show/517155/17:13
leshastejterrace, this is what pypy -m cProfile gives me http://pastebin.com/d0RdKdUQ17:13
fijalthose are my results17:14
stakkars_ (~tismer@dslb-088-074-053-099.pools.arcor-ip.net) left irc: Quit: schnarch17:14
hruske (~Gasper@188-230-156-183.dynamic.t-2.net) left irc: Quit: Leaving17:14
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds17:14
Fluxidthen I'll run cpython and pypy outside virtualenvs and make sure mako is newest tomorrow17:15
fijalFluxid: it's using mako from lib17:16
leshastejterrace, the joins are also pointless :)17:16
leshastejterrace, only there as the hamdist function can't cope with lists17:16
Fluxidah, mako is included with benchmarks17:17
aurojit (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) left irc: Remote host closed the connection17:17
Fluxidso it's 0.3.617:17
apanda (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) joined #pypy.17:17
Fluxidfijal: can you check what results do you get with mako 0.5.0 or 0.4.1?17:17
fijalhttp://paste.pocoo.org/show/517157/17:18
fijalthose are tannit benchmarks17:18
fijalso I'll double check why on the server is so much faster than on my laptop17:18
fijalit takes roughly the same time on cpython, but pypy is so much faster there17:18
fijal'some obscure GC options maybe17:18
fijalit might be that the larger cache size make it go *really* fast or so17:18
leshastejterrace, thanks for taking a look.. I'll try to improve the code tomorrow.. no clever compiler can compensate for rubbish python skills :)17:18
fijalFluxid: it's not cross platform (the results) at least :(17:19
Fluxidi can't tell ;)17:19
fijalFluxid: how much cache do you have on your machine?17:19
jterraceleshaste: heh, yeah you can probably make that hamdist function much faster17:19
jterracecould post on stackoverflow, people there love optimizing17:19
fijalFluxid: :/17:19
leshastejterrace, yes and once I have got rid of the joins too it will be a lot better17:19
Fluxidfijal: didn't check, i'll check tomorrow. (how do i do it btw?)17:19
fijalI'll investigate17:19
fijalcat /proc/cpuinfo if you're on linux17:19
Fluxidk17:19
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Ping timeout: 252 seconds17:20
leshaste (~lesshaste@87-194-206-189.bethere.co.uk) left irc: Quit: Leaving17:21
fijal (~fijal@41.55.46.72) left irc: Read error: Connection reset by peer17:21
apanda (~apanda@c-24-7-101-62.hsd1.ca.comcast.net) left irc: Ping timeout: 240 seconds17:21
fzzzy (~donovan@nat/mozilla/x-aubzsvdtmlvajwkj) joined #pypy.17:22
tito (~tito@pdpc/supporter/bronze/tito) left irc: Remote host closed the connection17:22
strtok (~strtok@thales.strtok.net) joined #pypy.17:23
mkerrin (~mkerrin@wg1-nat.fwgal01.deri.ie) left irc: Quit: mkerrin17:24
k_bx (~k_bx@195.20.130.1) left irc: Ping timeout: 255 seconds17:26
tito (~tito@81-67-113-37.rev.numericable.fr) joined #pypy.17:26
tito (~tito@81-67-113-37.rev.numericable.fr) left irc: Changing host17:26
tito (~tito@pdpc/supporter/bronze/tito) joined #pypy.17:26
k_bx (~k_bx@195.20.130.1) joined #pypy.17:26
santagada (~leonardo@177.18.63.214) left irc: Quit: Ex-Chat17:29
jnoller (~jnoller@vpn.nasuni.com) joined #pypy.17:29
tilgovi (~randall@69.111.59.167) joined #pypy.17:29
tilgovi (~randall@69.111.59.167) left irc: Changing host17:29
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.17:29
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.17:30
k_bx (~k_bx@195.20.130.1) left irc: Ping timeout: 252 seconds17:31
K-oshi! just out of interest: how were the benchmarks on http://speed.pypy.org chosen? are those the simply the ones where pypy looks best? ;-)17:34
overminder (~overminde@wf116-092.ust.hk) left irc: Ping timeout: 252 seconds17:35
exarkunK-os: Many of them were taken from the Unladen Swallow project.17:35
exarkunK-os: Others were contributed by real Python projects.17:35
K-osyes, i saw that too. but for example not all are shown online17:36
exarkunnot all what are shown online where?17:36
exarkunPyPy looks good on them because the team has been using them to direct many optimization efforts.17:36
Fluxidhttps://bitbucket.org/pypy/benchmarks/src ?17:36
exarkunIf you have benchmarks for which PyPy performs poorly, many PyPy developers would be interested to see them.17:36
K-oshehe, for example i don't see the bm_regex_* benchmarks on the website, only in the repository17:37
K-osor am i misunderstanding something17:38
exarkunThere's some delay between someone adding a benchmark and it getting onto speed.pypy.org17:38
K-osdelay > 1 year?17:38
exarkunI'm not familiar with the bm_regex benchmarks.  Where are they?17:39
K-osjust on the side: i don't want to flame around, i'm just curious. and i love the job, you're doing with pypy :-)17:40
K-osbenchmarks / unladen_swallow / performance 17:40
exarkunAh.  I don't know why those aren't up.17:41
K-os:-)17:41
nedbat (~nedbat@python/psf/nedbat) joined #pypy.17:52
darksideplanetgu (~dirk@p549E7414.dip.t-dialin.net) joined #pypy.18:06
darksideplanetgu (~dirk@p549E7414.dip.t-dialin.net) left irc: Remote host closed the connection18:06
darksideplanetgu (~dirk@p549E7414.dip.t-dialin.net) joined #pypy.18:07
darksideplanetgu (~dirk@p549E7414.dip.t-dialin.net) left irc: Client Quit18:07
darksideplanetgu (~dirk@p549E7414.dip.t-dialin.net) joined #pypy.18:08
darksideplanetgu (~dirk@p549E7414.dip.t-dialin.net) left irc: Remote host closed the connection18:10
fijal (~fijal@197.108.35.172) joined #pypy.18:10
fijalK-os: b_regexp, calls, pickle etc. were not chosen because we thought they're uninteresting microbenchmarks18:12
fzzzy (~donovan@nat/mozilla/x-aubzsvdtmlvajwkj) left irc: Quit: fzzzy18:12
fijalI think pickle is interesting now, but it wasn't at the time18:12
kennethreitz (~kennethre@204.14.152.118) joined #pypy.18:12
K-osok. thanks.18:12
fzzzy (~donovan@nat/mozilla/x-xanvdhdbdnoftkse) joined #pypy.18:12
fijalto be precise18:13
fijalsome of them are there because they were in US repo and they were interesting18:13
fijalsome of them are there because they present an interesting python real-load18:13
K-osI don't know the regex benchmark. but i would have thought regex performance would also be interesting to many people18:13
fijalsome of them are there because pypy was performing poorly18:14
fijalyes18:14
fijalfor example html5lib is mostly based on regexes18:14
K-os:-)18:14
K-osah, ok18:14
fijalyeah, we have some regex workload there18:14
fijalthey're not supposed to represent anything18:15
fijaland as of now they're faster than average, but just because this is what we were optimizing on18:15
fijallook at the history of some18:15
kenaan12alex_gaynor default 111a419a40ed62 15/pypy/module/micronumpy/: whitespace and other small codecleanups18:16
fijalfor example regex compilation is IMO very uninteresting18:16
Alex_Gaynorfijal: FYI there's getting to be a bit of untested code in interp_numarray, particularly in error cases18:16
fijalAlex_Gaynor: I'm not 100% sure if some of it is coverage being confused18:16
fijalto be checked18:16
K-osyes, sure. compilation isn't very interesting but maybe matching.18:17
Alex_Gaynoruh, it looks mostly ok, so I think it's us not having enough tests18:17
jer (jtregunna@unaffiliated/jer) left #pypy.18:17
kennethreitz (~kennethre@204.14.152.118) left irc: Quit: Textual IRC Client: http://www.textualapp.com/18:17
fijalK-os: look at those benchmarks18:17
fijalthey're kinda silly18:17
fijaland have unrolled loops by hand18:17
nedbatfijal, Alex_Gaynor: if you need coverage.py consultation, I'm glad to help...18:17
fijalor other auto-generated code18:17
fijalnedbat: I guess it might be me blaming others for good measure :)18:18
K-osyes, i've just started to take a closer look at the benchmarks, because i'll be holding a small talk about pypy in our workingroup tomorrow :-)18:18
Alex_Gaynornedbat: we need someone to write more tests :)18:18
fijalbut thanks, I'll get back to you if this seems to be the case18:18
fijalK-os: so I guess this is it18:18
kennethreitz (~kennethre@204.14.152.118) joined #pypy.18:18
nedbatAlex_Gaynor: I've been doing that too.. but only for the sandbox18:19
Alex_Gaynornedbat: well, then I'll keep holding fijal responsible for this ;)18:19
fijalK-os: pypy is exceptionally bad at hand-unrolled loops18:20
fijalwe also decided it's not really an issue18:20
K-osmakes sense :-)18:20
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) left irc: Quit: Leaving18:21
K-osin my workload pypy performs 3 times faster than cpython. so i like pypy :-)18:21
gutworththis ncar internship description is not very informative18:25
gutworthfor example it doesn't tell when it is until you it "apply"18:25
gutworthnothing tells you what's in the way of renummeration..18:27
fijal (~fijal@197.108.35.172) left irc: Read error: Connection reset by peer18:28
fijal (~fijal@197.172.212.37) joined #pypy.18:29
fijalgutworth: ping18:29
gutworthpong18:29
fijalI don't actually know when they require you to apply18:30
fijalI can probably tell you a bit what it would be about but it's irrelevant18:30
gutworthah, ha "no benefits"18:30
fijalwhat are benefits?18:31
kenaan12alex_gaynor default 110ccb237a9de2 15/pypy/module/micronumpy/: Implement varargs for ndarray.reshape and add a test for a missing error case.18:31
Alex_Gaynorfijal: health insurance mostly18:31
fijaland where did you find it?18:31
Alex_Gaynorgutworth: that's not unusual for a temporary position though18:31
gutworthI know18:31
gutworthI'm more interested in pay actually18:31
fijalgutworth: ?18:32
gutworthI'm trying to understand if its unpaid or not18:32
fijalI would claim it's paid18:33
fijalwell, I know it is18:33
fijalhow much I have no idea18:34
fijalgutworth: I think most NCAR visitors are paid like 2k$/months18:34
fijalper month18:34
fijalbut I don't knwo for sure18:34
gutworthat any rate, I'm not eligible18:34
stakkars_ (~tismer@dslb-088-074-053-099.pools.arcor-ip.net) joined #pypy.18:35
fijalwhere did you find it?18:35
voidspace (~voidspace@python/psf/voidspace) left irc: Quit: Uhm... gotta go18:35
fijalcan you reply?18:35
gutworthhttps://www2.cisl.ucar.edu/siparcs/requirements18:36
gutworthI suppose I could mail the director18:37
Action: fijal mails him now18:37
fijaldeadline is Feb18:39
fijalgutworth: he asked me if I would be willing to help with mentorship18:39
gutworthso no rush18:39
gutworthah18:39
fijalso I guess there is some cooperation involved18:39
DasIch (~dasich@p3E990997.dip.t-dialin.net) left irc: Ping timeout: 244 seconds18:41
lahwrananyone know of a very efficient opengl wrapper that works with pypy?18:42
fijalAlex_Gaynor: https://bitbucket.org/pypy/benchmarks/src/2bd97aedb97e/unladen_swallow/performance/bm_regex_v8.py18:42
fijalit should die in pain18:43
lahwranpreferably something that does not wrap it beyond exposing function calls18:43
fijallahwran: no idea18:43
fijalpyopengl is bad?18:43
lahwranuses ctyles18:43
lahwranctypes*18:43
fijaland?18:43
fijalctypes is good18:43
lahwranwhich according to a blog post on the pypy site is slow18:43
fijalwell no, but it's mostly fast on pypy18:43
fijalyou were reading a very old post18:43
lahwranah18:43
fijalto be precise, some ctypes is fast on pypy18:43
fijalsome is slow18:44
kenaan12alex_gaynor default 11d62de66be4aa 15/pypy/module/micronumpy/: remove some dead code, and added missing tests18:44
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) left irc: Quit: etrepum18:44
lahwranhow precise :/18:44
lahwranwhich is slow? :p18:44
Alex_Gaynorthe struct API is slow, but calls are very fast18:45
fijalAlex_Gaynor: no :)18:45
fijalAlex_Gaynor: if you don't pass types it's slow18:45
Alex_Gaynoroh, that too18:45
fijalthe slowpath is *really slow* so you have to be careful to hit the fast path18:45
fijalI'm not very happy with that, but whatever18:45
fijallahwran: you can write fast ctypes in pypy18:46
fijalmost of the time18:46
lahwranI'm more interested in ensuring that the opengl I use is fast18:46
fijalI don't know18:46
exarkunPyOpenGL uses ctypes on CPython18:46
exarkunAnd ctypes on CPython isn't particularly fast18:46
fijalit's not like I can check whether it's even working, it comes with no tests18:46
exarkunBut you can still make rad OpenGL-using Python programs18:46
exarkunGetting good performance out of OpenGL is less about having a super awesome runtime/efficient C library bindings and more about doing the right OpenGL calls18:47
lahwranI suppose so18:47
exarkunI'm not even very good at OpenGL, but I have a PyOpenGL-based program that renders massive amounts of stuff while using about 3% of my CPU18:48
lahwranpypy or no?18:48
exarkunCPython18:48
amaury (~amaury_@74.125.122.49) joined #pypy.18:48
lahwranimpressive18:48
exarkunSo say PyPy is 10x slower for some reason - no big deal, my video card is still going to be the bottleneck.18:48
jterracere: 3d + pypy, i wonder if panda3d's new idea for parallelizing rendering operations (http://www.panda3d.org/blog/?p=206) could be modified to use pypy for the app-level processing18:51
jterracei think it heavily uses python <-> c interface, so not sure if it's possible, but it would be really awesome if it was18:51
exarkunjterrace: Data flow analysis is hard.18:52
exarkunI think it would be awesome if someone gave PyPy a million dollars to research that topic. :)18:52
jterracehah18:52
DasIch (~dasich@p4FFDFB48.dip.t-dialin.net) joined #pypy.18:53
ericflo (~ericflo@75.103.8.110) joined #pypy.18:54
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.18:57
kdas_ (~kdas@114.143.162.64) joined #pypy.19:04
kenaan12alex_gaynor numpy-dtype-refactor-complex 11e4002f73fb25 15/: merged default19:05
kenaan12alex_gaynor numpy-dtype-refactor-complex 11894a554a8053 15/pypy/module/micronumpy/: make things translate19:05
kdas_ (~kdas@114.143.162.64) left irc: Client Quit19:06
durin42 (~durin@adium/durin42) left irc: Ping timeout: 252 seconds19:08
durin42 (~durin@adium/durin42) joined #pypy.19:10
kdas_ (~kdas@114.143.162.64) joined #pypy.19:12
etrepum (~bob@accessnat4.mochimedia.net) joined #pypy.19:13
mattip (~chatzilla@bzq-79-179-96-122.red.bezeqint.net) joined #pypy.19:17
fijalexarkun: I think I would be good with 20k19:17
fijalbut that's irrelevant19:18
fijalbecause noone will give me that19:18
cpets (4f7588fa@gateway/web/freenode/ip.79.117.136.250) joined #pypy.19:18
mattiphello. I have done some cookie penance, and wrote a "is_contiguous" for BaseArray. Writing tests now, on a seperate branch.19:22
fijalAlex_Gaynor: can you run the mako benchmark on your laptop19:22
fijal?19:22
fijalmattip: cool, I didn't :)P19:22
mattipThere is an issue though.19:22
Alex_Gaynorfijal: command?19:22
mattip(Why is nothing simple?)19:22
fijalAlex_Gaynor: runner.py -b bm_mako -p pypy19:23
tlynn (~tlynn@212.69.41.1) joined #pypy.19:23
Alex_Gaynorbecause we're solving a hard problem.  Or we're solving an easy problem very badly.19:23
mattipIf an array is F order, or if it is a transpose, the natural order is reversed.19:23
fijallet's assume only C order can be contiguous for now19:23
mattipSo there is what I call, for lack of a better term, contiguous_lr and contiguous_rl.19:24
fijalbecause you can only copy them if order is the same19:24
Alex_Gaynorfijal: http://paste.pocoo.org/show/517210/19:24
fijalAlex_Gaynor: it's faster on tannit, slower on my machine19:24
fijalavg on mine is 1.0, tannit 2.019:24
fijalwtf?19:24
fijalyou have an 8 core laptop?19:24
fijalor 4 * 2?19:25
Alex_Gaynor4 real cores, 8 if you count the fake ones19:25
fijalbastard19:25
mattipWell, anyway, I'll push the branch after testing, feel free to throw away lr rl cruft.19:26
fijalmattip: how about it returns the right thing depending on order?19:27
fijalbut yeah, commit19:27
bgola (~bgola@c9518082.virtua.com.br) joined #pypy.19:31
antocuni (~antocuni@host129-122-dynamic.16-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds19:35
kenaan12ctismer win64-stage1 11ccaf375ee24b 15/pypy/translator/c/src/: more places where signed_defn.h is missing. Made it easier to add it more often19:39
kdas_ (~kdas@114.143.162.64) left irc: Quit: This computer has gone to sleep19:40
apollo13 (~apollo13@unaffiliated/apollo13) left irc: Quit: SIGILL19:45
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!19:47
JaRoel|4d (~jaroel|4d@2001:980:7133:1:317d:e87f:5636:fdbe) joined #pypy.19:52
kenaan12mattip matrixmath-dot 11c1c81fb495b9 15/: hg merge with default19:53
kenaan12mattip numpypy-is_contiguous 119166c3425257 15/pypy/module/micronumpy/interp_numarray.py: refactor calc_steps19:53
kenaan12mattip numpypy-is_contiguous 11b8c87c4cf664 15/pypy/module/micronumpy/: test, implement is_contiguous for C order19:54
fijal (~fijal@197.172.212.37) left irc: Ping timeout: 252 seconds19:57
harrisonhttp://savethevoxels.blogspot.com/20:04
exarkunharrison: What about it?20:05
rguillebertwhat's a voxel ?20:06
harrisonlahwran: do you have a link to more about your pyopengl stuff?20:06
exarkunrguillebert: https://en.wikipedia.org/wiki/Voxel :)20:07
amauryVolume + Pixel20:07
rguillebertI see20:09
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 252 seconds20:15
kennethreitz (~kennethre@204.14.152.118) left irc: Ping timeout: 252 seconds20:16
kenaan12ctismer win64-stage1 1141a380cec3d1 15/pypy/translator/c/src/signals.h: typo :-(20:17
mvt (~mvantelli@53530442.cm-6-4a.dynamic.ziggo.nl) joined #pypy.20:18
arigato (~arigo@89.204.137.112) joined #pypy.20:20
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) joined #pypy.20:21
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) left irc: Changing host20:21
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.20:21
harrisonour voxels are a bit different20:23
harrisonmore advanced20:23
harrisonrecursivey20:23
kdas_ (~kdas@114.143.162.64) joined #pypy.20:25
nedbat (~nedbat@python/psf/nedbat) left irc: Disconnected by services20:31
nedbat_ (~nedbat@python/psf/nedbat) joined #pypy.20:31
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.20:32
harrisoni meant exarkun20:33
stakkars_arigato: I'm seeing a problem, suddenly, because I probably created a prebuilt constant of unsigned -1.20:35
asmeurer___ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy.20:35
stakkars_is that really so: I cannot have a prebuilt const of 2^^32-1 ?20:36
jnoller (~jnoller@vpn.nasuni.com) left irc: Quit: Leaving...20:37
stakkars_when exactly are PBCs created? I'm getting that when testing rmmap20:37
arigatothere are some issues, but it depends on what exactly is written in the RPython code20:38
arigatoindeed, a prebuilt constant that is a long, greater than sys.maxint, is not supported20:39
arigatobut you can have a prebuilt constant that is a r_uint(-1)20:39
arigatoor r_uint(2**32-1)20:39
kenaan12alex_gaynor default 114d94de61c725 15/pypy/module/micronumpy/interp_numarray.py: fix translation20:39
arigatoas long as making the r_uint is done outside the RPython code20:39
stakkars_aeh& if I use a constant r_uint(-1) in the module code, i.e. before RPython digs in, isn't then exactly it becoming a PBC?20:42
arigatoyes20:43
arigatothis is not supported?20:43
stakkars_or should the global module code contain -1, and when I use it, cast it dynamically in the RPython code20:44
lizardo (~lizardo@189.2.128.130) left irc: Quit: Leaving20:44
stakkars_I'm not sure when it happens, actually, just a random error in the annotator20:44
arigatono, it works for me20:45
stakkars_so if I make r_uint(-1) in the RPy code, _then_ I get that problem?20:46
int3__ (~int3__@st401-246.subnet-244.amherst.edu) joined #pypy.20:46
stakkars_then I have a different perception of 'pre'20:46
stakkars_pre to what?20:47
stakkars_ah, I use a constant -1 in the code, and a r_uint() of this -1 in the code, and that becomes a PBC?20:48
arigatosorry, I can't help you without seeing the code20:48
stakkars_working on trunk, something I did, and testing rmmap20:50
int3__ (~int3__@st401-246.subnet-244.amherst.edu) left irc: Remote host closed the connection20:50
kkris (~kris@80-123-36-77.adsl.highway.telekom.at) left irc: Quit: Leaving.20:53
kdas_ (~kdas@114.143.162.64) left irc: Quit: This computer has gone to sleep20:58
etrepum_ (~bob@accessnat4.mochimedia.net) joined #pypy.21:00
etrepum_ (~bob@accessnat4.mochimedia.net) left irc: Remote host closed the connection21:02
etrepum_ (~bob@accessnat4.mochimedia.net) joined #pypy.21:02
etrepum (~bob@accessnat4.mochimedia.net) left irc: Read error: No route to host21:04
Nick change: etrepum_ -> etrepum21:04
mattipI started a blog entry "A dummy's guide to hacking on pypy" here http://paste.pocoo.org/show/517268 21:07
mattipdoes this look worthy of working on and posting?21:08
rguillebertmattip, you might want to use flake821:09
rguillebertit combines pep8 pyflakes and something else21:10
apollo13 (~apollo13@unaffiliated/apollo13) joined #pypy.21:10
kdas_ (~kdas@114.143.162.64) joined #pypy.21:12
mattiprguilllebert: thanks. But is this interesting enough for a blog entry, or has it already been hashed over elsewhere and I missed it?21:12
rguillebertI don't know but you can post it on your blog and put it on reddit I guess :)21:14
rguillebertYou don't need to have anyone's approval at least21:16
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Read error: Operation timed out21:16
mattip;)21:17
ojii (~ojii@40-34.60-188.cust.bluewin.ch) left irc: Remote host closed the connection21:17
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) joined #pypy.21:17
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) left irc: Changing host21:17
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.21:17
rguillebertbut if you want to post it on the official blog I'm not the one you should ask to :)21:17
stakkars_arigato: if I have the global21:19
Nick change: nedbat_ -> nedbat21:20
stakkars_INVALID_SOCKET = r_uint(INVALIG_SOCKED)21:20
stakkars_does that generate a bad PBC?21:20
stakkars_IOW, do I create that too early?21:21
mat^2 (~mathias@212.130.113.35) joined #pypy.21:21
tlynn (~tlynn@212.69.41.1) left irc: Ping timeout: 240 seconds21:23
fzzzy (~donovan@nat/mozilla/x-xanvdhdbdnoftkse) left irc: Quit: fzzzy21:24
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds21:26
arigatostakkars_: no, that's fine21:26
kdas_ (~kdas@114.143.162.64) left irc: Quit: This computer has gone to sleep21:31
cpets (4f7588fa@gateway/web/freenode/ip.79.117.136.250) left irc: Quit: Page closed21:31
mattip (chatzilla@bzq-79-179-96-122.red.bezeqint.net) left #pypy.21:32
Shanita (~John@osbk-4db15ac2.pool.mediaWays.net) left irc: Ping timeout: 240 seconds21:35
kenaan12ctismer win64-stage1 112e5389e1972f 15/pypy/module/micronumpy/: merge default21:35
kenaan12ctismer win64-stage1 11e3ea42ebc83e 15/pypy/translator/c/src/signals.h: merge21:36
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 260 seconds21:39
fprimex (~fprimex@brent-macbook.sc.fsu.edu) left irc: Quit: http://www.fprimex.com21:40
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) left irc: Quit: Leaving...21:41
Shanita (~John@osbk-4db15ac2.pool.mediaWays.net) joined #pypy.21:42
stakkars_arigato: ah, I just see that it wasn't me. It happens on default, not my branch. Interesting, will find out :-)21:42
ousado_ (~ousado@hmbg-5f76557a.pool.mediaWays.net) left irc: Ping timeout: 244 seconds21:42
arigato:-/21:42
kenaan12arigo default 11245a08e9eb96 15/pypy/interpreter/baseobjspace.py: "Clarify."  Quotes intended.21:43
Alex_Gaynorarigato: this version will be slower!21:43
arigatoin fact, I think that it will be faster21:44
arigato:-)21:44
Alex_Gaynorshouldn't itertools.chain be very nicely JIT'd?21:44
kennethreitz (~kennethre@204.14.152.118) joined #pypy.21:44
arigatoitertools in general is JITted, but not very nicely21:44
Alex_Gaynorchain will be nicely JIT'd though I think21:45
Action: arigato doesn't find any "jit" in interp_itertools21:45
arigatobut possibly21:46
Alex_Gaynorthe common case for chain.next has no loop21:46
Alex_Gaynorit simply goes forward on teh current iterator21:46
Alex_GaynorI suspect it really doesn't matter for this code though :)21:46
arigato:-)21:46
arigatoyes, indeed21:47
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.21:47
ronnyisnt itertools.chain and itertools.chain.from_iterable bascally the same loop expressable as  return x for x in items for items in args21:49
ronny(unless i got the genexpr for order wrong again))21:49
Alex_Gaynorronny: I think you do have them in the wrong order, but yes basically21:49
ronnyhmm21:50
ronnydammit21:50
arigatoin other words it's equivalent to the incredible code:21:50
arigatofor items in args:21:50
arigato  for x in items:21:50
arigato     yield x21:50
arigatowhere you cannot get the order wrong21:50
ronnyheh21:51
kenaan12arigo default 119873bcdab29c 15/pypy/module/itertools/interp_itertools.py: Hum, I'll revert this if I'm wrong, but I don't see how w_iterables can ever be None.21:51
ronny#pypy has a amazing concentration of people that are more right than me by default21:51
kenaan12arigo default 11bb527bdf9535 15/pypy/interpreter/astcompiler/optimize.py: Kill! Kill!  Arrr arr arr.21:58
Rhy0lite (dje@nat/ibm/x-ymiaplfkvqbopvxt) left irc: Quit: Leaving21:58
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.22:02
Alex_Gaynorarigato: why the itertools hate?22:03
arigatoif you want a serious answer: I don't particularly hate it, it's just that in 245a08e9eb96 I hit the line for unrelated reasons22:04
arigatoand it was still failing (of course) once I removed itertools22:04
arigatobut then I thought, well, no reason to have itertools there22:04
arigatoand then bb527bdf9535 was due to me fooling around with a "grep 'import itertools'" and finding this one and unique other usage,22:05
arigatowhere the non-itertools version is much shorter22:05
arigatoin general, that's what I don't like of itertools: it makes program rather unreadable if you're not used to it22:08
CIA-1303nirbheek 07roundup * 10#747/fcntl.lockf() is broken with (fcntl.LOCK_EX | fcntl.LOCK_NB): 22:11
CIA-13[resolved] I can verify that with 1.7 this test no longer fails and the issue has been22:11
CIA-13resolved, thanks!22:11
CIA-13(I hope I closed the bug properly) * 14https://bugs.pypy.org/issue74722:11
Circlefusion (~cf@74-138-199-109.dhcp.insightbb.com) left irc: Ping timeout: 252 seconds22:12
ousado_ (~ousado@31-18-121-45-dynip.superkabel.de) joined #pypy.22:14
cheater (~cheater@178-26-42-68-dynip.superkabel.de) left irc: Quit: leaving22:15
stakkars_ (~tismer@dslb-088-074-053-099.pools.arcor-ip.net) left irc: Quit: schnarch22:17
mvt (~mvantelli@53530442.cm-6-4a.dynamic.ziggo.nl) left irc: Quit: This computer has gone to sleep22:18
Circlefusion (~cf@74-138-199-109.dhcp.insightbb.com) joined #pypy.22:18
Nick change: ousado_ -> ousado22:19
ousado (~ousado@31-18-121-45-dynip.superkabel.de) left irc: Changing host22:19
ousado (~ousado@unaffiliated/ousado) joined #pypy.22:19
amaury (~amaury_@74.125.122.49) left irc: Ping timeout: 244 seconds22:23
kenaan12alex_gaynor default 11648537e5b016 15/pypy/module/micronumpy/interp_dtype.py: typo fix22:27
kenaan12alex_gaynor default 1133ec28c6d811 15/pypy/: merged upstream22:27
CIA-1303nirbheek 07roundup * 10#747/fcntl.lockf() is broken with (fcntl.LOCK_EX | fcntl.LOCK_NB): 22:27
CIA-13[chatting] Err, I checked the wrong test here. This test case still fails with 1.7,22:27
CIA-13reopening... * 14https://bugs.pypy.org/issue74722:27
kenaan12amauryfa py3k 11f65121a60b37 15/pypy/rlib/rsocket.py: Return bytes for Linux abstract namespace sockets22:29
kenaan12amauryfa py3k 11fc9f0d9b77db 15/pypy/module/_socket/: socket.inet_aton() should return bytes22:29
kenaan12amauryfa py3k 11df2edb52b29e 15/pypy/module/_socket/: socket.getaddrinfo() accept strings for the 'port' parameter22:29
kenaan12amauryfa py3k 11c0288af60001 15/pypy/: Fixes in _ssl module22:29
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.22:29
voidspace (~voidspace@python/psf/voidspace) joined #pypy.22:30
stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) joined #pypy.22:32
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 248 seconds22:34
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/73622:37
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/16222:37
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/96622:37
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds22:50
_pyre_1 (~stian@137.241.189.109.customer.cdi.no) joined #pypy.22:51
_pyre_1calls from struct.Struct() class is super slow22:51
_pyre_1in pypy22:51
_pyre_1while their alot faster than making calls to struct.unpack in CPython22:52
Alex_Gaynorcan you write a script to demonstrate the slowdown, I thought I'd cleaned up the performance problem with the struct module22:52
_pyre_1I'll post it, one moment22:53
Alex_Gaynorthanks22:53
stakkars_ (~tismer@p5DDB7E68.dip.t-dialin.net) left irc: Quit: schnarch22:53
whitelynx|work (~whitelynx@63.241.75.144) left irc: Quit: Ex-Chat22:55
antocuni (~antocuni@host129-122-dynamic.16-79-r.retail.telecomitalia.it) joined #pypy.22:55
_pyre_1Here, it's a messy script, but I incuded some generator code as well for the format I try to read: http://pastebin.com/brfj0Aq922:57
_pyre_1Using struct.Struct: 2.37065622:57
_pyre_1Using struct.unpack: 1.05901422:57
Alex_Gaynorwhat are the times under CPython for each?22:58
voidspace_ (~voidspace@87-194-212-65.bethere.co.uk) joined #pypy.22:58
voidspace_ (~voidspace@87-194-212-65.bethere.co.uk) left irc: Changing host22:58
voidspace_ (~voidspace@python/psf/voidspace) joined #pypy.22:58
_pyre_1CPython use about a second less using struct.Struct than struct.unpack, while it's still 7-8 times slower than pypy using struct.unpack22:58
_pyre_17.391604 with struct.Struct22:58
_pyre_18.507426 with struct.unpack22:58
Alex_Gaynorok so it's werid, but we're not actually slower or anything, good :)  I'll still investigate22:59
_pyre_1pretty stable results, only varry with 0.02s22:59
_pyre_1i little more on pypy tho, with 0.0423:00
Alex_GaynorHAH, I see why, this is silly, hrm23:00
voidspace (~voidspace@python/psf/voidspace) left irc: Ping timeout: 240 seconds23:00
Nick change: voidspace_ -> voidspace23:00
Alex_Gaynorthis isn't going to be easy to fix :/23:00
arigato? :-)23:01
Alex_Gaynorarigato: so struct.unpack is very fast because the JIT checks if the format str is constant and if it is unrolsl the loops23:01
arigatoah, bah23:01
Alex_Gaynorarigato: however if you have x = Struct("format str"); for i n xrange(100): x.unpack(i) or so, the string isn't constant23:01
Alex_Gaynorit'd be ok if Struct was a global or something23:01
Alex_GaynorI would if Struct instances should just promote their format and be done with it23:02
arigatoand even not, because x.format is not a constant attribute23:02
Alex_Gaynoris it constant on CPython?23:02
arigatoI suppose, but right now it's written at app-level23:02
arigatoin pypy23:02
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 240 seconds23:02
Alex_Gaynorarigato: yes, I guess it might make sense to move it to interplevel so we can make it const and do the promotion23:03
Alex_GaynorI think promote makes sense here23:03
arigatoyes and no, as usual23:03
Alex_Gaynorindeed23:03
arigatoe.g. what if someone does x = Struct("i" * n)23:03
arigatowell I suppose that he's punished and it's his fault, or something23:04
Alex_Gaynorindeed, as always the mess of heuristics :/23:04
arigatoyes23:04
exarkunI did that this afternoon.23:04
arigatoah no, it's potentially a bad idea to promote a string, anyway23:05
JaRoel|4d (~jaroel|4d@2001:980:7133:1:317d:e87f:5636:fdbe) left irc: Remote host closed the connection23:05
arigatoI think it's promoting the pointer, not the actual string content23:05
Alex_Gaynorarigato: err, yes, we have string_promote nowadays though23:05
arigatoah, ok23:06
DanielHolth_ (~dholth@ip98-180-34-112.ga.at.cox.net) left irc: Quit: Ex-Chat23:06
Alex_Gaynor_pyre_1: I'm not going to fix this ATM, but I'm going to make a ticket about this, thanks23:06
stakkars (~tismer@p5DDB7E68.dip.t-dialin.net) left irc: Quit: stakkars23:06
_pyre_1Ok :)23:06
squiddy (~squiddy@f053081179.adsl.alicedsl.de) left irc: Quit: Leaving23:07
ronnywouldnt it make sense to have a __pypy__.constant descriptor one can use to declare attributes23:07
_pyre_1CPython:23:08
_pyre_1struct.Struct('').format = "H"23:08
_pyre_1AttributeError: attribute 'format' of 'Struct' objects is not writable23:08
_pyre_1on pypy it clears23:08
CIA-1303agaynor 07roundup * 10#955/move struct.Struct to interplevel: 23:09
CIA-13[new] There are a few reasons and issues to be fixed:23:09
CIA-131) format should be immutable on the Struct instance23:09
CIA-132) format should be promoted so tha ... * 14https://bugs.pypy.org/issue95523:09
stakkars (~tismer@p5DDB7E68.dip.t-dialin.net) joined #pypy.23:09
arigato (~arigo@89.204.137.112) left irc: Quit: See you23:11
dracman (~draco@212.255.26.64) left irc: Ping timeout: 252 seconds23:21
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Read error: Connection reset by peer23:31
aurojit (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.23:32
aurojit (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Read error: Connection reset by peer23:32
etrepum_ (~bob@accessnat4.mochimedia.net) joined #pypy.23:33
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.23:33
apanda (~apanda@dhcp-44-187.EECS.Berkeley.EDU) left irc: Read error: Connection reset by peer23:34
aurojit (~apanda@dhcp-44-187.EECS.Berkeley.EDU) joined #pypy.23:34
etrepum_ (~bob@accessnat4.mochimedia.net) left irc: Read error: No route to host23:34
etrepum__ (~bob@accessnat4.mochimedia.net) joined #pypy.23:34
etrepum (~bob@accessnat4.mochimedia.net) left irc: Ping timeout: 240 seconds23:36
Nick change: etrepum__ -> etrepum23:36
kennethreitz (~kennethre@204.14.152.118) left irc: Read error: Connection reset by peer23:43
--- Wed Dec 7 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!