#pypy IRC log for Thursday, 2011-12-15

PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood00:19
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.00:20
whyking (~quassel@128.148.231.16) left irc: Ping timeout: 244 seconds00:22
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) joined #pypy.00:25
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) left irc: Changing host00:25
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.00:25
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace00:29
kenaan12jterrace numpy-full-fromstring 11d5854741cdef 15/pypy/module/micronumpy/: Fix most review comments, two have responses in the notes sections of REVIEW00:29
whitelynx (~whitelynx@63.241.75.144) left irc: Quit: Ex-Chat00:30
voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.00:33
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host00:33
voidspace (~anonymous@python/psf/voidspace) joined #pypy.00:33
jaredw (~jared@122-59-91-107.jetstream.xtra.co.nz) joined #pypy.00:37
jterraceAlex_Gaynor: numpy-full-fromstring and numpy-concatenate should be good to go now (made some notes in fromstring REVIEW)00:37
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) joined #pypy.00:39
jaredw (~jared@122-59-91-107.jetstream.xtra.co.nz) left irc: Remote host closed the connection00:40
maxyz (~maxy@186.18.94.35) left irc: Ping timeout: 245 seconds00:46
jterrace (~jterrace@aegis.CS.Princeton.EDU) left irc: Quit: Leaving.00:46
maxyz (~maxy@186.18.94.35) joined #pypy.00:48
lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) left irc: Ping timeout: 240 seconds00:51
fprimex (~fprimex@brent-macbook.sc.fsu.edu) left irc: Quit: http://www.fprimex.com00:59
lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) joined #pypy.01:03
kenaan12ned nedbat-sandbox-2 1147a28aeffae1 15/pypy/translator/sandbox/: Make sandlib logging conditional, and only import pypy.tool.ansi_print if we are logging.01:03
Nick change: lesshaste -> Guest6671701:04
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/17001:05
rekamso (~textual@63-228-127-70.dia.static.qwest.net) left irc: Read error: Operation timed out01:09
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace01:12
mtigas (~mtigas@users.spokesman.com) left irc: Quit: mtigas01:14
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/74501:19
ericflo (~ericflo@75.103.8.110) left irc: Quit: ericflo01:25
etrepum (~bob@accessnat4.mochimedia.net) left irc: Ping timeout: 240 seconds01:29
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/188801:35
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/63201:35
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/64301:35
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/30501:35
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/116001:35
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/82701:35
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/146901:35
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/75001:35
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/29101:35
bbot213Exception: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/29101:35
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection01:41
jterrace (~Adium@dynamic-oit-vapornet-b-1405.Princeton.EDU) joined #pypy.01:49
jterrace (~Adium@dynamic-oit-vapornet-b-1405.Princeton.EDU) left irc: Client Quit01:52
papercrane (~papercran@75.101.111.82) left irc: Ping timeout: 248 seconds01:57
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) joined #pypy.02:00
MrSmile (~tamer@g227036073.adsl.alicedsl.de) joined #pypy.02:02
MrSmilehi people!02:02
MrSmileI am interisted to port pypy on 64 bit windows. 02:02
MrSmileIs it meanwhile possible, or are there still problems?!02:03
MrSmileand the other question, IF NOT SUPPORTED... is it because of the Visual C++ compiler or more or less because the include headers and libraries?!02:06
lvh (~lvh@python/psf/lvh) left irc: Ping timeout: 255 seconds02:09
lvh (~lvh@46.21.169.62) joined #pypy.02:10
Nick change: lvh -> Guest911202:10
MrSmile (tamer@g227036073.adsl.alicedsl.de) left #pypy.02:14
canta (~canta@77-20-123-240-dynip.superkabel.de) joined #pypy.02:16
tilgovi (~randall@173-13-135-107-sfba.hfc.comcastbusiness.net) joined #pypy.02:24
tilgovi (~randall@173-13-135-107-sfba.hfc.comcastbusiness.net) left irc: Changing host02:24
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.02:24
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!02:35
harrison (~quassel@adsl-69-209-218-183.dsl.chcgil.sbcglobal.net) left irc: Read error: Connection reset by peer02:39
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/97802:47
jae (~jae@dedicated.jaerhard.com) left irc: Ping timeout: 240 seconds02:49
xiaochen (xiaochen@nat/redhat/x-ldxhjbwpbsmopnbk) joined #pypy.02:50
canta (~canta@77-20-123-240-dynip.superkabel.de) left irc: Quit: Leaving.02:52
dracman (~draco@212.255.46.227) left irc: Ping timeout: 252 seconds02:54
kenaan12ned nedbat-sandbox-2 11585b46142c3e 15/pypy/translator/sandbox/sandlib.py: Move another import so we don't require 'import py'03:34
nedbatAnyone around that can help me interpret this?  http://paste.pocoo.org/show/521108/03:36
gutworthpassing a non-integer string to int()?03:39
nedbatgutworth: does your question mark mean you aren't sure, or that it's obvious?03:46
jae (~jae@dedicated.jaerhard.com) joined #pypy.03:47
Ademan (~dan@adsl-71-141-224-79.dsl.snfc21.pacbell.net) joined #pypy.03:48
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/188803:49
mat^2 (~mathias@212.130.113.35) left irc: 03:51
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 248 seconds03:55
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) left irc: Ping timeout: 248 seconds04:01
anish_ (anish@nat/redhat/x-qmytbvhzybmknyrw) joined #pypy.04:16
mwhudson (~mwh@120.136.5.22) left irc: Ping timeout: 240 seconds04:26
ltrottier (~Adium@128.54.254.180) joined #pypy.04:47
ltrottierhey all04:47
ltrottierso quick question: what is the right approach for getting macports to use pypy to compile (future versions of pypy)04:48
ltrottieralternatively04:48
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/146904:48
ltrottier& does it make sense to download the pypy binary, and then use it to speed up macports' build of pypy?04:48
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/82704:49
papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) joined #pypy.04:53
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/64304:54
canta (~canta@77-20-123-240-dynip.superkabel.de) joined #pypy.05:00
gutworth (~benjamin@dhcp-128-36-84-91.central.yale.edu) left irc: Ping timeout: 255 seconds05:08
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 252 seconds05:14
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/116005:17
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/63205:23
xiaochen (xiaochen@nat/redhat/x-ldxhjbwpbsmopnbk) left irc: Quit: leaving05:28
ltrottier&. nvm05:29
xiaochen (xiaochen@nat/redhat/x-prefcvbgtglmnoyx) joined #pypy.05:31
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Computer has gone to sleep.05:44
squiddy (~squiddy@f053084237.adsl.alicedsl.de) joined #pypy.05:45
xiaochen (xiaochen@nat/redhat/x-prefcvbgtglmnoyx) left #pypy.05:48
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.05:53
aleksi (~aleksi@85.235.191.82) joined #pypy.05:59
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) joined #pypy.06:07
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) left irc: Changing host06:07
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.06:07
Nick change: Gulopine -> Gulaway06:16
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/75006:21
ltrottier (Adium@128.54.254.180) left #pypy.06:21
fijal (~fijal@41.53.118.135) joined #pypy.06:32
fijalhi06:36
papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep.06:37
fijalfzzzy: ping?06:42
fzzzyfijal: pong06:42
fijalfzzzy: have few minutes? see PM06:42
Alex_Gaynorhey fijal 06:58
fijalhey06:58
mvt (~mvantelli@87.213.45.85) joined #pypy.07:04
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 240 seconds07:06
fijalAlex_Gaynor: the refactor-signature branch is going somewhere07:16
fijalI'll fix views now07:16
Alex_Gaynorfijal: great07:16
Alex_Gaynorfijal: do you mean make them handle non-concrete parents, or something else?07:17
fijalit'll still takes a few days I think07:17
fijalyes07:17
fijalit fits better in the picture07:17
Alex_Gaynoreh, yes which?07:17
fijalnon-concrete parents07:17
Alex_Gaynorcool, let me know if you need review/rubber duck or anything07:17
fijalno, not for now07:17
Alex_Gaynork07:18
fijalit's just that I disabled broadcast & flatiter07:18
fijaland have to put them back at some point07:18
Alex_Gaynorput them back eventually please :)07:18
kenaan12fijal refactor-signature 119400300d69d3 15/pypy/module/micronumpy/signature.py: make viewsignatures always differ07:28
fijalhum, no that was not what I meant07:30
kenaan12alex_gaynor numpy-full-fromstring 11d0c17d270f6c 15/pypy/: merged default07:31
kenaan12alex_gaynor numpy-full-fromstring 1169e0e93f244d 15/pypy/module/micronumpy/REVIEW: remaining review notes, almost there07:31
Alex_Gaynorfijal: are you doing array sharing, or just iterator sharing for now?07:31
fijaliterator sharing07:31
fijalI'm trying to cheat and failing07:31
Alex_Gaynork, we can do array sharing on top of it later07:31
fijalyes07:33
fijalthat would be a different numbering07:33
Alex_Gaynoryes, essentially two sets of numbering, something for later07:34
Alex_Gaynorin theory you can have array sharing that doesn't share iterators, but also iterator sharing that doesn't share arrays07:34
Alex_Gaynore.g. numpy.arange(10) + numpy.ones(10) can share iterators07:34
fijalyes07:34
fijalok, good07:35
fijalit passes all the tests without zjit07:35
Alex_Gaynordid you unskip all the skips you added?07:35
nedbat (~nedbat@python/psf/nedbat) joined #pypy.07:35
fijalnope07:35
fijalbut that's beyond the point07:35
fijalI'll make sure test_zjit passes07:36
fijalso I actually did something07:36
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy07:36
fijalthe rest is just boring engineering07:36
Alex_Gaynorzjit should be better than passing07:36
kenaan12fijal refactor-signature 117cbc688e4439 15/pypy/module/micronumpy/signature.py: this is what I meant07:36
Alex_Gaynorit should be failing because it is so good :)07:36
fijalys07:36
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.07:40
Guest66717 (~lesshaste@87-194-206-189.bethere.co.uk) left irc: Quit: Leaving07:46
kenaan12hakanardo extradoc 11f37efa0faa47 15/talk/iwtc11/benchmarks/convolution/convolution.py: numpy version07:50
fijalhakanardo: how is it doing?07:52
hakanardo_http://paste.pocoo.org/show/521145/07:52
hakanardo_did I make some mistake, I was expecting them to get fairly close?07:53
fijallook at traces07:54
fijalare you using multidim arrays?07:54
hakanardo_traces for simpler loops ar looking good with numpy07:55
Alex_Gaynorcan you paste a trace?07:55
hakanardo_there is an applevel Array2D class that implementes 2 dimensional arrays ontop of array.array07:55
fijalthen why is it slow :)07:55
Alex_Gaynoreh, why should indexing be slower07:56
hakanardo_Alex_Gaynor: it shouldn't :)07:56
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/30507:56
nedbat (~nedbat@python/psf/nedbat) left irc: Quit: Leaving07:57
fijalhakanardo_: then there is a bug07:57
Alex_Gaynortimes are much closer for me but I still see it07:59
fijalAlex_Gaynor: enumerate is RPython, but it fails to do some casts, so llinterp does not like it07:59
Alex_Gaynorfijal: so did we merge some un-rpython code?08:00
fijalno, it's all well and good08:02
fijalbut it sometimes explodes in llinterp08:02
Alex_Gaynorfijal: so jitviewer is exploding ATM for me08:03
Alex_Gaynorhttp://paste.pocoo.org/show/521147/08:03
fijalnice08:03
Alex_Gaynorlog produced with: PYPYLOG=jit-log-opt,jit-backend:x.pypylog PYTHONPATH=~/projects/pypy/ /tmp/pypy-c-jit-50524-5a9a29b9c0ae-linux64/bin/pypy `pwd`/runner.py convolution/convolution.py conv3x3_numpy 1000 100008:03
fijalit can't find some code08:04
fijalbut it finds the source08:04
fijaljust not quite what it wants08:04
fijalsomething like applevel or so08:04
fijalyou can just disable seeing source in the code, or just have an except: that puts FuncNoCode in that place08:05
Alex_Gaynorfijal: could it be a bug with the new label syntax things?08:05
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.08:06
fijalno, it's probably not08:08
Alex_Gaynorwell, I'm not particularly in the mood to fight with the jitviewer tonight08:09
fijalAlex_Gaynor: do you want me to tell you exactly what to change?08:09
Alex_Gaynorperhaps some other time, not tonight though08:09
kenaan12fijal refactor-signature 1178db9015b5f0 15/pypy/module/micronumpy/: RPythonization08:09
kenaan12fijal refactor-signature 11876091f5df5a 15/pypy/module/micronumpy/: rpythonization, remove one int_add08:09
fijalAlex_Gaynor: hg pull -u?08:11
Alex_Gaynor?08:11
kenaan12fijal jitviewer 11ca0dcdbd8036 15/bin/jitviewer.py: insist a bit more on displayng loops08:11
Action: fijal figths with llinterp08:12
fijalnow explodes with TypeError :/08:12
fijalthe traceback makes no sense08:15
fijalhttp://paste.pocoo.org/show/521154/08:15
fijalAlex_Gaynor: feel like debugging?08:15
Alex_GaynorATM, not especially, did you try calling the funciton directly08:15
Alex_Gaynorand see if the python version raises08:16
fijalyes08:17
fijaland no it does not08:17
fijalthere is an identical test in test_compile08:17
Alex_Gaynorno idea then, it's too late for me08:17
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds08:17
fijaloh noes, I'm a moron08:18
fijaleh, am I?08:18
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.08:18
hakanardo_optimizeop/heap does not seam to care about interiorfield's?08:19
hakanardo_so a setinteriorfield_raw will clear the heap cache?08:19
Alex_Gaynorhakanardo_: heh, that's no good, shouldn't happen, you can't mutate GC memory with _raw operations08:20
hakanardo_so it should probable be added to the list of ops in OptHeap.emitting_operation08:21
Alex_Gaynoryes08:21
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy.08:26
xiaochen (xiaochen@nat/redhat/x-uywqmnhytazzzmpv) joined #pypy.08:26
kenaan12fijal refactor-signature 11c0dff290b339 15/pypy/module/micronumpy/test/test_compile.py: an extra test just because08:28
kenaan12fijal refactor-signature 11cd58a567d17d 15/pypy/module/micronumpy/test/test_zjit.py: remove some int_add's08:28
fijalhakanardo_: ping?08:43
fijalhakanardo_: can I request a JIT optimization?08:44
fijal;-)08:44
Alex_Gaynorfijal: whats the request?08:45
fijalI'm not sure wait ;)08:45
fijalI'm trying to make bridges in max/min better08:46
Alex_Gaynorone thing at a time :)08:47
fijalyes, sure08:47
fijalbut it's easier to do things now that I disabled some stuff08:47
fijalthan later08:47
Alex_Gaynoreasiest thing would be to figure out the right string of ASM for MAXSD/MINSD with NaNs :)08:47
fijalAlex_Gaynor: it's just that things fall nicely in places now :)08:49
fijallike you just have to fill in slots08:49
Alex_Gaynor?08:49
fijalthe shape of the code is just right now08:50
fijalit does not feel awkward to specialize08:50
fijalif you want to specialize on X, X must be a different signature08:50
fijalfor example a forced thing would just have an ArraySignature08:51
fijalbecause it makes tons of sense08:51
Alex_Gaynoryes sure, my point is can't bridge disucssion wait until later?08:52
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Remote host closed the connection08:52
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) joined #pypy.08:53
fijalsure08:54
fijalI'm off it right now08:54
Alex_Gaynorok ust wanted to make sure I understand08:54
kenaan12fijal refactor-signature 11dfe668607a47 15/pypy/module/micronumpy/: Improve a bit on max() bridge08:54
fijalAlex_Gaynor: ^^^ check out how the bridge looks like now08:55
Alex_Gaynordoes this require pygame08:55
fijalyes08:55
fijal:)08:55
Alex_Gaynornot gonig to happen, pygame still can't be installed by pip08:55
fijalwould be cool to be able to use webgl for that08:56
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.08:58
fijalAlex_Gaynor: I get only few unnecessary things for the bridge09:02
fijalbecause the rest is on signature (== constant)09:02
Alex_Gaynorcool09:03
hakanardo_fijal: sure :)09:07
fijalhakanardo_: I'm not sure I need it09:08
fijalit seems I'm doing just fine with storing more stuff on greenkey09:08
anish_ (anish@nat/redhat/x-qmytbvhzybmknyrw) left irc: Ping timeout: 252 seconds09:09
hakanardo_ok09:09
fijalhakanardo_: ah, I want to be able to check bridge09:10
fijalnot check_resops09:10
hakanardo_count the ops in one specific bridge?09:10
hakanardo_how would you specify which?09:11
fijal"all bridges that go out of simple_loop"09:11
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.09:11
fijalAlex_Gaynor: for example this is test_already_forced09:12
fijalhttp://paste.pocoo.org/show/521167/09:12
fijalI cut down on number of resops in preamble a bit09:13
fijal(that does mean stuff for bridges)09:13
hakanardo_fijal: ok, that shouldt be too hard to implemet I suppose09:13
Alex_Gaynorcan I get a sorted version of this, a bit hard to read09:13
fijalit's not ideal, but instead of having 22 guard_class we have 809:13
fijalhttp://paste.pocoo.org/show/521168/09:13
fijalthe diff09:13
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.09:14
kenaan12fijal refactor-signature 11c2da6bb0433f 15/pypy/module/micronumpy/: have ForcedSignature instead of ArraySignature (it's the same otherwise), cut down on number of ops09:17
fijalhm09:18
fijalthat's not very good, we don't share array and forced09:18
fijalmaybe ok though09:18
fijalbecause they have different layout09:18
anish_ (anish@nat/redhat/x-ychfmgmnjtxalzmb) joined #pypy.09:23
Guest9112 (~lvh@46.21.169.62) left irc: Quit: leaving09:24
mkerrin (~mkerrin@wg1-nat.fwgal01.deri.ie) joined #pypy.09:29
mkerrin (mkerrin@wg1-nat.fwgal01.deri.ie) left #pypy.09:29
arigato (~arigo@89.204.154.6) joined #pypy.09:32
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) left irc: Quit: asmeurer__09:37
fijalhi armin09:37
arigatohi09:38
fijalAlex_Gaynor: ok, feel like being a rubber duck?09:38
Alex_Gaynorfijal: sure09:39
kenaan12arigo counter-decay 11642e93456721 15/pypy/: Write the parametrisation of decaying counters, and a test. Missing the actual implementation.09:39
fijalso09:39
kenaan12arigo counter-decay 11b84533eb4350 15/pypy/: Implement decaying.  See comments.09:39
kenaan12arigo counter-decay 116e5fb0038df9 15/pypy/jit/metainterp/: Not tested: clean up the jitcell_dicts that are used internally.  This is probably important to ensure that a...09:39
kenaan12arigo counter-decay 11ba32c0042356 15/pypy/jit/metainterp/: Fix yet another obscure case in which we can end up with an apparent leak.09:39
kenaan12arigo counter-decay 11f7182dd71d78 15/pypy/jit/metainterp/: Test and fix.09:39
fijalI propose to have two kinds of arrays09:39
fijalVirtualView and NDimSlice09:39
fijaltwo kinds of views09:39
fijalthey're both completely separate09:40
Alex_Gaynorthats in addition to concrete arrays09:40
fijalyes09:40
Alex_Gaynorseems sane-ish09:40
fijalI'll make concrete array first09:40
fijalas a base class09:40
Alex_Gaynorthat's what we have now right?09:40
fijalright now we have only concrete view09:40
fijalbut implicitely so09:40
Alex_Gaynorok so you just propose to fix up slice09:41
fijalwell yes09:41
fijalbut they'll have different signatures09:41
fijal(obviously so)09:41
fijalbut then a concrete view is just a non-contiguous array09:41
fijalok?09:41
Alex_Gaynordoesn't have to be noncontigious, but right09:42
fijal[there is no question]09:42
fijalright09:42
fijalso when do we can share the view iterators?09:42
fijalcan we share09:42
fijalu[:-1] with u[1:]09:42
fijal?09:42
aleksi (aleksi@85.235.191.82) left #pypy.09:43
Alex_Gaynorno09:43
Alex_Gaynorbecuase first starts at 0, second at 109:43
Alex_Gaynorso you're at different indexes09:43
fijalwell, you can if you try hard enough09:43
Alex_Gaynorif you move the base array ptr of one you mean?09:44
fijalyou need to do addition at each iteration though09:44
Alex_Gaynorthen you have 2 array ptrs, which seems worse than 2 counters09:44
fijalwe don't do pointer addition now at all anyway09:44
Alex_Gaynorright09:44
Action: nirbheek wonders if he can coax someone to fix https://bugs.pypy.org/issue95609:45
antocuniarigato: what if counter-decay about? To kill the old jitted code?09:46
fijalnirbheek: you can try to fix it yourself :)09:46
fijal"However, Googlers work every day on increasingly more complex problems"09:47
fijallike a new widget in the top left corner of gmail09:47
nirbheekfijal, I don't even know where to begin :S09:47
fijalnirbheek: can you reproduce it with py.py for example?09:47
arigatoantocuni: yes, it's about the fact that counters should get exponential decays because that's theoretically nice, with the practical side-effect of cleaning up "jitcell_dict" from time to time09:47
arigatoantocuni: note that jitcell_dict is not used for the normal jitdriver because we store jitcells on the PyCode objects09:48
fijalif you can provide a failing test that does not require translation09:48
nirbheekfijal, I found that binary, but what is that?09:48
fijalthat would be a good first step09:48
arigatoanish_: but it's used for all other jitdrivers09:48
fijalthis is an untranslated interpreter09:48
arigatoantocuni: sorry09:48
nirbheekfijal, okay, trying now09:48
arigatoanish_: sorry, meant to say "antocuni".09:48
fijalAlex_Gaynor: so the broadcast view is just another view09:49
fijalwhich can also be both concrete and not09:49
fijalit just has different strides09:49
antocuniand why it's theoretically nice?09:49
Alex_Gaynorfijal: sounds sane09:49
fijalnow, how do I represent a virtual view?09:49
Alex_Gaynorwhat do you mean?09:49
Alex_Gaynorshow my python code which generates a virtual view?09:49
arigatoantocuni: mostly because decaying counters have been used by all JITs from the Self one09:50
fijal(a + a)[::2]09:50
nirbheekfijal, yep, I can reproduce it with py.py as well09:50
Alex_Gaynorso it should be a slice which happens to have virtual contents09:50
fijalnirbheek: good, now make it into a failing test in test_posix then09:50
fijalpypy/module/posix/test/...09:50
fijalAlex_Gaynor: yes, but you can't compute strides really09:50
fijalbecause you have no concrete storage09:50
arigatoantocuni: actually the practical benefit is probably larger: it avoids a leak if you keep making new regexps that you call only once09:50
Alex_Gaynorfijal: you can get the dimensions of a virtual array without computing contents though09:51
fijalyes, but not strides09:51
nirbheekfijal, ./pypy/module/posix/test/test_posix2.py ?09:51
antocuniyes, that's nice09:51
fijalnirbheek: yes09:51
Action: nirbheek is on it09:51
Alex_Gaynorfijal: why not strides?09:51
antocuniwhy don't we store the JItCell on the compiled regex object btw?09:51
fijalAlex_Gaynor: because strides assume some location09:51
fijalni the space09:51
Alex_Gaynorhow do you mean?09:52
arigatoantocuni: ah, also, it should fix test_pypy_c.test_generator out of the box09:52
fijalwhile you have no space, you can't have the location09:52
dracman (~draco@212.255.36.109) joined #pypy.09:52
arigatoantocuni: simply because when we compile the first loop, the second's counter is decayed a bit09:52
Alex_GaynorI'm confused09:52
arigatoantocuni: so it won't be immediately compiled09:52
fijals = strides[i] means elements in dimension i are s bytes apart09:52
fijalor s * sizeof(dtype) bytes apart09:52
fijalright?09:52
Alex_Gaynoryes, why can't you do that without actual contents of the array09:52
anish_arigato, its ok09:53
fijalit's a bit pointless09:53
antocuninice09:53
arigato:-)09:53
arigatoantocuni: good question09:53
fijalarigato: yeah, I would not be against requiring to provide a storage09:53
arigatoantocuni: ah yes, because it's a mess because there are a number of jitdrivers, not just one09:53
fijalAlex_Gaynor: because you compute the offset, but then you have to unwrap it again and recompute09:53
fijalI mean offset is a bit meaningless if you don't have a concrete storage09:53
Alex_Gaynorfijal: yes, but you can compute stride, and then use it on some later-computed concrete storage09:54
arigatofijal: that was meant for Alex_Gaynor instead of me, right?09:54
fijalarigato: no, that was to you. I would not mind if jitcell_dict would die completely09:54
fijaland the jitdriver author would have to provide a storage for jitcells09:54
arigatofijal: ah, well, I certainly would09:54
fijalbecause it's messy?09:55
arigatoit would be a mess in rsre_core09:55
arigatoand in *all* tests09:55
fijalAlex_Gaynor: you would need stride recomputation09:55
Alex_Gaynorfijal: alternatively, maybe strides should measure something besides bytes09:55
fijala biiit pointless09:55
fijalit's probably better to just do something simpler09:55
fijallike remember the way slice was created or so09:55
Alex_GaynorI highly endorse simpler things09:55
nirbheekfijal, do I just add an fdopen to the pre-existing test_openpty (hence testing two things at once) or should I add a new test_fdopen_openpty?09:56
fijalnirbheek: add a new test09:56
nirbheekalrighty09:56
fijalAlex_Gaynor: thank you, that does not really answer any of my issues :)09:57
xiaochen (xiaochen@nat/redhat/x-uywqmnhytazzzmpv) left irc: Ping timeout: 252 seconds09:57
fijalpom pom pom09:57
Alex_Gaynorfijal: I'm a duck, solve your own problems :)09:57
fijalhakanardo_: are you fixing the _raw forcings?09:57
arigatonirbheek: I'm unsure but I guess that it's trying to use seek() internally09:58
arigatono clue why the problem doesn't show up with pipes09:58
Action: nirbheek did find it curious that it *only* showed up with openpty09:59
nirbheekfijal, I did this: http://dpaste.com/673636/10:01
nirbheekhow do I make sure that the testsuite runs that test?10:01
fijalrun py.test10:02
fijalor pytest.py10:02
fijalor however it is named these days10:02
nirbheekit'll pick up the new test automatically?10:02
Action: nirbheek was unsure how pytest.py worked10:03
fijalyes10:03
nirbheekokay, thanks :)10:03
arigatopython test_all.py module/posix/test/test_posix2.py10:03
fijalit'll find everything that's in the directory/file10:03
nirbheekarigato, I ran it as: ./pytest.py ./pypy/module/posix/test/test_posix2.py10:04
fijalAlex_Gaynor: ok, I'll make a break and think a bit on the surfboard10:04
fijalusually helps10:04
arigatonirbheek: ok10:05
fijalnirbheek: good10:05
arigatonirbheek: ah, your test deadlocks10:06
arigatoyou have to close to slave_fd first10:06
nirbheekSo *that's* why it's just sitting there doing nothing :p10:06
arigato:-)10:06
nirbheekoh crap!10:06
nirbheekhow silly of me10:07
nirbheekthere, it's giving a nice ....F.... now10:08
arigatonirbheek: ah10:09
arigatoactually it's a behavior of Linux10:09
arigatohttp://paste.pocoo.org/show/521184/10:09
Action: arigato looks how CPython hides it10:10
arigatobeh, it's probably the libc that hides it10:10
Action: nirbheek does not understand this :S10:13
nirbheekShouldn't an EOF give an IOError anyway?10:14
arigatono, normally os.read() should return ''10:15
arigatothat's what occurs for regular files and for pipes10:15
nirbheekhmmm, I see10:17
khs (~khs@151.43.202.84.customer.cdi.no) joined #pypy.10:18
amaury_ (~amaury_@74.125.57.34) joined #pypy.10:21
arigatouh?10:22
fijalhow do I get the common prefix of two strings?10:23
arigatoah, the definition of file.read() without argument is:10:23
arigatoif we get some data but then an error, it's returning that data and ignoring the error10:23
arigatoI see the point but that's still quiiiite obscure10:23
nirbheekSo if it gets no data and an error, it propagates that error?10:24
arigatoyes10:24
nirbheekthis bug seems related: http://bugs.python.org/issue538010:25
fijaluh10:25
nirbheekHowever, it looks like that bug was somehow fixed10:25
fijalarigato: help10:25
amaury_ (~amaury_@74.125.57.34) left irc: Read error: Operation timed out10:25
arigato?10:25
fijalarigato: I'm trying to fix caching of modules10:25
fijalwhich are based on eci10:25
fijalso10:26
fijalone ECI is10:26
fijalseparate_module_files=('/tmp/usession-default-37/module_cache/module_0.c',)10:26
fijaland the other is the same except a different number10:26
fijalshould we just ignore separate_module_files?10:26
arigato? no10:26
arigatothat looks bogus10:27
arigatoyou could read the file's content and put it in the md5 computation10:27
fijalyes, it's already done10:28
arigatothen of course it makes no sense to also keep the string '/tmp/usession-37/..'10:28
nirbheekarigato, most interestingly, I can reproduce this bug on python3 as well!10:30
nirbheekHow strange10:30
arigatonirbheek: uh, how?  I tried and failed10:31
nirbheekarigato, I ran this test: http://dpaste.com/673641/10:31
nirbheekwhich should be equivalent10:31
nirbheek(slightly modified version of the test attached to the issue)10:32
arigatonirbheek: ah bah, I understood what's going in Python 2.x10:32
arigatonirbheek: let me check10:32
arigatoyes10:34
arigatoif len(test_string) == 6 for example, then it crashes on Python 2.x too10:35
voidspace (~voidspace@python/psf/voidspace) joined #pypy.10:37
arigatoso you should not use this idiom because it's broken on CPython in some cases10:37
arigatoand you should report it as a bug to CPython, I suppose10:37
kenaan12fijal default 11976baeae86ed 15/pypy/tool/gcc_cache.py: improve caching10:38
nirbheekargh!10:38
nirbheekHow silly10:38
arigatoyes :-(10:38
arigatoit makes no sense that it works or fails based on internal buffering issues10:39
arigatobut then it's a bit unexpected that (on Linux at least) the pty does not return "" when reading past the end10:39
nirbheekwait, what did you mean by <arigato> if len(test_string) == 6 for example, then it crashes on Python 2.x too ?10:40
nirbheekI changed the length of test_string and it still passes on python210:40
arigatouh10:40
arigatonot for me :-(10:40
arigato32-bit python 2.510:41
arigatoah no10:41
arigato32-bit python 2.7 trunk10:41
arigatoindeed, with python 2.6 it works10:41
nirbheekpasses for me on 2.6.6 and 2.7.210:41
arigatofails for me with Python 2.7.2+10:41
nirbheekhorrible, horrible10:42
arigatoyes10:42
arigatoindeed, 2.7.1 works too10:42
nirbheekI think it's time to get a test case for this into python trunk :p10:42
arigatoah, but that's just a fluke10:42
arigatoit fails with a string of length 8192 on 2.7.110:42
nirbheekNow you're just screwing with me.10:42
arigatono10:43
arigatoit's the size of the buffer10:43
nirbheekIt does!10:43
nirbheekoh man10:43
arigato:-(10:43
nirbheekthis was actually reduced from a test failure in gentoo portage's testsuite10:47
nirbheekI'm going to talk to the portage devs and see what the correct thing to do is10:47
fijalnice10:47
CIA-1303arigo 07roundup * 10#956/os.fdopen() is broken with fds returned by os.openpty(): 10:47
CIA-13[chatting] A bug of CPython too? The included test file fails on CPython <= 2.7.2 if10:47
CIA-13len(test_string) == 8192, and on CPython 2.7 trunk if len(t ... * 14https://bugs.pypy.org/issue95610:47
fijalarigato: so, out of 66s of running test_numarray.py on top of pypy, 32s was spent tracing10:47
arigatonot too surprizing10:48
fijalno?10:48
fijalhttp://paste.pocoo.org/show/521192/10:48
fijaland please don't tell me about interpreters on top of interpretrs, because I don't buy it :)10:49
nirbheekarigato, thanks a lot for your help! I'll be back after pestering the portage devs about this :)10:50
fijalarigato: if I run it twice in a row, I spent 60s tracing10:52
arigatonirbheek: :-)10:52
fijalarigato: don't you think it's bad?10:54
fijalarigato: what's the different between ops and recorded ops?10:56
arigatofijal: "ops" includes the constants-during-tracing10:58
azanella_ (~azanella@189.6.80.207) joined #pypy.10:58
arigatofijal: bad: yes, I think it's bad, but it's nothing new as far as I can tell10:58
fijalit's certainly not new10:58
fijalit was always around10:59
fijalbut I claim this is the single biggest obstacle for *my* pypy adoption10:59
fijalarigato: for example it seems we have single functions that exceed trace limit11:02
fijalwe gonna trace such functions again and again11:02
fijalnever compiling11:05
antocunidon't we have a flag to disable tracing such functions?11:05
fijalwe only mark them as "trace from start"11:05
antocunibut then we keep tracing from start again and again11:06
antocunipff11:06
fijalyes11:07
fijalfor example11:07
fijalarigato: I claim there is some more info to be got from there :)11:07
bivab (~david@fwstups.cs.uni-duesseldorf.de) joined #pypy.11:08
voidspace (~voidspace@python/psf/voidspace) left irc: Quit: Uhm... gotta go11:10
hakanardo_fijal: yes11:23
anish_ (anish@nat/redhat/x-ychfmgmnjtxalzmb) left irc: Ping timeout: 248 seconds11:23
fijalhakanardo_: cool11:24
kenaan12hakanardo default 1194f53881b7bf 15/pypy/jit/metainterp/optimizeopt/: setinteriorfield_should_not_clear_cache11:24
kenaan12hakanardo default 11ee80fd67c26a 15/pypy/tool/gcc_cache.py: hg merge11:24
fijalhakanardo_: it's everything _raw11:25
fijalthat has no effects on caches11:25
fijalI believe11:25
hakanardo_which are missing?11:26
fijalI dunno11:27
fijalare there?11:27
fijaleh11:27
fijalI'm missing a way to say where was the trace aborted :/11:27
Action: fijal fishes oldish branches11:27
ojii (~ojii@40-34.60-188.cust.bluewin.ch) joined #pypy.11:28
voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.11:29
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host11:29
voidspace (~anonymous@python/psf/voidspace) joined #pypy.11:29
hakanardo_fijal: I think they are all there, setinteriorfield was the onlyone missing11:30
fijalok11:30
fijal (~fijal@41.53.118.135) left irc: Read error: Connection reset by peer11:31
nedbat (~nedbat@python/psf/nedbat) joined #pypy.11:31
bgola (~bgola@c9518082.virtua.com.br) left irc: Ping timeout: 252 seconds11:39
mikefcooo. 5x11:48
Action: mikefc is behind the times11:48
davidpk (~davidpk@0278be3d.bb.sky.com) joined #pypy.11:49
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) joined #pypy.11:52
kenaan12arigo counter-decay 119df7dbac5802 15/: close branch for merge11:58
kenaan12arigo default 115309a1389556 15/pypy/: Merge the counter-decay branch:  * add a theoretically useful "exponential decaying" of loop counters  * it is usef...11:58
kenaan12arigo default 11e790db7af776 15/pypy/jit/metainterp/warmstate.py: add an assert.11:58
kenaan12arigo default 1115811e23d71a 15/pypy/jit/metainterp/memmgr.py: fix11:58
kenaan12arigo default 110a4ac8556799 15/pypy/jit/metainterp/optimizeopt/: merge heads11:58
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) joined #pypy.12:04
amaury_ (amaury_@nat/google/x-nszzirxcrlvlovib) joined #pypy.12:18
fijal (~fijal@41.177.93.12) joined #pypy.12:18
canta (~canta@77-20-123-240-dynip.superkabel.de) left irc: Quit: Leaving.12:18
kenaan12fijal virtualizable-experiments 1140d47febc01e 15/: close branch that got nowhere, I don't even remember what it was about12:24
kenaan12fijal faster-json 114d5064194771 15/: close merged branch12:24
kenaan12fijal numpy-share-iterators 1183a3724c72f3 15/: superseeded by refactor-signature12:24
kenaan12fijal better-jit-hooks 11e3c7e20c107b 15/: merge in default12:24
Action: fijal cleaned up his branches a bit12:24
fijalarigato: I assume you have no deep feelings about those tests12:24
Action: arigato would love to know which tests12:25
fijal (~fijal@41.177.93.12) left irc: Ping timeout: 240 seconds12:29
arigato<spam>12:38
kenaan12arigo closed-branches 11ecda4725a574 15/: Merge closed head 6d7644a6fd38 on branch no-force-guard-lazy-set12:38
kenaan12arigo closed-branches 11fd60224546b1 15/: Merge closed head 105fe17b27f7 on branch gmp12:38
kenaan12arigo closed-branches 118c402b1fe7b9 15/: Merge closed head 48639d927f25 on branch releasegil-effectinfo12:39
kenaan12arigo closed-branches 11b873a48dd65a 15/: Merge closed head 7a58d06641f3 on branch numpy-minilang12:39
kenaan12arigo closed-branches 11dce334355118 15/: Merge closed head 0ede8b92968e on branch rgc-mem-pressure12:39
kenaan12arigo closed-branches 1148804171531e 15/: Merge closed head 9d40404468cf on branch win64 test12:39
kenaan12arigo closed-branches 113a0dec36ccd3 15/: Merge closed head ca0f81ea74b5 on branch win64_gborg12:39
kenaan12arigo closed-branches 11caa80f4bb7b8 15/: Merge closed head ee74794f5464 on branch numpy NDimArray12:39
kenaan12arigo closed-branches 119f9d57f80be2 15/: Merge closed head 9d59a8501c00 on branch win64_gborg12:39
kenaan12arigo closed-branches 11acc499164a5c 15/: Merge closed head 03816b1a901c on branch win64 test12:39
kenaan12arigo closed-branches 117847dc9bc66c 15/: Merge closed head d0d86e088b8b on branch numpy-multidim-shards12:39
kenaan12arigo closed-branches 1116c7651dc366 15/: Merge closed head 2f25f1d0005c on branch numpy-monkeyaround12:39
kenaan12arigo closed-branches 118b343089086b 15/: Merge closed head 1ea6c70d9971 on branch micronumpy-resync12:39
kenaan12arigo closed-branches 117be0184f6d38 15/: Merge closed head df45bf0c2ee7 on branch numpy-multidim-exp12:39
kenaan12arigo closed-branches 11ae7bb10725a5 15/: Merge closed head 9ab059335d1f on branch jit-refactor-tests12:39
kenaan12arigo closed-branches 11103191efcd4a 15/: Merge closed head 78795591bf65 on branch matrixmath12:39
kenaan12arigo closed-branches 119ee3e533c081 15/: Merge closed head c8b9ca8d5c25 on branch matrixmath-reshape12:39
kenaan12arigo closed-branches 11e0b744896524 15/: Merge closed head 83b5695339e8 on branch matrixmath-reshape-merge12:39
kenaan12arigo closed-branches 114fa228cccdda 15/: Merge closed head 2db53dd9c46e on branch temp212:39
kenaan12arigo closed-branches 1133b58ee5d2da 15/: Merge closed head 2eb520555de5 on branch nedbat-sandbox12:39
kenaan12arigo closed-branches 11619b713a5bb6 15/: Merge closed head a48a35a5c618 on branch SpecialisedTuples12:39
kenaan12arigo closed-branches 111c5de2c38245 15/: Merge closed head 53d2c2028ef3 on branch numpy-pi-sum-min-max12:39
kenaan12arigo closed-branches 110b366011ad13 15/: Merge closed head 539833042097 on branch jit-simplify-backendintf12:39
kenaan12arigo closed-branches 11dc1585148802 15/: Merge closed head b391b4c1ef21 on branch numpy-identity12:39
kenaan12arigo closed-branches 1191e205942073 15/: Merge closed head 40d47febc01e on branch virtualizable-experiments12:39
kenaan12arigo closed-branches 113ef224190801 15/: Merge closed head 4d5064194771 on branch faster-json12:39
kenaan12arigo closed-branches 117fe3a58c1918 15/: Merge closed head 83a3724c72f3 on branch numpy-share-iterators12:39
kenaan12arigo closed-branches 114f16091a6497 15/: re-close this branch12:39
kenaan12arigo default 118c37cae3aea2 15/pypy/tool/clean_old_branches.py: fix to handle branch names with special characters, like spaces12:39
stakkars_ (~tismer@p5DDB72FD.dip.t-dialin.net) joined #pypy.12:48
stakkars (~tismer@p5DDB72FD.dip.t-dialin.net) left irc: Read error: Connection reset by peer12:48
Nick change: stakkars_ -> stakkars12:48
nedbatI'm getting this error when I run the sandbox after updating my tree: http://paste.pocoo.org/show/521108/12:50
nedbatI didn't think I changed the code I was running in the sandbox, but it's more likely to be my fault than anyone else's.  I just don't know how to track it down.12:51
arigatonedbat: it's just a ValueError because you call int("some non-integer string")12:55
nedbatarigato: in my Python code running in the sandbox?12:56
arigatono, in the RPython code from targetpypystandalone.py12:56
arigato(strangely enough)12:56
nedbatarigato: I don't think I changed any of that, how would I track it down further?12:57
arigatoe.g. you pass "--heapsize foobar"12:57
nedbathmm, ok.12:57
arigatoor "--heapsize 5M"12:57
arigatoit just expects to pass the argument to int()12:57
arigato(--heapsize is a not-really-supported option)12:57
nedbatah, I see, yes, pypy_interact accepts more interesting strings. ok12:57
nedbatarigato: how did you know it was the heapsize argument?12:58
oal (u4126@gateway/web/irccloud.com/x-mrrmdxqibenxbofl) left irc: Remote host closed the connection12:58
alexsuraci (u1962@gateway/web/irccloud.com/x-fmgkunpvlhywhrbp) left irc: Remote host closed the connection12:58
Alex_Gaynor (u1246@gateway/web/irccloud.com/x-unwydkcmqavjigvx) left irc: Remote host closed the connection12:58
hongminhee (u295@gateway/web/irccloud.com/x-xwvmgozxcovfnhfl) left irc: Remote host closed the connection12:58
bfirsh (u1308@gateway/web/irccloud.com/x-bxagoggwardmynye) left irc: Read error: Connection reset by peer12:58
oal (u4126@gateway/web/irccloud.com/x-mnsqyvfaeeiuegvd) joined #pypy.12:58
arigatot's the only place in targetpypystandalone.py that calls int()12:58
arigatowhich you can see in the traceback, from the name of the function ll_int()12:59
nedbatarigato: and you knew it was in that file because translator_goal_targetpypystandalone.c is the C translation of translator/goal/targetpypystandalone.py12:59
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.12:59
arigato(but I agree that it's a bit involved)12:59
arigatonedbat: yes12:59
nedbatok12:59
nedbatand ll_int__rpy_stringPtr_Signed means, "create a native int from an rpy string"?13:00
azanella (~azanella@189.6.80.207) left irc: Quit: Ex-Chat13:01
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.13:01
arigatoyes, it's the mangled name of the function ll_int() that is defined in rpython/lltypesystem/rstr.py13:01
nedbatarigato: thanks, this was very helpful.13:01
arigatopypy_interact.py should not let non-integer values through --heapsize, but maybe you hacked that13:02
hongminhee (u295@gateway/web/irccloud.com/x-rfobsaryywiizdur) joined #pypy.13:03
nedbatarigato: yes, that's me.13:04
aboudreault__ (~alanb@199.91.205.186) joined #pypy.13:04
lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) joined #pypy.13:06
arigato (~arigo@89.204.154.6) left irc: Ping timeout: 252 seconds13:07
Nick change: aboudreault -> Guest2125413:07
Nick change: aboudreault__ -> aboudreault13:08
Guest21254 (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 268 seconds13:08
aboudreault (~alanb@199.91.205.186) left irc: Changing host13:08
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.13:08
jnoller (~jnoller@vpn.nasuni.com) joined #pypy.13:09
alexsuraci (u1962@gateway/web/irccloud.com/x-utxyxvaiaizvxxfy) joined #pypy.13:12
bfirsh (u1308@gateway/web/irccloud.com/x-zjbqpjgbhoifdbza) joined #pypy.13:13
G2P (~G2P@fw-asn1.ornis.com) left irc: Ping timeout: 240 seconds13:21
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.13:22
cwillu_at_work (~cwillu@cwillu-1-pt.tunnel.tserv13.ash1.ipv6.he.net) joined #pypy.13:37
dgl (~dgl@109.86.165.231) joined #pypy.13:42
dgl (~dgl@109.86.165.231) left irc: Client Quit13:42
overminder (~overminde@143.89.167.109) joined #pypy.13:44
derdon (~derdon@p5DE8A908.dip.t-dialin.net) joined #pypy.13:45
harrison (~quassel@adsl-76-217-35-156.dsl.chcgil.sbcglobal.net) joined #pypy.13:47
ch_jyx (~ch_jyx@csl2wk29.cse.ust.hk) joined #pypy.13:48
harrison (~quassel@adsl-76-217-35-156.dsl.chcgil.sbcglobal.net) left irc: Remote host closed the connection13:50
overminder (~overminde@143.89.167.109) left irc: Quit: Leaving13:50
hwangad (~hwangad@csl2wk05.cse.ust.hk) joined #pypy.13:51
hwangad (~hwangad@csl2wk05.cse.ust.hk) left irc: Client Quit13:52
ch_jyx (~ch_jyx@csl2wk29.cse.ust.hk) left irc: Quit: using sirc version 2.211+KSIRC/1.3.1213:53
alexsuraci (u1962@gateway/web/irccloud.com/x-utxyxvaiaizvxxfy) left irc: Remote host closed the connection13:58
Nick change: Gulaway -> Gulopine13:59
alexsuraci (u1962@gateway/web/irccloud.com/x-fhdmjxumxcvrqyin) joined #pypy.14:00
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Quit: Leaving14:03
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.14:03
gutworth (~benjamin@dhcp-128-36-84-91.central.yale.edu) joined #pypy.14:06
whitelynx (~whitelynx@63.241.75.144) joined #pypy.14:09
Alex_Gaynor (u1246@gateway/web/irccloud.com/x-qgyzmgmavjyscztf) joined #pypy.14:11
#pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services.14:11
logan` (~loganchie@canlover.m6.ntu.edu.tw) joined #pypy.14:17
Nick change: Gulopine -> Gulaway14:21
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.14:25
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) left irc: Read error: Connection reset by peer14:25
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.14:28
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.14:28
arigato (~arigo@fwstups.cs.uni-duesseldorf.de) joined #pypy.14:29
mattip (c1a9466d@gateway/web/freenode/ip.193.169.70.109) joined #pypy.14:32
mattip (c1a9466d@gateway/web/freenode/ip.193.169.70.109) left irc: Client Quit14:34
Eventh (evenwiik@caracal.stud.ntnu.no) joined #pypy.14:36
fprimex (~fprimex@brent-macbook.sc.fsu.edu) joined #pypy.14:42
gutworth (~benjamin@dhcp-128-36-84-91.central.yale.edu) left irc: Ping timeout: 255 seconds14:42
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Computer has gone to sleep.14:44
berdario (~Tiibiidii@217.220.118.217) joined #pypy.14:46
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) joined #pypy.14:48
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.14:57
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Client Quit15:00
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace15:01
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.15:03
kkris (~kris@80-123-35-252.adsl.highway.telekom.at) joined #pypy.15:03
cwillu_at_work (~cwillu@cwillu-1-pt.tunnel.tserv13.ash1.ipv6.he.net) left irc: Remote host closed the connection15:06
cwillu_at_work (~cwillu@cwillu-1-pt.tunnel.tserv13.ash1.ipv6.he.net) joined #pypy.15:11
kenaan12antocuni buildbot 11d4fc18217372 15/bot2/pypybuildbot/master.py: add new builder/slave to run benchmarks on speed.python.org15:21
kenaan12antocuni buildbot 114d1b872dbf60 15/bot2/pypybuildbot/master.py: hg merge default15:21
cwillu_at_work (~cwillu@cwillu-1-pt.tunnel.tserv13.ash1.ipv6.he.net) left irc: Remote host closed the connection15:25
santagada (~leonardo@189.27.187.218.dynamic.adsl.gvt.net.br) joined #pypy.15:26
bivab (~david@fwstups.cs.uni-duesseldorf.de) left irc: Quit: bivab15:26
antocunibah15:26
antocunisomeone made some change in buildmaster's wc and did not checkin15:26
antocuniarigato: is it you maybe who added "snakepit32" to the slavenames?15:27
cwillu_at_work (~cwillu@cwillu-1-pt.tunnel.tserv13.ash1.ipv6.he.net) joined #pypy.15:28
kenaan12antocuni buildbot 11922a46875231 15/bot2/pypybuildbot/master.py: fix the config for JITBENCH64_2, and add snakepit32 to the slavenames (this change was actually lying in the WC...15:29
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Textual IRC Client: http://www.textualapp.com/15:37
logan` (~loganchie@canlover.m6.ntu.edu.tw) left irc: Remote host closed the connection15:38
Ademan (~dan@adsl-71-141-224-79.dsl.snfc21.pacbell.net) left irc: Quit: leaving15:40
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/0 [12antocuni]15:41
Kaskuka (~John@osbk-4db14206.pool.mediaWays.net) left irc: Ping timeout: 240 seconds15:42
Kaskuka (~John@osbk-4d08b9ca.pool.mediaWays.net) joined #pypy.15:42
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.15:44
antocunicool, now we have the super-fast speed.python.org as a slave15:44
antocuni64bit only for now15:44
antocuniand only for benchmarks15:44
bbot24Failure: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/0 [12antocuni]15:47
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/1 [12antocuni]15:51
voidspace (~anonymous@python/psf/voidspace) joined #pypy.15:55
arigatoantocuni: cool15:56
arigatosorry, it may be me15:56
antocuninp, I checked it in15:56
antocunipff, but now we have a problem with locks15:57
antocunibecause the all JITBenchmark builds acquire the tannit master lock15:57
kenaan12arigo op_malloc_gc 118497ccd603ae 15/: hg merge default15:57
arigato:-/15:57
Action: antocuni adds even more workarouds15:58
arigatojust run the JITBENCH64_2 during the U.S. night instead15:58
arigatois there any way to coordinate between multiple projects using the machine?15:58
antocuniso far we are the only one :-)15:59
arigatoeasy :-)15:59
voidspace (~anonymous@python/psf/voidspace) left irc: Client Quit15:59
voidspace (~anonymous@python/psf/voidspace) joined #pypy.15:59
antocuniI think that in the fullness of time the solution would have to have a buildmaster just for speed.python.orgh15:59
arigatoyes16:00
antocunibut for now, we are just reusing pypy's one so that we can quickly put it on use16:00
kenaan12antocuni buildbot 11f550f9557270 15/bot2/pypybuildbot/: bah, make sure that benchmarks on speed.python.org don't acquire the lock for tannit16:03
kenaan12antocuni buildbot 11c34d2583efd4 15/bot2/pypybuildbot/master.py: hg merge default16:03
kenaan12hager ppc-jit-backend 113167c1547eef 15/pypy/jit/backend/ppc/ppcgen/: (bivab, arigo, hager): implemented COND_CALL_GC_WB16:03
mvt (~mvantelli@87.213.45.85) left irc: Quit: This computer has gone to sleep16:04
kenaan12hager ppc-jit-backend 113ea35dae52a6 15/pypy/jit/backend/ppc/ppcgen/: implemented COND_CALL_GC_WB_ARRAY16:18
davidpk (~davidpk@0278be3d.bb.sky.com) left irc: Quit: davidpk16:21
jterrace (~jterrace@aegis.CS.Princeton.EDU) joined #pypy.16:23
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) left irc: Quit: Leaving...16:31
witulski (~stupsi@134.99.16.21) joined #pypy.16:47
kenaan12arigo op_malloc_gc 1130a943d165f9 15/pypy/jit/backend/llsupport/: In-progress.16:51
kenaan12arigo op_malloc_gc 110304705f6571 15/pypy/jit/backend/llsupport/: Progress...16:51
kenaan12arigo op_malloc_gc 116caef25ddb53 15/pypy/jit/backend/llsupport/: fixes16:51
witulski (stupsi@134.99.16.21) left #pypy.16:52
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.16:54
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.16:58
kushal (~kdas@fedora/kushal) joined #pypy.17:00
davidpk (~davidpk@0278be3d.bb.sky.com) joined #pypy.17:04
davidpk (~davidpk@0278be3d.bb.sky.com) left irc: Client Quit17:04
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Ping timeout: 240 seconds17:06
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection17:07
G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving.17:09
arigato (~arigo@fwstups.cs.uni-duesseldorf.de) left irc: Quit: See you17:20
bgola (~bgola@c9518082.virtua.com.br) joined #pypy.17:32
fijal (~fijal@197.170.170.208) joined #pypy.17:38
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) left irc: Quit: etrepum17:39
fcorrea (~fcorrea@187.38.33.117) joined #pypy.17:41
fijalantocuni: ping17:42
amaury_ (amaury_@nat/google/x-nszzirxcrlvlovib) left irc: Ping timeout: 252 seconds17:43
jimbaker (~jbaker@canonical/jimbaker) left irc: Ping timeout: 240 seconds17:44
jterracefijal: what's with all the closed branches?17:46
jterracenumpy-identity got closed17:46
fijalwas it merged?17:46
fijaldidn't I merge it?17:46
jterraceoh nevermind17:46
jterracei was thinking numpy-concatenate which wasn't merged yet17:46
jterracebut that one survived17:46
jimbaker (~jbaker@c-75-71-80-146.hsd1.co.comcast.net) joined #pypy.17:48
jimbaker (~jbaker@c-75-71-80-146.hsd1.co.comcast.net) left irc: Changing host17:48
jimbaker (~jbaker@canonical/jimbaker) joined #pypy.17:48
kushal (kdas@fedora/kushal) left #pypy ("Leaving").17:53
papercrane (~papercran@75.101.111.82) joined #pypy.17:56
papercrane (~papercran@75.101.111.82) left irc: Read error: Connection reset by peer17:58
papercrane (~papercran@75.101.111.82) joined #pypy.18:00
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) joined #pypy.18:01
JaRoel (~jaroel|4d@sink.jaroel.nl) joined #pypy.18:03
papercrane (~papercran@75.101.111.82) left irc: Client Quit18:05
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) left irc: Ping timeout: 244 seconds18:06
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace18:12
etrepum (~bob@accessnat4.mochimedia.net) joined #pypy.18:14
Eventh (evenwiik@caracal.stud.ntnu.no) left irc: Quit: Lost terminal18:16
jimbaker` (~jbaker@c-75-71-80-146.hsd1.co.comcast.net) joined #pypy.18:17
jimbaker (~jbaker@canonical/jimbaker) left irc: Ping timeout: 255 seconds18:19
bbot24Failure: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/1 [12antocuni]18:28
santagada (~leonardo@189.27.187.218.dynamic.adsl.gvt.net.br) left irc: Ping timeout: 240 seconds18:34
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/2 [12fijal]18:37
DasIch (~dasich@p3E990E33.dip.t-dialin.net) left irc: Ping timeout: 240 seconds18:39
santagada (~leonardo@189.27.182.208.dynamic.adsl.gvt.net.br) joined #pypy.18:46
DasIch (~dasich@p4FFDED50.dip.t-dialin.net) joined #pypy.18:51
fijal (~fijal@197.170.170.208) left irc: Ping timeout: 252 seconds18:54
mwhudson (~mwh@linaro/mwhudson) joined #pypy.18:54
berdario (~Tiibiidii@217.220.118.217) left irc: Ping timeout: 252 seconds19:04
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.19:15
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) left irc: Remote host closed the connection19:16
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.19:16
jterraceAlex_Gaynor: ping19:27
Alex_Gaynorjterrace: pong19:27
jterracehey question for you about dtype('int')19:27
Alex_Gaynorsure19:27
jterracewhen i create an array in numpy with dtype=int, it becomes int6419:28
jterracesame in numpypy19:28
Alex_Gaynorit has int64 for a string, but try numpy.dtype(int) is numpy.dtype("int64")19:28
jterracebut when i use dtype=int in fromstring, it's using the Integer box19:28
jterraceinstead of the Integer64 box19:28
jterraceoh19:28
jterracehmm19:29
jterracedtype becomes Integer instead of UInt64 when i set dtype=int in fromstring19:29
Alex_Gaynorerr sorry it's dtype(int) is dtype(long)19:29
jterraceso what format code should i use in Integer?19:29
jterraceis it always 64-bit, or is it platform dependent?19:29
Alex_Gaynorplatform dependent19:30
Alex_Gaynorit's 32-bit on 32-bit systems, and 64 on 6419:30
Alex_GaynorI'm not sure if rstruct has a format code for that, so you might need to put a branch19:30
jterracehmm ok19:30
jterracealso, you're right that fromstring doesnt work with dtype=bool and a sep value19:30
jterracebut it does for numpypy19:30
jterracedoes it need to arbitrarily not work for that case, or is it okay to have improved functionality?19:31
Alex_Gaynorah "l" maybe19:31
Alex_Gaynorjterrace: improved is fine with me, and if upstream ever fixes it we'll just make sure our behavior is inline with theirs19:31
jterraceok19:32
rekamso (~textual@216.190.27.14) joined #pypy.19:32
jterraceyeah confirmed that fromstring with int is 32-bit on 32-bit19:33
jterraceinteresting19:33
jterraceif i run pytest.py with a 32-bit python, does it run the tests with 32-bit?19:34
Alex_GaynorI guess, yes19:35
Alex_GaynorI'm not 100% sure though19:35
jterraceyes it does19:36
jterraceawesome19:36
jterraceon mac, it's easy with just arch -i386 python2.7 ...19:36
Alex_Gaynor"it's easy with just [long strong of confusing flags]"19:38
Alex_Gaynor:)19:38
jterracewell i'm familiar with them because they have been a pain in the ass before19:38
jterraceso at least they come in handy for once19:38
azanella_ (~azanella@189.6.80.207) left irc: Read error: Operation timed out19:42
jterrace'l' works nicely on both 32-bit and 64-bit19:43
kenaan12jterrace numpy-full-fromstring 11bca13f8d517c 15/pypy/module/micronumpy/: Updated tests for bool, added tests for int, all tests pass on 32-bit and 64-bit19:45
jterraceAlex_Gaynor: that ^ should address the rest of the review notes19:45
jterracei think numpy-concatenate is good to go also19:45
Alex_Gaynorawesome19:45
bugfree (~azhang@adsl-99-63-77-95.dsl.pltn13.sbcglobal.net) joined #pypy.19:47
Alex_Gaynorjterrace: format_code shouldn't be on Integer, it should be on Long19:47
Alex_GaynorInteger is a base class for all int types19:47
jterraceoh19:48
jterracemy bad19:49
Action: jterrace reruns tests19:49
Alex_Gaynorthey'll pass, since everything subclasses it and then overrides the attr19:49
jterracealways good to double check19:49
jterracewas wondering about numpy's ufunc.accumulate function19:50
jterraceany idea how to implement that?19:50
Alex_Gaynorno, but I have no idea what it is :)19:51
Action: Alex_Gaynor looks it up19:51
jterraceit would allow me to trivially add cumsum function19:51
Alex_Gaynorlooks pretty easy, take a look at the psuedo-code for it in the docstring19:52
jnoller (~jnoller@vpn.nasuni.com) left irc: Quit: Leaving...19:53
jterraceyeah i saw19:53
jterracethat's only for 1-dim though19:53
jterracebut maybe could be generalized for multidim with an axis19:53
Alex_GaynorI suppose so, weneed an axis iterator or something internally19:53
jterraceyeah19:53
kenaan12jterrace numpy-full-fromstring 11cf46c4c19b2f 15/pypy/module/micronumpy/types.py: Moved generic int code from Integer to Long19:54
jterracei kind of fudged it with concatenate19:54
Alex_Gaynordoes ulong have a format code?  if yes, feel free to merge19:54
jterraceyeah19:54
jterracei think i need to delete REVIEW file19:54
Alex_Gaynoroh yeah, please do that19:54
mattip (~chatzilla@bzq-79-183-120-145.red.bezeqint.net) joined #pypy.19:55
mattipjterrace: ping19:56
jterracemattip: pong19:56
mattipI see you want to take on cumsum19:56
kenaan12jterrace default 1165311ed125b7 15/pypy/module/micronumpy/: Merged numpy-full-fromstring19:56
mattipThere's a whole class of functions that take an axis argument, cumsum, sum, diff, ...19:57
Alex_Gaynorjterrace: don't forget to close the branch and remove the REVIEW file19:57
kenaan12jterrace numpy-full-fromstring 110fe83ac4f0da 15/: Closing merged branch19:57
jterracei think i just did19:57
Alex_Gaynorah ok :)19:58
jterracemattip: yeah, those are missing them19:58
mattipAre you looking to implement the axis functionality? Or can you do without it?19:58
jterracei dont think i need them right now19:58
jterracebut we should still implement them19:58
mattipyes, I have a branch matrixmath-dot that has a proof of concept implementation of double-iterators19:59
jterraceoh nice19:59
mattipOnce signatures/iterators are up to speed I plan to refactor and finish it.20:00
mattipor I guess someone else could do it too.20:01
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds20:03
mattipAlex_Gaynor: the frompyfunc branch is shaping up nicely. You seemed to want it, do you have a particular use case I could test?20:05
Alex_Gaynormattip: nope, I just think it's a cool case study of something that can be awesome on pypy, but not on CPython20:05
mattipyes, it seems to be an easy way to implement ufuncs from applevel20:06
ronnywhat does frompyfunc do?20:07
mattipsomething like f = frompyfunc(abs, 1, 1); out = f(arange(-5,5));20:08
mattipout==[5,4,3,2,1,0,1,2,3,4]20:08
kenaan12alex_gaynor default 112b3d72c181dd 15/pypy/module/micronumpy/: fixes for ulong dtype and some small other cleanups20:14
jterracemattip: check out the numpy-concatenate branch20:20
jterraceit iterates through an axis20:20
jterracebut it's pretty cheezy20:20
mattipok thanks. I'll look.20:20
Alex_Gaynorjterrace: that's written at applevel though, does numpy have an API for axis iteration at applevel?20:20
jterraceyou're looking at it20:20
jterraceas far as i know, i dont think so20:21
Alex_Gaynorah, ok20:21
jterraceyeah doesn't look like it: http://stackoverflow.com/questions/1589706/iterating-over-arbitrary-dimension-of-numpy-array20:21
jterracei basically just took advantage of that you can index with a tuple20:21
jterracex[:,0:3,:] == x[(slice(None), slice(0,3), slice(None))]20:22
jterraceit's not bad for concatenate because it's only O(N) where N = number of arrays being concatenated20:22
jterracesomething like cumsum would have to iterate over the entire lenght of the array20:22
jterraceit might just get JITed, but i think it would be better at rpython level20:23
jterrace(if im understnading it correctly)20:23
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Ping timeout: 240 seconds20:24
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Disconnected by services20:24
fijal (~fijal@197.172.87.224) joined #pypy.20:25
fijalhi20:25
Alex_Gaynorhey fijal 20:26
JaRoel (~jaroel|4d@sink.jaroel.nl) left irc: Ping timeout: 240 seconds20:26
ojii (~ojii@40-34.60-188.cust.bluewin.ch) left irc: Quit: Leaving20:28
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.20:31
fijalAlex_Gaynor: what's the roadmap for numpy?20:31
Alex_Gaynorfijal: add feature that people want.  make it faster.20:32
Alex_Gaynorfijal: I don't understand the question ;)20:32
squiddy (~squiddy@f053084237.adsl.alicedsl.de) left irc: Quit: Leaving20:32
fijal;)20:32
pjenvey1) make it work 2) ? 3) profit20:32
jterracethere's so much to be done, it seems the only sane approach is to fix things as people request them?20:32
Alex_Gaynorjterrace: at some point we want to start using the original numpy tests as well20:33
Alex_Gaynorthat should help in finding important missing features20:33
jterraceyeah20:33
jterracebut that's going to find a million missing htings20:33
mattipI like the 80-20 rule. Many of those tests are for corner features that are applicable in specialized problems20:35
fijaljterrace: how much are we actually missing btw?20:37
jterracea ton20:37
jterracethat's a technical term btw20:37
fijalthat's not very specific :)20:37
Alex_Gaynorwe're missing "stuff"20:38
bbot24Failure: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/2 [12fijal]20:38
jterracethe whole linalg package for one20:39
fijalthe whole linalg package is mostly implemented in C20:39
fijalwe don'20:39
fijalt need to reimplement that20:39
fijalwe'll just use it :)20:39
jterracewhen this function works, ill be happy: https://github.com/pycollada/meshtool/blob/master/meshtool/filters/simplify_filters/sander_simplify.py#L108-14420:40
Alex_Gaynora lot of it is pure python as well20:40
Alex_Gaynorfirst missing feature is indexing by None/Ellipsis20:41
fijalwe have everything but outer and linalg from the functions?20:41
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) joined #pypy.20:41
jterraceall the hstack, dstack, vstack, etc functions are missing20:42
jterracegrr, is scipy down for others or just me?20:43
Alex_Gaynorappears down (or at least slow) for me too20:44
santagada (~leonardo@189.27.182.208.dynamic.adsl.gvt.net.br) left irc: Quit: Ex-Chat20:44
fijaljterrace: they're not particularly hard20:44
jterracefijal: so here's a fun exercise20:45
jterraceprint this page out: http://web.archive.org/web/20110511204128/http://scipy.org/Numpy_Example_List_With_Doc20:45
fijalI agree there is a ton of functions out there20:45
jterracecover your eyes and point randomly20:45
fijalwell20:45
fijalI'm busy doing some other stuff :)20:45
fijallike making it run fast20:45
jterraceid bet good money the function you land on is not in numpypy :)20:45
jterracethat's what i mean by a "ton"20:45
fijalyes, but functions are easy20:45
fijalsome of them are 2-line-long20:46
jterraceno, a lot of them are not trivial20:46
Action: fijal landed on add20:46
Alex_Gaynorsome are, some aren't20:46
Alex_Gaynorhehe20:46
jterraceand a lot of the ones already in there (sum, min, max) are not even implemented correctly20:46
Alex_GaynorI agree, but, propgress, slowly :)20:46
fijaljterrace: well, ok20:47
jterraceagreed20:47
jterracei was just trying to answer his question :)20:47
fijalI mean the hard parts are mostly done20:47
jterraceyeah20:47
fijalwhat we need is another 90%20:47
fijalyes, sure20:47
fijalbut all of those functions can be implemented by whoever-needs-them20:47
fijalbecause they're mostly simple-to-trivial20:47
jterraceactually, making a table of those 413 functions and which ones are done would be nice to keep track of20:47
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.20:49
jterraceill keep helping out where i can - i have a large project that uses a large portion of numpy that i want to eventually get running on pypy20:49
jterracebecause it's slow and i want it to be fast20:50
fijaljterrace: good target :)20:51
Action: fijal has not much running on numpy20:51
jterracewell my first target is pycollada20:51
fijalbut I'm in love with SSE :)20:51
jterracewhich uses a small subset20:51
fijaljterrace: to put things in perspective, for me it's cooler to write a blog post that we overpowered C than to have it fully done20:51
Alex_Gaynorfijal: array sharing first, then sse20:52
fijalunless someone pays money20:52
mattipI like the money idea.20:52
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.20:53
Action: Alex_Gaynor -> dentist20:53
mattipI also like fast.20:54
jterraceyou're a dentist?20:54
fijalmattip: chances are fast will bring people who want complete20:55
fijallike jterrace :)20:55
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) left irc: Client Quit20:55
jterracewell i should be working on other stuff20:55
jterracebut pypy is fun20:55
mattipThen they might have money behind them.20:55
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.20:56
fijalmattip: I haven't found any yet, but maybe20:58
fijalanyway, it's way too easy to just wait20:58
mattipAnd some can only add more features (like me) with no idea of how to make it fast.20:58
fijalthat's fine20:58
fijalthe pypy architecture kinda makes it nice20:59
fijalobviously not entirely so20:59
fijalbut still20:59
mattipfijal: I took a look at the refactor-signature branch.21:00
mattipI think I can figure out broadcasting. Are you to a point of wanting code from others?21:01
jterraceindexing of arrays with other arrays and the string type are the big ones i need i think21:01
fijalsure, if you can figure out why not?21:01
fijalI don't want broadcasting to be special21:02
fijalI want it to be just another view21:02
fijalbut I want to implement virtual views first21:02
mwhudson (~mwh@linaro/mwhudson) left irc: Quit: Leaving21:04
mattipIt seems to me that a view of view of a concrete should reduce to a view, the strides should just meld together.21:05
mattipSo a broadcast should just be a view into a concrete as well, just with wierd strides.21:05
mwhudson (~mwh@linaro/mwhudson) joined #pypy.21:06
mattipBut maybe I'm missing something, as usual.21:07
fijalyes, you're completely correct21:07
fijalbut we also could have a view into non-concrete21:08
fijaland I'm thinking how to implement this21:08
fijalif you want to implement a broadcast of an array or of a concrete view, go ahead21:08
fijaleverything is already there21:08
fijalAlex_Gaynor: did you see in the logs about test runs?21:08
arigato (~arigo@89.204.130.86) joined #pypy.21:09
fijalarigato: hi, I can explain what tests I had in mind :)21:13
Action: arigato only half there21:13
fijalok :021:14
fijalI can explain some other time21:14
voidspace (~anonymous@python/psf/voidspace) joined #pypy.21:18
lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) left irc: Quit: Leaving21:19
mattipjterrace: ping21:22
jterracepong21:22
mattipCan I add a failing test to concatenate? it returns dtype=float for int arrays21:22
jterraceoh sure21:23
mattipSo I'll commit some tests on the branch?21:23
jterracethat should be an easy fix21:23
jterraceyeah that sounds good21:23
mattipok.21:23
ixokai (~ixokai@rrcs-24-199-8-246.west.biz.rr.com) joined #pypy.21:24
ixokai (~ixokai@rrcs-24-199-8-246.west.biz.rr.com) left irc: Changing host21:24
ixokai (~ixokai@pdpc/supporter/student/ixokai) joined #pypy.21:24
jterracemattip: i think it's actually harder than i thought to fix21:26
jterracenumpy casts up21:26
jterraceso concatenating (int, float) arrays -> float result21:26
jterracehave to figure out how to implement that21:26
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.21:26
Action: mikefc hopes to gain enough knowledge to work on numpypy. 21:27
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Computer has gone to sleep.21:28
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.21:28
fijal (~fijal@197.172.87.224) left irc: Ping timeout: 276 seconds21:31
mikefcother things like 'work' and 'life' keep getting in my way.21:32
mattipjterrace: find_binop_result_dtype in interp_ufunc, but it only handles two dtypes, you can concatenate more than two arrays21:36
kenaan12mattip numpy-concatenate 11ce05e49e8eda 15/pypy/module/micronumpy/test/test_numarray.py: add failing test21:39
mikefc need to find a uni lecturer teaching linear algebra and set numpy.linalg as student assignments.  21:43
mikefcor, you know, just call the C library...21:44
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.21:45
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1470 [12amaury, py3k]21:51
fprimex (~fprimex@brent-macbook.sc.fsu.edu) left irc: Quit: http://www.fprimex.com21:53
kkris (~kris@80-123-35-252.adsl.highway.telekom.at) left irc: Quit: Leaving.21:56
jterracemattip: couldn't you just use find_binop_result_dtype along the arguments?21:58
jterracefor arrays a,b,c, run it with (a,b)->r and then (r,c)21:58
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Computer has gone to sleep.22:02
mattipyes, I think so. I need this too for frompyfunc so if you do it as a function I can reuse it.22:09
mattipfind_multiop_result22:10
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds22:15
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace22:16
nedbat (~nedbat@python/psf/nedbat) joined #pypy.22:24
ixokai (~ixokai@pdpc/supporter/student/ixokai) left irc: Quit: Bye!22:24
voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.22:25
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host22:25
voidspace (~anonymous@python/psf/voidspace) joined #pypy.22:25
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) joined #pypy.22:28
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Quit: Leaving22:32
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/74622:34
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/97922:34
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/17122:34
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/322:34
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds22:41
mattip (~chatzilla@bzq-79-183-120-145.red.bezeqint.net) left irc: Quit: ChatZilla 0.9.87 [Firefox 8.0/20111115183158]22:44
arigato (~arigo@89.204.130.86) left irc: Ping timeout: 252 seconds22:45
DanKluev (~root@78-24-224-073-udsl.vntc.ru) left irc: Ping timeout: 252 seconds22:45
legogris (~legogris@c80-217-207-157.bredband.comhem.se) left irc: Ping timeout: 252 seconds22:51
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) left irc: Remote host closed the connection22:58
whitelynx (~whitelynx@63.241.75.144) left irc: Quit: Ex-Chat22:59
amaury_>>>> import ast; e = ast.expr(); e.lineno = 5; print e.lineno23:02
amaury_AttributeError: 'expr' object has no attribute 'lineno'23:02
derdon (~derdon@p5DE8A908.dip.t-dialin.net) left irc: Remote host closed the connection23:08
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 252 seconds23:09
exarkunamaury_: Hi.  Were you hoping for me to apply the patch you supplied the other day?23:09
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Read error: Connection reset by peer23:09
PiotrSikora_ (~none@nginx/adept/piotrsikora) joined #pypy.23:09
amaury_exarkun: I don't think it was finished at the time23:10
exarkunAny interest in finishing it? :)23:10
amaury_not tonight, will go in bed soon23:10
exarkunI guess what I'm curious about is whether I should apply it to the branch and then find someone else to finish up the work, or if you're interested in eventually doing that23:11
exarkunI'm not in a terrible rush, but it would be nice to see some progress in this direction.23:11
exarkunAlternatively, I wouldn't mind talking to someone about alternate implementation strategies for SSL on PyPy (but not tonight).23:12
etrepum (~bob@accessnat4.mochimedia.net) left irc: Ping timeout: 240 seconds23:12
amaury_I think the patch was in the right direction23:12
amaury_if anything, it shows that support for buffers is incomplete in pypy23:13
pjenveyobligatory snark: that's not saying much, buffers are incomplete on cpython too!23:15
exarkunI could also offer a bounty for Twisted/SSL/PyPy support.  Open source bounties often far undervalue programmer labor though, and I'm not made of money so this one might do that too.23:17
exarkunAnyway, just wanted to get those things out there, now I gotta go.23:18
gehho2 (~gehho2@X190178.ppp.dion.ne.jp) joined #pypy.23:18
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!23:23
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace23:28
jerithAren't bounties supposed to be incentives and motivators rather than fair compensation for the work involved?23:29
nedbat (~nedbat@python/psf/nedbat) joined #pypy.23:30
etrepum (~bob@c-67-180-192-170.hsd1.ca.comcast.net) joined #pypy.23:39
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Ping timeout: 240 seconds23:41
legogris (~legogris@c80-217-207-157.bredband.comhem.se) joined #pypy.23:56
--- Fri Dec 16 201100:00

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