#pypy IRC log for Wednesday, 2012-01-04

stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.00:01
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer00:01
Nick change: stakkars_ -> stakkars00:01
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.00:02
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer00:02
Nick change: stakkars_ -> stakkars00:02
kennethreitz (~kennethre@static-71-166-241-209.washdc.east.verizon.net) left irc: Quit: Computer has gone to sleep.00:03
asmeurer___ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.00:04
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.00:09
bugfree (azhang@adsl-99-63-77-95.dsl.pltn13.sbcglobal.net) left #pypy.00:11
Goron (~LC@h171.163.191.173.dynamic.ip.windstream.net) joined #pypy.00:34
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 244 seconds00:36
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) joined #pypy.00:44
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds01:06
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.01:07
moreati (~alex@5ad0d588.bb.sky.com) left irc: Ping timeout: 240 seconds01:14
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) left irc: Ping timeout: 248 seconds01:15
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/2101:25
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/19301:28
nettok (~quassel@200.119.156.188) joined #pypy.01:31
rekamso (~textual@67.51.82.66) left irc: Quit: Computer has gone to sleep.01:32
derdon (~derdon@p5DE89A12.dip.t-dialin.net) left irc: Remote host closed the connection01:36
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/76501:48
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/191002:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/66302:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/32702:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/31602:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/66502:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/118102:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/84802:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/150002:00
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/77502:00
nedbat (~nedbat@python/psf/nedbat) joined #pypy.02:19
tilgovi (~randall@ool-4357ca60.dyn.optonline.net) joined #pypy.02:21
tilgovi (~randall@ool-4357ca60.dyn.optonline.net) left irc: Changing host02:21
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.02:21
fijal (~fijal@197.104.28.197) left irc: Read error: Connection reset by peer02:24
nettok (~quassel@200.119.156.188) left irc: Read error: Connection reset by peer02:26
Shanita (~John@osbk-4db14907.pool.mediaWays.net) joined #pypy.03:04
Kaskuka (~John@osbk-4db155a1.pool.mediaWays.net) left irc: Ping timeout: 240 seconds03:04
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/100203:09
ericflo (~ericflo@75.103.8.110) left irc: Quit: ericflo03:12
dracman (~draco@212.255.38.134) left irc: Ping timeout: 252 seconds03:18
mikefc3sigh03:20
mikefc3tis lonely in here during the day.03:20
gutworthit's not like this channel is for emotional support03:22
dracman (~draco@212.255.47.158) joined #pypy.03:25
mikefc3gutworth: module/emotional_support.py03:36
mikefc3i like my emotional support JITted...03:36
mikefc3i was just looking to slack off from work :)03:36
timotimoso basically a jitted eliza?03:47
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood03:56
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.03:58
gutworth (~benjamin@64-131-9-52.usfamily.net) left irc: Ping timeout: 240 seconds04:08
gutworth (~benjamin@64-131-9-52.usfamily.net) joined #pypy.04:10
bbot23Success: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/191004:12
Goron (LC@h171.163.191.173.dynamic.ip.windstream.net) left #pypy.04:16
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 252 seconds04:39
anish_ (anish@nat/redhat/x-hdpqlcauwalskbmr) joined #pypy.04:43
lmoura (lmoura@gateway/indt/session) left irc: Ping timeout: 244 seconds04:56
lmoura (lmoura@gateway/indt/session) joined #pypy.05:08
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/84805:16
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/150005:17
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.05:30
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer05:30
Nick change: stakkars_ -> stakkars05:30
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/118105:32
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/66505:36
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/66305:43
[dax!dax@freenode/staff/dax] away 06:04
[dax!dax@freenode/staff/dax] oops, you're not the bot I was looking for >.>06:04
anish_ (anish@nat/redhat/x-hdpqlcauwalskbmr) left irc: Ping timeout: 276 seconds06:06
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 240 seconds06:14
anish_ (anish@nat/redhat/x-qpkfujoualazpwhm) joined #pypy.06:15
mcdonc_ (~mcdonc@ip72-209-213-54.dc.dc.cox.net) left irc: Remote host closed the connection06:19
mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) joined #pypy.06:20
fermianyon (~lane@c-71-229-21-197.hsd1.al.comcast.net) left irc: Ping timeout: 240 seconds06:21
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/32706:43
bbot23Success: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/77506:44
gutworth (~benjamin@64-131-9-52.usfamily.net) left irc: Ping timeout: 252 seconds06:46
apollo13 (~apollo13@unaffiliated/apollo13) left irc: Ping timeout: 268 seconds06:51
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.07:02
apollo13 (~apollo13@unaffiliated/apollo13) joined #pypy.07:08
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.07:11
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer07:11
Nick change: stakkars_ -> stakkars07:11
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.07:16
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer07:16
Nick change: stakkars_ -> stakkars07:16
ousado (~ousado@unaffiliated/ousado) left irc: Remote host closed the connection07:20
ousado (~ousado@31-18-121-45-dynip.superkabel.de) joined #pypy.07:20
mat^2 (~mathias@87.55.0.198) left irc: Ping timeout: 240 seconds07:24
fijal (~fijal@41.49.91.195) joined #pypy.07:38
fijalhakanardo: ping07:40
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/31607:41
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) joined #pypy.07:44
fijal (~fijal@41.49.91.195) left irc: Read error: Connection reset by peer08:07
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.08:14
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds08:18
squiddy (~squiddy@g224192142.adsl.alicedsl.de) joined #pypy.08:23
mattip (~mattip@bzq-79-182-123-101.red.bezeqint.net) joined #pypy.08:31
fijal (~fijal@196-215-5-11.dynamic.isadsl.co.za) joined #pypy.08:31
fijalwuhu internet!08:31
mattiphi08:31
mattipADSL problems?08:32
fijalwell08:32
fijalit started working08:32
fijalnot necesarilly "problems"08:32
mattipnjoy. I'm jealous of people with good internet.08:33
fijal"good"08:33
fijalit's a 4M line08:33
fijalwhich costs tons of money08:33
mattipfijal: I was intrigued about the idea of applevel numpy08:39
mattipIs there a way to expose BaseArray to applevel code08:40
mikefcthere are lots of staticmethods that are easy to cut-and-paste from numpy into applevel numpypy08:40
mikefcbut i'll wait for everyone else to think about how best this is done :)08:41
mikefcmy .cross() was just a sample08:41
fijal (~fijal@196-215-5-11.dynamic.isadsl.co.za) left irc: Ping timeout: 240 seconds08:42
mattipBut cross() is not an attribue of BaseArray08:42
mikefcmattip: applevel instance methods would be a good way to get numpypy "complete" relatively quickly. any old git (like me) could add to numpypy08:43
mikefcmattip: i know that cross is not a basearray instance method.  I was referring to it as an example of a staticmethod (probably not the right term) of the numpypy module.08:43
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy.08:44
mattipI know, I am trying to work out if pypy supports adding attributes at applevel to interplevel classes08:44
mikefcmattip: I too would like applevel basearray methods. so I could add cumsum().  I think cumsum may be easy with your axisops now though?08:46
Action: mattip trying to figure out what cross does for 3d arrays or 2d arrays larger than 3x308:47
mattipdeterminant?08:47
mikefcmattip: it only seem to do the 1-d version. and if pass an N*3 or N*2 array, it just does N lots of the 1D cross product.08:48
mikefcmattip: alex and fijal had a brief word this morning (um.  12 hours ago) about pulling in the applevel stuff for staticmethods.  I'll have to wait until someone figures out how to do it properly before I try adding anything.08:50
mikefcthere must be a "proper" way to do it rather than my adhoc cut and paste for cross()08:50
mattip"when you point at someone else, four fingers are pointing at you"08:51
mikefcswapaxes() I could also write if we had applevel basearray methods.  but this too is probably something that works with an iterator/axisops at interp level.08:51
mikefc(not sure how interators/axisops work yet though).08:52
mikefcs08:52
mikefco 08:52
mikefcm08:52
mikefcuch08:52
mikefc 08:52
mikefcso much to learn in pypy08:52
mattipOnce I get done with axisops, I thought I'd finish multidim dot.08:52
mattipthen maybe swapaxis.08:52
mattipIs there a usecase for all this?08:53
mikefcwell, I'm prepared to go through numpy and pull in all the module level methods from numpy into numpypy at the applevel08:53
mattiplike a potential benchmark?08:53
mikefcmattip: for numpypy?08:53
mikefcoh08:53
mikefcum08:53
mikefci was only looking at cross() because someone filed it as a bug.08:53
mikefcand it seemed small enough for me to understand08:53
mattipbenchmarks / usecases provide motivation08:54
mattip"let's be faster (or use less memory) than X"08:54
mikefcmattip: I'm aiming for "complete enough for more than just me to be using it". Is that bad?08:55
mattipfijal reminded me that "good" and "bad" are relative about 15 minutes ago.08:55
Action: mikefc will think about a sexy usecase08:56
lesshastegood morning09:07
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy09:26
JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy.09:27
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 252 seconds09:28
hakanardofijal: pong09:32
mattipmikefc: you might want to take a look at the separate-applevel-numpy, it seems to be the direction you were proposing.09:38
mattipthe separate-applevel-numpy branch 09:39
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds09:41
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.09:42
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) joined #pypy.09:44
fijal (~fijal@41-133-63-76.dsl.mweb.co.za) joined #pypy.09:49
Action: mikefc is a bit slow and has only just found https://bitbucket.org/pypy/extradoc/src/extradoc/planning/micronumpy.txt09:50
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.09:51
fijalmikefc: might be a bit outdated09:52
mikefcmaybe. it's hard for me to tell.09:53
mikefcmattip: there doesn't seem to be much in separate-applevel-numpy09:53
fijalmattip: what's abc=kil line doing there?09:55
fijalno, we should recreate this branch probably09:55
ernop (~ernie@114.250.129.15) joined #pypy.09:55
hakanardofijal: repong09:56
fijalhakanardo: did you see assertionerror in import_from_state?09:56
fijalor is this fixed by now?09:57
hakanardowhen?09:57
fijalhttps://bugs.pypy.org/issue97509:57
fijal3 days ago apparently :)09:57
hakanardothere is no mention about how to reproduce it...09:59
fijalmattip: the reduce.compute is kinda broken09:59
fijalit has to use find_sig09:59
fijalI'll explain to you when you're online10:00
fijalhakanardo: no :(10:00
mattipfijal: I should have gone to bed before commiting debug cruft to hg10:02
fijalhaha :)10:02
fijalmattip: it's unmergeable as it is now :)10:02
mattipI tried using find_sig but got a recursive call to get_concrete10:02
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds10:02
fijalwhy does find_sig call get_concrete?10:02
fijalcalling create_sig would mean that you compile new assembler each time you call this function10:03
fijalnot really great10:03
Action: mattip thinking before speaking this time10:03
fijal:]10:03
fijalI should learn that10:03
fijalone day...10:03
fijalwould be cool to have test_zjit for that btw10:04
ousado (~ousado@31-18-121-45-dynip.superkabel.de) left irc: Changing host10:04
ousado (~ousado@unaffiliated/ousado) joined #pypy.10:04
fijalI have to look deeper later how this can be nicely done10:04
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.10:05
fijal (~fijal@41-133-63-76.dsl.mweb.co.za) left irc: Ping timeout: 260 seconds10:11
mikefcmattip/fijal: there's an easy .std() .var() patch for numpy waiting on the mailing list :)10:12
mikefcoops10:12
mikefcmissed him10:12
mattipfijal: http://paste.pocoo.org/show/529744/10:17
voidspace (~anonymous@python/psf/voidspace) joined #pypy.10:17
mattipCall2._invent_array_numbering calls arr.get_concrete() which calls arr.compute which calls arr.find_sig wich calls arr.ceate_sig(...,arr) which calls invent_array_numbering, which brings us back to the beginning.10:20
mattip:(10:21
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Remote host closed the connection10:23
kenaan12mattip numpypy-axisops 11e318c72388d9 15/pypy/module/micronumpy/interp_numarray.py: remove debug cruft10:24
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.10:25
mattip (~mattip@bzq-79-182-123-101.red.bezeqint.net) left irc: Quit: Leaving10:27
voidspace (~anonymous@python/psf/voidspace) left irc: Ping timeout: 240 seconds10:33
mfoord (~anonymous@python/psf/voidspace) joined #pypy.10:40
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds10:47
k_bx (~k_bx@195.20.130.1) joined #pypy.10:50
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.10:52
k_bxHello everyone! It's a little bit offtopic, but I thought maybe still be somehow relevant to pypy. Today, when map/reduce-like systems are widely used, there's clear need in serializing/deserializing functions with their dependencies. Today's solution is just a piece of text, which is then evaluated as JS-function on server. Is there a chance that pypy is suitable for giving opportunity to break that and give users a way to serialize/deserialize/transmit functions?10:52
k_bx(maybe simplified and with some restriction). Thanks.10:52
ronnyk_bx: how about just pushing the code to the remote side and calling entrypoints10:55
ronnyk_bx: execnet already can do it in a simplified form, and part of my diplome thesis will consist of having a more extended form of those entrypoints10:56
k_bxronny: you mean like keeping copies of code on all the database servers?10:56
voidspace (~voidspace@python/psf/voidspace) joined #pypy.10:56
voidspace (~voidspace@python/psf/voidspace) left irc: Client Quit10:57
Nick change: mfoord -> voidspace10:57
ernop (~ernie@114.250.129.15) left irc: Ping timeout: 240 seconds10:58
k_bxronny: there are currently two approaches: 1. keep remote code synchronized -- that's what celery does, for example. and it's confusing and non well in terms of server/workers restart on code update. 2. give function as string -- that's how map/reduces do these days -- because it's just simpler than synchronizing code. but it's such a pain for testing / writing that code.10:59
k_bxronny: I'll look at execnet, thanks11:00
ronnyk_bx: no, i mean having a import hook that loads the code from a pristine source instead of storing it everywhere11:00
k_bxronny: yes, the question is about "loads the code" part. how does it do that? transmits code as strings?11:04
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) joined #pypy.11:10
antocuniNO11:14
antocuniops :-)11:14
ronnyk_bx: how else would it do that, the code IS basically a string11:24
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) left irc: Ping timeout: 268 seconds11:44
hruske (~Gasper@2001:15c0:66e9:14:9c82:51ff:fee8:af82) joined #pypy.11:49
hruske (~Gasper@2001:15c0:66e9:14:9c82:51ff:fee8:af82) left irc: Client Quit11:49
Nick change: asmeurer___ -> asmeurer_11:50
Nick change: asmeurer_ -> asmeurer11:51
mattip (~mattip@bzq-79-182-123-101.red.bezeqint.net) joined #pypy.11:58
JaRoel (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 240 seconds12:00
mattip (~mattip@bzq-79-182-123-101.red.bezeqint.net) left irc: Quit: Leaving12:05
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.12:05
mattip (~mattip@bzq-79-182-123-101.red.bezeqint.net) joined #pypy.12:06
mattiphelp M12:12
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.12:13
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) joined #pypy.12:13
mattip_ (~mattip@bzq-109-65-107-23.red.bezeqint.net) joined #pypy.12:28
mattip (~mattip@bzq-79-182-123-101.red.bezeqint.net) left irc: Ping timeout: 252 seconds12:28
Dulak (~michael@unaffiliated/dulak) left irc: Read error: Connection reset by peer12:38
Dulak (~michael@unaffiliated/dulak) joined #pypy.12:38
mattip_ (~mattip@bzq-109-65-107-23.red.bezeqint.net) left irc: Quit: bye12:43
mattip_ (~mattip@109.65.107.23) joined #pypy.12:44
Nick change: mattip_ -> mattip12:45
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Remote host closed the connection12:54
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.12:59
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.12:59
kenaan12hager ppc-jit-backend 11634bd8357b6e 15/pypy/jit/backend/ppc/: (bivab, hager): adjusted code to meet the latest refactoring, made first test pass13:02
kenaan12hager ppc-jit-backend 11bd859a60caa8 15/pypy/jit/backend/ppc/ppcgen/ppc_assembler.py: renamed function13:13
k_bx (~k_bx@195.20.130.1) left irc: Ping timeout: 260 seconds13:13
k_bx (~k_bx@195.20.130.1) joined #pypy.13:15
kenaan12hager ppc-jit-backend 11f95845ede0d3 15/pypy/jit/backend/ppc/ppcgen/ppc_assembler.py: don't print debug output13:23
kenaan12hager ppc-jit-backend 1145e11554d5ad 15/pypy/jit/backend/: replace LoopToken with JitCellToken and kill unused functions from runner13:23
kenaan12hager ppc-jit-backend 11cc871ff2daed 15/pypy/jit/backend/ppc/ppcgen/ppc_assembler.py: merge13:23
asmeurer (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) left irc: Quit: asmeurer13:34
k_bx (~k_bx@195.20.130.1) left irc: Ping timeout: 252 seconds13:46
k_bx (~k_bx@195.20.130.1) joined #pypy.13:47
k_bx (~k_bx@195.20.130.1) left irc: Read error: Connection reset by peer13:47
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds13:53
mattip (mattip@109.65.107.23) left #pypy ("bye").13:58
kenaan12hager ppc-jit-backend 1196c640b286a3 15/pypy/jit/backend/ppc/ppcgen/: (bivab, hager): adjust implementation of JUMP, implement LABEL13:59
kenaan12hager ppc-jit-backend 11fe53dc8f2aea 15/pypy/jit/backend/ppc/ppcgen/ppc_assembler.py: add some methods and fix little bugs13:59
mikefc3 (9665bcc5@gateway/web/freenode/ip.150.101.188.197) left irc: Ping timeout: 258 seconds14:00
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds14:07
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.14:11
kenaan12hager ppc-jit-backend 11e719f82117db 15/pypy/jit/backend/test/runner_test.py: (bivab, hager): fix test14:15
mat^2 (~mathias@212.130.113.35) joined #pypy.14:16
anish_ (anish@nat/redhat/x-qpkfujoualazpwhm) left irc: Quit: Leaving14:34
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds14:36
jnoller (~jnoller@vpn.nasuni.com) joined #pypy.14:37
whitelynx (~whitelynx@63.241.75.144) joined #pypy.14:44
anish_ (anish@nat/redhat/x-fmxrfyvskxvhuoou) joined #pypy.14:44
fprimex (~fprimex@brent-macbook.sc.fsu.edu) joined #pypy.14:46
kenaan12hakanardo jit-usable_retrace_2 115bf961464624 15/pypy/: debugging14:48
kenaan12hakanardo default 11656cfd21d520 15/pypy/jit/metainterp/optimizeopt/test/test_multilabel.py: passing test14:48
kenaan12hakanardo default 1183a92dcf51e3 15/pypy/jit/metainterp/optimizeopt/unroll.py: some more debug prints14:48
kenaan12hakanardo default 115afb4fd1f372 15/pypy/: hg merge14:48
kenaan12hakanardo jit-usable_retrace_2 11ca3d9cd45ea5 15/pypy/: hg merge default14:48
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.14:49
CIA-5303hakanardo 07roundup * 10#975/A jit bug crashes my app: 14:52
CIA-53Is there any way we could reproduce the bug? I've added some more debug prints to14:52
CIA-53shed light on whats happening here. Maybe you could reproduce ... * 14https://bugs.pypy.org/issue97514:52
nedbat (~nedbat@python/psf/nedbat) joined #pypy.15:00
Rhy0lite (dje@nat/ibm/x-sgkneqngmrwovwqr) joined #pypy.15:01
gutworth (~benjamin@64-131-9-52.usfamily.net) joined #pypy.15:01
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.15:03
dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) left irc: Quit: Leaving15:06
bogner (~bogner@li325-42.members.linode.com) left irc: Ping timeout: 240 seconds15:09
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Read error: Operation timed out15:12
anish_ (anish@nat/redhat/x-fmxrfyvskxvhuoou) left irc: Quit: Leaving15:14
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds15:33
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.15:34
bogner (~bogner@2600:3c03::f03c:91ff:fedf:7ef4) joined #pypy.15:36
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds15:47
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.15:53
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds15:53
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy.15:56
rguillebert (~rguillebe@fec76-1-78-234-124-105.fbx.proxad.net) left irc: Ping timeout: 240 seconds16:04
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.16:08
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 252 seconds16:11
rguillebert (~rguillebe@2a01:e34:eea7:c690:21f:c6ff:fe12:4dee) joined #pypy.16:18
andrewfr (~andrew@modemcable211.101-70-69.static.videotron.ca) joined #pypy.16:26
andrewfr (andrew@modemcable211.101-70-69.static.videotron.ca) left #pypy.16:27
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) left irc: Ping timeout: 240 seconds16:31
CIA-5303SimonHova 07roundup * 10#979/Pickle fails in pypy, works in CPython: 16:32
CIA-53[new] When trying to use the shelve module, I noticed that it kept hanging while16:32
CIA-53saving a rather large (~5 MB) object. I followed the source cod ... * 14https://bugs.pypy.org/issue97916:32
kkris (~kris@80-123-40-226.adsl.highway.telekom.at) joined #pypy.16:36
bgola (~bgola@189.100.41.65) joined #pypy.16:41
dmalcolm (david@nat/redhat/x-zoubchatvwgwopkg) joined #pypy.16:43
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.16:49
kenaan12hager ppc-jit-backend 1151d8610b324a 15/pypy/jit/backend/ppc/ppcgen/: (bivab, hager): implement bridges16:51
antocuni_ (~antocuni@host208-85-dynamic.7-79-r.retail.telecomitalia.it) joined #pypy.16:57
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds16:59
Shanita (~John@osbk-4db14907.pool.mediaWays.net) left irc: Read error: Connection reset by peer17:01
Kaskuka (~John@osbk-4db14907.pool.mediaWays.net) joined #pypy.17:04
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection17:10
G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving.17:15
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) left irc: Read error: Operation timed out17:16
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy.17:18
fermianyon (~lane@c-71-229-21-197.hsd1.al.comcast.net) joined #pypy.17:19
durin42 (~durin@adium/durin42) left irc: Ping timeout: 240 seconds17:21
durin42 (~durin@173-15-14-174-Illinois.hfc.comcastbusiness.net) joined #pypy.17:23
durin42 (~durin@173-15-14-174-Illinois.hfc.comcastbusiness.net) left irc: Changing host17:23
durin42 (~durin@adium/durin42) joined #pypy.17:23
kkris1 (~kris@188-22-117-228.adsl.highway.telekom.at) joined #pypy.17:29
kkris (~kris@80-123-40-226.adsl.highway.telekom.at) left irc: Ping timeout: 240 seconds17:30
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Ping timeout: 276 seconds17:33
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.17:35
fijal (~fijal@196-215-5-11.dynamic.isadsl.co.za) joined #pypy.17:37
zain (~textual@cpe-68-173-70-124.nyc.res.rr.com) joined #pypy.17:44
Niedar (~nnscript@ip68-231-248-182.hr.hr.cox.net) left irc: Read error: Connection reset by peer17:50
Niedar (~nnscript@ip68-231-248-182.hr.hr.cox.net) joined #pypy.17:50
fijalhey17:52
JaRoel|4d (~jaroel|4d@2001:980:7133:1:5da:b012:85b1:4c1b) joined #pypy.17:54
CIA-5303fijal 07roundup * 10#979/Pickle fails in pypy, works in CPython: 17:54
CIA-53[chatting] PyPy does not have cPickle, only pickle, so it's likely it's just a bug in pickle17:54
CIA-53or it's slow. Can you provide an example, preferabl ... * 14https://bugs.pypy.org/issue97917:54
dmalcolm (david@nat/redhat/x-zoubchatvwgwopkg) left irc: Read error: Connection reset by peer18:03
dmalcolm (david@nat/redhat/x-mxcyzpicwkqvhntz) joined #pypy.18:03
lmoura (lmoura@gateway/indt/session) left irc: Ping timeout: 240 seconds18:07
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds18:08
JaRoel|4d (~jaroel|4d@2001:980:7133:1:5da:b012:85b1:4c1b) left irc: Remote host closed the connection18:09
lmoura (lmoura@gateway/indt/session) joined #pypy.18:09
CIA-5303SimonHova 07roundup * 10#979/Pickle fails in pypy, works in CPython: 18:14
CIA-53I am pickling a list of classes, and it only seems to choke on the big ones. I'll18:14
CIA-53try to create a simple example for you now. * 14https://bugs.pypy.org/issue97918:14
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace18:19
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy18:22
durin42 (~durin@adium/durin42) left irc: Ping timeout: 268 seconds18:22
durin42 (~durin@173-15-14-174-Illinois.hfc.comcastbusiness.net) joined #pypy.18:24
durin42 (~durin@173-15-14-174-Illinois.hfc.comcastbusiness.net) left irc: Changing host18:24
durin42 (~durin@adium/durin42) joined #pypy.18:24
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.18:24
jnoller (~jnoller@vpn.nasuni.com) left irc: Quit: Leaving...18:24
antocuni_ (~antocuni@host208-85-dynamic.7-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds18:24
zain (~textual@cpe-68-173-70-124.nyc.res.rr.com) left irc: Quit: Computer has gone to sleep.18:32
Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Ping timeout: 240 seconds18:33
CIA-5303SimonHova 07roundup * 10#979/Pickle fails in pypy, works in CPython: 18:33
CIA-53I don't know how to submit my code without the problematic file, so I am going to18:33
CIA-53dump it all.18:33
CIA-53Test program attached. * 14https://bugs.pypy.org/issue97918:33
CIA-5303SimonHova 07roundup * 10#979/Pickle fails in pypy, works in CPython: EDI module attached. * 14https://bugs.pypy.org/issue97918:34
mtigas (~mtigas@users.spokesman.com) joined #pypy.18:35
CIA-5303SimonHova 07roundup * 10#979/Pickle fails in pypy, works in CPython: EDI File parser. * 14https://bugs.pypy.org/issue97918:35
CIA-5303SimonHova 07roundup * 10#979/Pickle fails in pypy, works in CPython: 18:37
CIA-53I'm not really certain how to show the problem without all the code, so I am18:37
CIA-53including it.18:37
CIA-53Basically, my EDI module makes a mutable dict obje ... * 14https://bugs.pypy.org/issue97918:37
setmeaway (stemearay@119.201.52.190) left irc: Read error: Connection reset by peer18:42
setmeaway (setmeaway3@119.201.52.190) joined #pypy.18:43
rekamso (~textual@63-228-127-70.dia.static.qwest.net) joined #pypy.18:46
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds18:47
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) joined #pypy.18:49
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.18:51
jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Read error: Connection reset by peer18:51
jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.18:53
lesshastehi all18:59
lesshastefijal: ping19:00
lesshastefijal: I made a nice example of what we talked about following your suggestion19:00
lesshastewhich is quite convincing19:00
fijallesshaste: pong19:00
lesshastewhat's a nice pastebin?19:00
lesshasteI'll put the examples there19:00
ronnypaste.pocoo.org?19:02
lesshastethanks19:03
lesshastefijal: ok.. so example one http://paste.pocoo.org/show/529979/19:03
fijaland?19:03
ericflo (~ericflo@75.103.8.110) joined #pypy.19:04
lesshasteexample two http://paste.pocoo.org/show/529981/19:04
lesshastefijal: the first takes 56 seconds and the second 1919:04
lesshasteyou will notice that all I did was unroll the loop 19:04
lesshaste(exactly as you suggested)19:04
fijaloh cool19:04
lesshasteit's a completely trivial unrolling19:04
lesshastethe loop iterates 5 times19:05
fijalhow much down did you want to go btw?19:05
lesshasteas in how much faster did I want it to go?19:05
fijalyes19:05
lesshaste:) is there any limit 19:05
lesshasteI would be happy with this 3 times speedup for the moment :)19:05
fijalbtw19:06
fijalthis is an optimization that gcc couldn't get to the point of including it in the default for the past 20 years19:07
fijalor so19:07
lesshastesadly I don't know the internals of pypy well enough not to make stupid suggestions.. but.. :) .. I would guess there are two obvious and possibly wrong solutions. a) unroll loops a bit b) change the interpreter to annotate constant sized loops usefully19:07
fijal2519:07
lesshaste:) standing on the shoulders of giants?19:07
fijalwell, no, I just mean it's complex19:07
fijaland gcc pace is not particularly great :)19:08
lesshasteah yes :)19:08
fijalwe're better off, because we know runtime data19:08
lesshasteI still remember the egcs debacle19:08
fijalor at least we can know19:08
fijalyes, we don't have such problems19:08
fijallesshaste: I would suggest something smarter, have a counter on loops that have very few runs19:08
lesshastefijal: sounds better all round19:08
fijaland unroll them if so19:09
lesshasteyou mean count how many times the loop itself is called?19:09
fijalwe do that for debugging as well19:09
fijalbut I want a counter at a bridge that ends up with finish()19:09
fijallesshaste: feel like creating an issue with that btw?19:09
fijalwith both pastes and some summary19:10
Rhy0litefijal: it's not always a win19:11
fijalRhy0lite: of course :)19:11
fijalthat's why I'm saying it's complex19:11
fijalthat might be a cool thing where runtime-info gives you enough to make it always a win19:11
fijalor at least to come up with a useful heuristic19:11
fijalRhy0lite: note that in case of pypy, the win is bigger because you don't have to create frame19:11
fijalanother option would be to get rid of virtualizables alltogether19:12
CIA-5303fijal 07roundup * 10#976/Implement numpypy.cross(): 19:13
CIA-53[wontfix] I'm closing this issue as numpy.cross is implemented in pure python. We should19:13
CIA-53just reuse it. The issue describing "reuse pure python ... * 14https://bugs.pypy.org/issue97619:13
lesshastefijal: err. in the issue tracker I can add a file19:14
lesshastefijal: how can I add two files? :)19:14
fijaladd one and then add another one19:14
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.19:14
CIA-5303lesshaste 07roundup * 10#980/Loop with small number of iterations much faster after unrolling: 19:16
CIA-53[new] In the following examples, the function hamdistX is called many times. If it is19:16
CIA-53implemented as a simple loop then it is 3 times slower than ... * 14https://bugs.pypy.org/issue98019:16
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) left irc: Ping timeout: 252 seconds19:16
lesshasteok19:17
lesshasteit's a curious issue tracker19:17
lesshastewhat happened to bugzilla :)19:17
lesshastehttps://bugs.pypy.org/issue98019:18
CIA-5303lesshaste 07roundup * 10#980/Loop with small number of iterations much faster after unrolling: [chatting] Timing on my PC are 56 seconds for testham.py and 19 seconds for testham-unrolled.py * 14https://bugs.pypy.org/issue98019:19
lesshastefijal: so there is no point getting the interpreter to annotate the source?19:19
fijalof course not19:20
fijalbut there are hints in the jit19:20
fijalso jit knows if stuff is constant or not19:20
fijal(even if it's not a compile-time constant)19:20
lesshasteI think hint ==annotate to me19:20
lesshastebut it seems the jit doesn't even know if something is a loop or not19:21
lesshasteas it is trace-based 19:21
lesshasteor is that wrong?19:21
Action: lesshaste suspects it is19:21
fijaljit does not know anything about python in the first place :)19:23
fijalbut anyway19:23
fijalconstantness is few levels down19:23
lesshasteok19:24
mattip (~chatzilla@109.65.107.23) joined #pypy.19:24
lesshastefijal: if I need to improve the issue report just let me know19:24
fijalmattip: hi19:24
mattiphi. I changed it (push coming)19:25
kenaan12mattip numpypy-axisops 11ca13cff50c3a 15/pypy/module/micronumpy/interp_numarray.py: change create_sig to find_sig19:25
fijalthere is more :)19:26
fijalwait a sec, I'm in the middle of a commit19:27
fijalcan you explain the loop in the meantime19:27
fijalwhat is it doing?19:27
fijalwhy two loops, why different chunks etc.19:27
mattipOK. Say you want to sum along the columns. So the outer loop sets up a frame (using chunks) at the top of a column19:28
mattipThen frame.next to get the values along the column and sum them up in an inner loop.19:29
fijalso you create a new frame per what?19:29
mattipper column19:29
fijalthat might be prohibitevly expensive19:29
mattipis that expensive?19:29
mattipoh. :(19:30
kkris (~kkris@v2201109100636289.yourvserver.net) joined #pypy.19:30
mattipI wish there was a way for mortals to benchmark, I don't know when I am doing expensive stuff.19:30
kkris1 (~kris@188-22-117-228.adsl.highway.telekom.at) left irc: Quit: Leaving.19:30
fijalwell, there is19:31
fijaleven two :)19:31
kenaan12fijal better-jit-hooks 1188295e485a01 15/pypy/jit/metainterp/warmstate.py: remove some code that's not necessary any more - in-progress19:31
kenaan12fijal numpypy-axisops 11e38fface989e 15/pypy/module/micronumpy/interp_numarray.py: remove cruft19:31
fijalmight be potentially ok19:32
mattipand they are...?19:33
fijalcompile the entire interpreter19:34
fijalor add stuff to minilanguage and compile the mini interpreter19:34
aleksi (~aleksi@109-184-97-156.dynamic.mts-nn.ru) joined #pypy.19:35
fijalmattip: my question is19:35
kkris (~kkris@v2201109100636289.yourvserver.net) left irc: Quit: leaving19:36
kkris (~kkris@v2201109100636289.yourvserver.net) joined #pypy.19:36
fijalcan't you flip it outside out?19:36
fijalright now in a patological case you will have a bunch of allocations and loop operations per column19:37
fijalif column is say 10 elements19:37
fijalyou would do tons and tons of operations per each 10 elements19:37
mattipBut then I need some kind of signal when I have finished a column in order to restart value, increment output iterator.19:39
fijalmattip: with me so far?19:39
fijalyes19:39
fijalbut you can achieve that with just a single iterator19:39
fijalit's how it's done for example for multi-dimensional arrays no?19:40
mattipI thought I was creating the first iterator by dimension,19:41
mattipbroadcast just repeats a iterator-over-all-values19:42
mattipor are you talking about the internals of the ViewIterator with strides and whatnot?19:43
fijalyes, that's what I'm talking about19:43
fijalIMO you should not create two iterators19:43
fijalframe & signature should take care of it19:43
fijalso you might create a different frame that has a special .next for example19:43
fijalor so19:43
mattipyes, that is bad zen.19:43
fijalbut I think you should just have a thing like broadcast iterator19:44
fijaland call next always19:44
lesshastefijal: slightly surprisingly.. I just profiled the unrolled version and it still spends most of its time in the hamdist3 function19:44
fijalwhich would not do anything most of the time19:44
fijallesshaste: cProfile might lie a bit19:44
lesshasteah :)19:45
mattipOK, now that I have the class and the compute function, it is easy to play with.19:45
lesshasteit's still rather slow now I look at it 35,853,0922 calls of hamdist3 is basically 100 million comparisons19:45
lesshastewhich one might think would be done in under a second :)19:46
lesshasteoh err.19:46
lesshastethose figures are wrong19:46
mattipfijal: I will try again and get back to you. Thanks for the review.19:46
lesshaste175 million comparisons and 175 million +1 operations19:47
lesshastestill reckon it should take under a second :)19:47
Action: lesshaste is joking...sloghtly19:47
lesshasteslightly19:47
fijalmattip: np :)19:48
fijalmattip: sorry it took so long19:49
mattipfijal: could you explain "compile the mini interpreter"?19:49
fijallesshaste: you can look at traces19:49
lesshastefijal: how do I do that?19:49
fijallesshaste: run jitviewer and see how many asm instructions are there19:49
fijallesshaste: google jitviewer, follow instructions19:49
fijalmattip: you know test_zjit and test_compile right?19:49
fijalyou add stuff to test_compile, so the mini-interpreter knows about this19:50
fijaland then you create it19:50
lesshasteshould do this first?19:51
lesshastehg clone https://bitbucket.org/pypy/jitviewer19:51
fijalyes :)19:53
fijaland read the README :)19:53
lesshasteI am trying to follow the instructions.. the first line says 19:54
lesshastemkvirtualenv --python=/path/to/pypy pypy-viewer19:54
lesshasteis that really needed?19:54
lesshasteit doesn't exist on ubuntu19:54
fijalit does19:55
fijalvirtualenvwrapper19:55
fijalyou can create virtualenv by hand19:55
fijalit's highly recommended19:55
fijalotherwise you deal on your own with dependencies :)19:55
mattiptest_compile can give performance info as well as go/nogo asserts?19:55
lesshasteok thanks19:56
fijalmattip: you can compile this language19:56
fijalit probably needs some extra development to read from files and whatnot19:56
fijalcheck translator/goal/targetnumpystandalone.py19:57
fijalyou can change it to accept a filename for example19:57
fijalin that language19:57
fijaland then you can compile it and measure perf19:57
mattipOK, now I'm seeing the big picture. Thanks.19:58
mattiptest_zjit seems to be beyond me yet.19:58
fijaltest_zjit makes sense only if you want to test against a particular number of instructions19:59
fijal"does this assembler look right?"19:59
mattipI could get it to tell me how many instructions a run of a test using a particular implementation took20:00
mattipand then run again with another implementation20:00
fijalyes20:00
fijal--viewloops20:00
fijalwill show you loops20:00
fijalAlex_Gaynor: ping20:03
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.20:07
mattipfijal: thanks.20:09
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:12
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:12
Nick change: stakkars_ -> stakkars20:12
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:14
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:14
Nick change: stakkars_ -> stakkars20:14
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:14
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 240 seconds20:14
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:14
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:15
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:15
Nick change: stakkars_ -> stakkars20:15
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:16
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:16
Nick change: stakkars_ -> stakkars20:16
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:17
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:17
Nick change: stakkars_ -> stakkars20:17
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:18
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:18
Nick change: stakkars_ -> stakkars20:18
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.20:18
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:19
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:19
Nick change: stakkars_ -> stakkars20:19
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:20
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:20
Nick change: stakkars_ -> stakkars20:20
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:21
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:21
Nick change: stakkars_ -> stakkars20:21
fijalstakkars: what's up with your internet?20:21
fijalamaury_: hi20:21
amaury_Hi20:22
amaury_and Happy New Year to everybody!20:22
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:24
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:24
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:24
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:24
stakkars_ (~tismer@p5DDB7A14.dip.t-dialin.net) joined #pypy.20:26
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:26
Nick change: stakkars_ -> stakkars20:26
zain (~textual@cpe-68-173-70-124.nyc.res.rr.com) joined #pypy.20:26
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.20:39
mattiptargetnumpystandalone seems to have developed bitrot :(20:40
fijalmattip: yes20:41
fijalmattip: it requires a complete overhaul20:41
fijala bit intentionally left out20:41
Taggnostr (~quassel@host224-64-dynamic.116-80-r.retail.telecomitalia.it) left irc: Remote host closed the connection20:42
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds20:43
mattipI'll try to understand the zjit route with --viewloops. Much more potent magic.20:43
fijalboth are kinda necessary20:43
Rhy0lite (dje@nat/ibm/x-sgkneqngmrwovwqr) left irc: Quit: Leaving20:44
fijalzjit is what you want at first20:44
fijalanyway, your loop lacks jitdriver20:44
fijalI suggest running some other tests with --viewloops20:44
mattipdoing that now, pygame gui is nice.20:44
mattip (chatzilla@109.65.107.23) left #pypy ("g'nite").20:46
Taggnostr (~quassel@host224-64-dynamic.116-80-r.retail.telecomitalia.it) joined #pypy.20:49
fermianyon (~lane@c-71-229-21-197.hsd1.al.comcast.net) left irc: Ping timeout: 240 seconds21:10
zain (~textual@cpe-68-173-70-124.nyc.res.rr.com) left irc: Quit: Computer has gone to sleep.21:15
CIA-5303cheecheeo 07roundup * 10#974/channel is not iterable: (log message trimmed)21:17
CIA-53You bring up a good point. In my use case, what I'm doing is:21:17
CIA-53while True:21:17
CIA-53chan.recieve()21:17
CIA-53... * 14https://bugs.pypy.org/issue97421:17
fijalmikefc: are you mike c on pypy-dev?21:20
mikefcfijal: yes. I just posted a patch for a bunch of applevel module functions that only call class level methods and return.  relatively trivial.21:25
fijalokey21:25
mikefcand there's also a var/std() patch from last week.21:25
fijalnow the question is - why don't we just import numpy's pure python part?21:26
mikefcwe could, but21:27
mikefc(a) we're missing stuff. like swapaxes() and all the actual methods we need to call.21:27
CIA-5303fijal 07roundup * 10#961/Spurious (?) mercurial error messages from sandlib.py: [resolved] Closing the bug, seems to fix the problem enough for people not to care :) * 14https://bugs.pypy.org/issue96121:27
mikefc(b) there's some module boilerplate which we don't need?21:27
mikefcI really don't know the best way to go about it, which is why I made the patch so at least there's something to talk about :)21:28
CIA-5303fijal 07roundup * 10#959/numpypy array operation improperly jitted away: 21:29
CIA-53[resolved] There were other issues fixed in the meantime. I'll just close it unless someone21:29
CIA-53wants to pursure * 14https://bugs.pypy.org/issue95921:29
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.21:29
mikefc(c) there's really not much to this particular python code from numpy. all it does is 5 lines per function which say "if it's not already an ndarray then cast it. then just call the method"21:30
fijalok21:31
fijalmikefc: how about you create a branch that renames numpypy to _numpypy21:31
fijaland creates lib_pypy/numpypy21:32
fijalwhich will contain most of the import?21:32
fijalit can also contain imported tests21:32
mikefc(d) not all our BaseArray methods have the same signature. i.e. no 'axis' for most things as yet.21:32
fijalright21:32
fijalideally, we would ask axis everywhere that would raise NotImplementedError21:33
fijalpossibly to run vanilla version numpy21:33
mikefcfijal: for this patch that I've just sent, this is stuff that really is going to have to be in the module at the applevel.  It's like the max() function in app_numpy.py at the moment - all I'm doing is making sure that there is a place for a module function that calls the ndarray method.21:35
fijalI know and I'm trying to trick you to do more work ;-)21:36
mikefc:P21:36
fijaldo you want to give it a go?21:36
fijalI'm fine if you say "no"21:36
mikefcfijal: It's going to have to be 'no'.  I only have time to work on very small things at a time. and I just don't know  pypy/numpypy well enough at the moment to attempt anything major.21:37
fijalok, fair enough21:38
fijalfeel like uploading the patch to issue tracker?21:38
fijalI don't know the list moderation password21:38
mikefcok21:38
fijalso it won't make it (it's too big)21:38
CIA-5303mikefc 07roundup * 10#981/micronumpy 'fromnumeric' patch: 21:40
CIA-53[new] Hi All,21:40
CIA-53Numpy is somewhat derived from another array library called 'Numeric'. What were originally21:40
CIA-53functions in 'Numeric' have been p ... * 14https://bugs.pypy.org/issue98121:40
Action: mikefc also needs to work out hg21:40
fijalmikefc: ok, you can at least ping me if I forget about some issues21:42
mikefcshould I move my std()/var() patch onto the issue tracker as well?21:42
fijalyes please21:43
lahwranI need to figure out how to clone pypy with git, so that I can actually contribute21:43
fijallahwran: please use hg21:43
fijalsorry21:43
lahwranwhy?21:43
lahwrando you actually have a reason? :P21:43
fijalyes21:44
fijalautomatic tools tend to have problems21:44
lahwranah21:44
mikefcfijal: my cross() patch was just the pure python from numpy.  What do you want done differently?  do you want that lib_pypy/numpypy implemented?21:45
mikefc(i just saw the closed issue)21:45
fijalmikefc: yes, I want this stuff to be relatively vanilla copies21:46
fijalmostly because we'll have to redo it again when we support say axis arg21:46
mikefcwell it was, but it was just one function out of the source.  and to support the axis arg, you just had to uncomment the lines which were commented out 21:46
CIA-5303mikefc 07roundup * 10#982/micronumpy std()/var() patch: 21:47
CIA-53[new] Find attached a patch (to default branch) to add var and std to micronumpy21:47
CIA-53BaseArray.21:47
CIA-53Two simple tests were added, and all the micronum ... * 14https://bugs.pypy.org/issue98221:47
fijalmikefc: yes, but I mean if we start merging files into just app_numpy.py it'll get lost really quickly21:47
fijallet me do something quick21:48
fijalalthough I should sleep21:48
mikefcfijal: yeah, i realise there should be a better/nicer way of doing it.21:48
fijalyeah ok21:48
fijalI'll do it tomorrow21:48
fijalmikefc: how about I try to import numpy tomorrow and we'll see how it goes?21:49
mikefcok.  I think i see what you mean with lib_pypy/numpy but I'm not sure how to do it.21:50
mikefcyou were just going to copy the numpy/core directory over, but use our numpypy/BaseArray?21:50
fijalessentially current thing is a mixedmodule21:51
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds21:51
fijalwhich is an equivalent of C-implemented module in CPython21:51
fijalwhat it does on cpython, there is a python package that imports some C stuff21:52
fijalso what we'll do, we'll have the same python package that will import _numpypy (or whatever name we please)21:52
fijalthat contains things that are implemented in original numpy in C21:52
fijalwith me so far?21:52
fijalmikefc: your patch did not attach to the issue21:54
CIA-5303mikefc 07roundup * 10#982/micronumpy std()/var() patch: [chatting] patch attached. * 14https://bugs.pypy.org/issue98221:54
mikefcso instead of python/c. we'll be python/interp?21:54
fijalyes21:55
mikefci sort of understand, but I can't see how how it'll be laid out so that we're still using our BaseArray.  I might try and have a play with it this morning before breakfast.21:56
mikefcanother problem is that my machine is dreadfully slow.21:56
fijalwhy is this a problem?21:57
fijalso essentially you lay it down as I described21:57
fijalright now import numpypy imports a builtin module21:57
fijalyou can rename it to _numpypy (there is name in __init__.py)21:57
fijaland then put stuff in lib_pypy/numpypy/21:58
fijalthat would import _numpypy21:58
mikefcslow machine is like slow internet - it just makes everything a little more difficult.21:58
fijalah ok21:58
fijalmikefc: what's your timezone btw?21:58
lahwrannew machines for everyone!21:58
mikefcgmt+10.  I'm in australia.21:58
lahwranjust as long as someone else is paying.21:58
Action: mikefc has never done module/package related stuff in python before. Learning as I go...22:00
fijalmikefc: fellow inhabitant of the other side :)22:00
fprimex (~fprimex@brent-macbook.sc.fsu.edu) left irc: Quit: http://www.fprimex.com22:01
mikefc:)22:01
fijalmikefc: cool, thanks for your contributions22:01
fijaland please don't be annoyed if we don't accept it right away :)22:02
fijalI have a pretty hectic week22:02
mikefcthat's ok.  Two days after I submit something I realise there's a better way to do it.  I just get impatient and want more things in numpy straight away. but good design/code reuse counts.  I'll try to get numpy python stuff done nicely.  But don't hold your breath :)22:03
fijalok22:03
CIA-5303peterjc 07roundup * 10#981/micronumpy 'fromnumeric' patch: 22:03
CIA-53[chatting] This patch looks like it will address some of the missing NumPy functionality22:03
CIA-53reported in issue 913 * 14https://bugs.pypy.org/issue98122:03
fijalmikefc: seems people like your patches anyway22:04
fijalmikefc: I'm a bit sorry, but we're not in a mood to "get as much functionality as we can really quickly"22:04
mikefcfijal: what pterjc doesn't realise is that all his functions are NotImplemented yet :)22:05
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy22:05
mikefcfijal: I understand that you guys are all working on the guts of numpy.  I want more of the framework in place so that us mere mortals have code to work on.22:06
mikefccurrently it's pretty hard for an outsider (who is interested in numpy) to contribute to numpypy.  it's hard to see where things should go.22:08
fijalyou mean on the interp-level or which part?22:08
JaRoel|4d (~jaroel|4d@2001:980:7133:1:15c:aba2:a04c:4a8e) joined #pypy.22:09
mikefcfijal: in general. for example I was trying to figure out where to put "cumsum".  I could write one at the applevel. but there's no way to get applevel code into basearray.  but it took me an afternoon of farting around to find that out.   It's also hard to figure out what we don't have in basearray just yet - or what is needed to get some other stuff working.  For instance I thought cross() would be trivial, but it relies on swapaxes() whic22:10
mikefcdon't have.22:10
mikefcso all my patchwork is trying to explore what we have and don't have and how I can leave empty spaces in my patches which say "code goes here once someone figures out how to do it"22:11
fijalyou can do dir(array)22:11
mikefcsorry. it's hard for me to describe what I want to do.  I want it to be easier for people to see what is/isn't implemented *and* where that code may go.22:12
derdon (~derdon@p5DE8A9FE.dip.t-dialin.net) joined #pypy.22:13
fijalmikefc: you can ask :)22:13
mikefcand i realise that isn't what your/mattip/etc priorities lie.22:13
fijalanyway22:13
fijalit is my priority22:13
fijalit's just a bit hard to do22:13
fijalI'll try to document it a bit22:13
fijalok, I have to sleep22:14
mikefcI'm just trying to flesh numpypy out a bit more22:14
fijalmikefc: thanks for feedback, I'll give some docs22:14
fijalgive a try at some docs22:14
mikefcif it is just with docstrings and 'NotImplemented"s22:14
mikefcfijal: go to bed :)  gnight22:14
Action: mikefc preps for work22:14
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 240 seconds22:15
nedbat (~nedbat@python/psf/nedbat) left irc: Disconnected by services22:16
nedbat_ (~nedbat@python/psf/nedbat) joined #pypy.22:16
lesshastehow do I get jitviewer to know where pypy is?22:19
gutworthyou run it with pypy22:20
lesshasteI think I need python-virtualenv22:21
gutworthor do you mean the pypy tree?22:22
lesshasteargh... I am trying to do what it says at https://bitbucket.org/pypy/jitviewer22:22
lesshaste~/Desktop/jitviewer/bin/jitviewer.py log.pypylog 22:23
lesshaste/usr/bin/env: pypy: No such file or directory22:23
lesshasteso I tried virtualenv --python=/home/user/Desktop/pypy-1.7/bin/pypy pypy-viewer22:23
lesshastebut that fails too22:23
lesshasteecho $PYTHONPATH22:24
lesshaste/home/user/Desktop/pypy-1.7/bin/22:24
lesshastewhat should I do?22:24
gutworthfails how?22:24
lesshastehttp://paste.pocoo.org/show/530081/22:25
lesshasteany ideas?22:27
squiddywhat version of virtualenv are you using? it has to be at least 1.6.122:27
squiddyat least that's what the READMe says22:28
lesshaste1.4.5 :(22:28
lesshasteso I can try to do it manually..the paths that is22:28
lesshasteI now get22:28
lesshasteImportError: Could not import pypy module, make sure to add the pypy module to PYTHONPATH22:28
lesshastewhich file is the pypy module?22:28
fzzzy (~donovan@nat/mozilla/x-egdecpsujerlpihz) joined #pypy.22:29
squiddyyou need a source checkout, it contains a pypy directory22:30
lesshasteI have lib_pypy and lib-python and include22:31
lesshasteunder pypy-1.722:31
lesshastenone of those?22:31
squiddyI think that's a binary release, look at http://pypy.org/download.html#building-from-source for source packages22:32
lesshasteok22:32
Alex_Gaynorfijal: pong22:32
squiddyonce you've extracted the files, you'll see it contains a couple more directories like py, pypy22:33
lesshasteok22:33
lesshastehg is doing it's slow thing :)22:33
lesshasteI don't really understand what takes it so long22:33
lesshastes/it's/its22:33
lesshaste....22:36
lesshastehmm.. what is hg clone doing at the "adding manifests" stage22:39
lesshaste?22:39
lesshasteit just sits there22:39
lesshasteis that a very slow server?22:39
lesshastehmm.. still going :)22:44
asmeurer (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.22:45
gutworthjust wait until file changes!22:54
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) joined #pypy.23:00
kenaan12hager ppc-jit-backend 11c7d964550b10 15/pypy/jit/backend/test/runner_test.py: fix test_spilling23:00
kenaan12hager ppc-jit-backend 112116589f3b9e 15/pypy/jit/backend/ppc/ppcgen/regalloc.py: add get_loc_index23:00
_pyre_ (~stian@137.241.189.109.customer.cdi.no) joined #pypy.23:04
squiddy (~squiddy@g224192142.adsl.alicedsl.de) left irc: Quit: Leaving23:04
Nick change: _pyre_ -> _stian_23:05
Nick change: nedbat_ -> nedbat23:07
rguillebertlesshaste, you needs the pypy executable in your $PATH23:10
rguillebertand the source tree23:10
whitelynx (~whitelynx@63.241.75.144) left irc: Quit: Ex-Chat23:19
mikefc2 (9665bcc5@gateway/web/freenode/ip.150.101.188.197) joined #pypy.23:21
GDorn (~gdorn@li223-147.members.linode.com) joined #pypy.23:22
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Textual IRC Client: http://www.textualapp.com/23:32
fermianyon (~lane@c-71-229-21-197.hsd1.al.comcast.net) joined #pypy.23:36
dmalcolm (david@nat/redhat/x-mxcyzpicwkqvhntz) left irc: Quit: Leaving23:40
stakkars fijal: no idea23:46
ronan (~ronan@host86-143-209-182.range86-143.btcentralplus.com) joined #pypy.23:52
--- Thu Jan 5 201200:00

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