#pypy IRC log for Thursday, 2011-10-27

aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) left irc: Quit: Computer has gone to sleep.00:01
mcdonc_ (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds00:04
tos9It was *too fast*00:07
mcdonc_ (~mcdonc@cabana.palladion.com) joined #pypy.00:10
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection00:18
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) joined #pypy.00:19
eropple|Air (~ed@pool-74-104-187-201.bstnma.fios.verizon.net) joined #pypy.00:23
Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Quit: Ex+re00:24
rekamso (~textual@67.51.82.66) left irc: Ping timeout: 258 seconds00:27
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) left irc: Ping timeout: 276 seconds00:30
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) joined #pypy.00:34
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) left irc: Quit: Computer has gone to sleep.00:39
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) joined #pypy.00:42
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/180900:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/54300:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/59200:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/107100:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/77800:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/141000:49
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/67500:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/24200:49
bradford (~bradford@c-24-7-159-22.hsd1.ca.comcast.net) joined #pypy.00:53
bradford (bradford@c-24-7-159-22.hsd1.ca.comcast.net) left #pypy.00:53
etrepum_ (~bob@accessnat4.mochimedia.net) joined #pypy.00:55
etrepum (~bob@accessnat4.mochimedia.net) left irc: Ping timeout: 240 seconds00:58
etrepum_ (~bob@accessnat4.mochimedia.net) left irc: Ping timeout: 260 seconds00:59
asmeurer_ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Quit: asmeurer_01:03
asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy.01:05
espes (espes@2001:470:1f05:ca1::) left irc: Ping timeout: 240 seconds01:07
mtigas (~Adium@users.spokesman.com) left irc: Quit: Leaving.01:11
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/11701:12
mitchellh (~mitchellh@c-69-181-107-107.hsd1.ca.comcast.net) left irc: Quit: ZOMG BYE01:13
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.01:14
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) joined #pypy.01:24
Trundle (~andy@python/site-packages/trundle) left irc: Ping timeout: 240 seconds01:32
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) left irc: Quit: Computer has gone to sleep.01:38
mcdonc_ (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds01:47
ericflo (~ericflo@75.103.8.110) left irc: Quit: ericflo01:53
Moku (~John@osbk-4d087493.pool.mediaWays.net) joined #pypy.01:53
Nick change: Moku -> Guest1071101:53
Shinita (~John@osbk-4d08746a.pool.mediaWays.net) left irc: Ping timeout: 248 seconds01:54
Nick change: Guest10711 -> Shinita01:54
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/24201:58
Ademan (~dan@adsl-71-141-242-47.dsl.snfc21.pacbell.net) joined #pypy.02:09
DasIch (~dasich@p4FFDFEA3.dip.t-dialin.net) left irc: Ping timeout: 258 seconds02:09
DasIch (~dasich@p4FFDF3B2.dip.t-dialin.net) joined #pypy.02:10
MjrTom (MjrTom@azureus/MjrTom) left irc: Remote host closed the connection02:15
MjrTom (MjrTom@azureus/MjrTom) joined #pypy.02:16
JaredWigmorehi, what's the status of rpyc to connect pypy and cpython02:44
Alex_Gaynorprobably unchanged since whatever the last thing we wrote about it, I haven't seen any work on it in the last year02:44
JaredWigmoreI tried using it to connect OpenCog to pypy; which uses some C++ code and Cython to call complex Python algorithms that access several complex datastructures in C++ via Cython. Unsurprisingly it had lots of bugs :P I'll try just making it less perverse02:46
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/92002:47
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) left irc: Ping timeout: 252 seconds03:09
mat^2 (~mathias@212.130.113.35) joined #pypy.03:13
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) joined #pypy.03:25
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/180903:53
lizardo (~lizardo@201008207211.user.veloxzone.com.br) left irc: Quit: Leaving04:03
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Quit: BRB, MAYBE04:07
redsteg (~steven@129.215.36.46) left irc: Ping timeout: 240 seconds04:07
redsteg (~steven@129.215.36.46) joined #pypy.04:09
_pyre_ (~stian@137.241.189.109.customer.cdi.no) left irc: Quit: Leaving.04:10
gutworth (~benjamin@dhcp-128-36-89-186.central.yale.edu) left irc: Read error: Operation timed out04:11
JaredWigmoreusing rpyc + cython seems a bit messy, e.g. Cython methods don't like the proxy objects04:15
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.04:21
kvda (~kvda@124-171-36-124.dyn.iinet.net.au) joined #pypy.04:26
mwhudson (~mwh@linaro/mwhudson) left irc: Quit: Leaving04:31
mcdonc (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds04:36
CIA-2503bgola 07roundup * 10#923/Regular expressions with optional look-ahead operations failing to match: 04:44
CIA-25[new] I'm trying to run moin-2.0 on PyPy and I found this weird behavior that I think04:44
CIA-25is a bug.04:44
CIA-25Whenever you have a regexp that uses the look ... * 14https://bugs.pypy.org/issue92304:44
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.04:48
tomh (u501@gateway/web/irccloud.com/x-jtouxswgdbefglwe) left irc: Excess Flood04:52
tomh (u501@gateway/web/irccloud.com/x-ghkrqcijcpmjsvvr) joined #pypy.04:52
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/141004:55
mat^2 (~mathias@212.130.113.35) left irc: Ping timeout: 256 seconds04:55
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.04:59
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/54305:11
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/59205:12
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 276 seconds05:12
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/107105:13
mat^2 (~mathias@212.130.113.35) joined #pypy.05:13
mcdonc (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds05:13
espes (espes@2001:470:1f05:ca1::) joined #pypy.05:25
bgola (~bgola@c95185bc.virtua.com.br) left irc: Ping timeout: 260 seconds05:32
mat^2 (~mathias@212.130.113.35) left irc: 05:34
bgola (~bgola@c95185bc.virtua.com.br) joined #pypy.05:38
MjrTom (MjrTom@azureus/MjrTom) left irc: Disconnected by services05:39
MjrTom_ (MjrTom@azureus/MjrTom) joined #pypy.05:39
MjrTom (MjrTom@azureus/MjrTom) joined #pypy.05:40
MjrTom_ (MjrTom@azureus/MjrTom) left irc: Ping timeout: 248 seconds05:44
asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Quit: asmeurer05:59
stakkars_ (~tismer@dslb-088-072-196-002.pools.arcor-ip.net) joined #pypy.06:05
kushal (~kdas@114.143.166.239) joined #pypy.06:05
kushal (~kdas@114.143.166.239) left irc: Changing host06:05
kushal (~kdas@fedora/kushal) joined #pypy.06:05
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/77806:06
eropple|Air (~ed@pool-74-104-187-201.bstnma.fios.verizon.net) left irc: Quit: eropple|Air06:07
kdas_ (~kdas@114.143.166.239) joined #pypy.06:11
kushal (~kdas@fedora/kushal) left irc: Ping timeout: 258 seconds06:12
EnCuKou (~encukou@193.167.65.94) joined #pypy.06:12
stakkars_ (~tismer@dslb-088-072-196-002.pools.arcor-ip.net) left irc: Quit: schnarch06:26
JaRoel|4D (~jaroel|4d@sink.jaroel.nl) left irc: Remote host closed the connection06:26
guibou (~guibou@kraken.insecable.net) joined #pypy.06:37
mvt (~mvantelli@87.213.45.85) joined #pypy.06:38
Nick change: kdas_ -> kushal06:50
kushal (~kdas@114.143.166.239) left irc: Changing host06:50
kushal (~kdas@fedora/kushal) joined #pypy.06:50
guibouHello here. I'm wondering, is there any heuristic in pypy to optimise stuff such as sum(x * x for x in u). I have implemented a simple vector implementation, and it is slow as hell (ie, 2 times faster than cython, but 20 times slower than C) using generators like that. If I replace my function by something like self.values[0] * self.values[0] + self.values[1] * self.values[1] + self.values[2] * self.values[2], it fly like hell (ie, 2 times slower than th06:52
ronnyguibou: your message got cut off06:56
guibouha, so bad. return sum(x * x for x in self.values) is far far far (did i say far) slower than return self.values[0] * self.values[0] + self.values[1] * self.values[1] + self.values[2] * self.values[2]07:00
guibouIs there a way to get the speed of the later using a generic and nice syntax as the former ?07:00
guibou(by far far slower, I mean, first is 0.5 times cpython, 20 times C, later is 2 times C)07:01
MostAwesomeDudeguibou: FWIW that kind of expanded dot product is necessary in CPython and Cython too.07:01
guibous/cython/cpython in all my previous text ;)07:01
Action: Garen gave a Linux VM 6GB RAM, and it's still not enough to build pypy trunk without swapping. wtf07:02
guibouThe question I'm wondering is, why the jit does not realize that len(values) == 3 and does not generate the optimised version of the loop ? (I don't say it is trivial, I'm just trying to understand what the jit is really able to do)07:03
JaRoel|4D (~jaroel|4d@office.fourdigits.nl) joined #pypy.07:06
Garen (garen@69.76.18.3) left irc: Remote host closed the connection07:09
Garen (garen@69.76.18.3) joined #pypy.07:15
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/67507:15
Trundle (~andy@p578bfdcf.dip0.t-ipconnect.de) joined #pypy.07:17
Trundle (~andy@p578bfdcf.dip0.t-ipconnect.de) left irc: Changing host07:17
Trundle (~andy@python/site-packages/trundle) joined #pypy.07:17
amaury (~amaury_@74.125.57.34) joined #pypy.07:29
tlynn (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) joined #pypy.07:36
amaury (~amaury_@74.125.57.34) left irc: Ping timeout: 240 seconds07:37
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds07:38
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.07:40
amaury (amaury_@nat/google/x-votcdcxvaddsvnbl) joined #pypy.07:51
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 260 seconds07:52
nirbheek_ (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.07:52
Da_Blitzguibou: try writing a function called sum that adds the numbers up and use it instead07:53
Da_BlitzIIRC sum is optimised for adding 2 numbers together and is faster when rewritten as python code as opposed to the current rpython implemntation07:54
teknico (~quassel@88-149-208-230.dynamic.ngi.it) joined #pypy.07:56
ronnyhmm07:59
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.07:59
stakkars_ (~tismer@89.204.137.106) joined #pypy.08:01
arigato (~arigo@89.204.154.65) joined #pypy.08:03
verte (~verte@python/site-packages/verte) joined #pypy.08:03
antocuni (~antocuni@host200-47-dynamic.116-80-r.retail.telecomitalia.it) joined #pypy.08:04
arigatohi08:04
ronnyhmm08:05
arigatofijal: I think that a2b911e61392 is just a random pseudo-fix08:05
ronnywhere is collections.Counter implemented,08:05
arigatothe real issue is somewhere else08:05
ronnyhmm08:07
ronnyactually i need heapq.nlargest08:07
mwhudson_ (~mwh@121-73-77-183.cable.telstraclear.net) left irc: Quit: Ex-Chat08:10
fijalarigato: which part?08:12
fijalm08:12
fijalright it's silly08:12
Action: arigato tries to fix08:13
fijalarigato: done08:14
arigato?08:14
kenaan12fijal numpy-multidim 11cc8b141031c6 15/pypy/module/micronumpy/: shape08:14
kenaan12fijal numpy-multidim 114dd7d695536a 15/pypy/module/micronumpy/: tentative checkin08:15
kenaan12fijal numpy-multidim 1152416083d855 15/pypy/jit/backend/: a better fix for test_zll_random08:15
fijalgrrr08:15
fijalwrong branch08:15
arigatoeh? no, that's not enough08:15
fijalwhy?08:15
kenaan12fijal default 11b6e0fd170dae 15/pypy/jit/backend/: a better fix for test_zll_random08:15
arigatothe issue is that force_result_in_reg() should not be used that way08:16
arigatobecause it's not the result at all08:16
fijaldoes it care?08:16
arigatoI don't know, maybe it can also work in that way, but I would prefer seriously to avoid that08:16
fijalok08:16
fijalthe idea is the same as for result in say imul08:16
arigatoalso, you missed that there is a three-operands IMUL08:16
fijalI want this to be in register, but also I'll scratch this register08:17
verte (~verte@python/site-packages/verte) left irc: Ping timeout: 258 seconds08:17
arigatowhich can be better08:17
fijalok08:17
fijalwell, technically, for a lot of cases we can avoid operation alltogether08:17
arigatoyes, that's another issue08:17
fijaland just pass scale08:17
fijalbut dict's are 24 bits anyway08:17
fijalok, I did miss 3-arg imul08:17
arigatofor now I'm just trying to use the interface of llsupport/regalloc as documented08:17
arigatoI think it's already a good thing08:18
fijalyes :)08:18
ronnyyay, heapq.nlargest is broken08:18
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood08:19
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.08:19
ronnyand even better, the stdlib one is broken for retarded reasons08:19
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy08:25
amauryhmm, translated code is full of these: http://paste.pocoo.org/show/498876/08:26
amaury(the ^@ is actually a NUL character in the file)08:27
amauryI use a recent nightly build of pypy-jit, 64bit linux08:27
vak (3ef59c34@gateway/web/freenode/ip.62.245.156.52) joined #pypy.08:28
vakhi all08:28
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) left irc: Remote host closed the connection08:28
vakwhat happened to speed.pypy.org?08:29
fijalvak: it's broken08:30
fijalamaury: uh?08:30
fijalarigato: feel like reviewing a blog post?08:31
amauryI was translating default branch, btw08:31
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) joined #pypy.08:32
fijalpedronis: morning samuele08:32
pedronismorning08:32
ronnyi am tempted to put a _heapq.py in lib_pypy that does nothing but 'from itertools import islice as _islice;def islice(...): _islice(max(n,0),...)'08:32
fijalronny: why?08:33
ronnypure python version of heapq in the stdlib is broken08:33
fijalin what way?08:33
arigatofijal: ok08:34
k_bx (~k_bx@195.20.130.1) joined #pypy.08:34
fijalarigato: https://bitbucket.org/fijal/talk/src/8b99014b0107/blog/road-towards-scipy.rst08:34
ronnyfijal: heapq.nlargest/smallest uses islice without guarding for negatives, the c versions however do 08:34
fijalronny: file a bug upstream please08:34
ronnyhow do we want to deal with it?08:35
fijalarigato: exarkun said it's unclear whether this convinces anyone08:35
fijalronny: start with cpython bug issue08:35
arigatoronny: then copy heapq.py to modified-2.7 and fix it here08:36
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.08:36
kushal (~kdas@fedora/kushal) left irc: Quit: This computer has gone to sleep08:39
arigatofijal: I think such a blog post makes sense, even if it's real use is probably limited to: having an official place you can point people that argue "numpypy is bad because <bogus reason x>"08:40
fijalmeh :/08:41
fijalI'll post it on my private blog08:41
fijalI don't feel like there is a place on pypy's blog for that08:41
arigatoyes, that's why I said *you* as in fijal08:42
fijalhaha08:42
CIA-2503ronny 07roundup * 10#924/stdlib heapq islice missuse: 08:42
CIA-25[new] http://bugs.python.org/issue1327408:42
CIA-25needs a copy in modified-2.7 and tests * 14https://bugs.pypy.org/issue92408:42
fijalI'm fine with saying "it's nonsense"08:42
fijalI'm borderline :)08:42
fijalronny: I meants bugs.python.org really08:42
fijalronny: for us feel free to just copy heapq to modified and fix it08:42
ronnyfijal: see the bug text08:43
ronnyi cant fix right now08:43
fijalok08:43
kenaan12arigo default 11478d71b47de8 15/pypy/jit/backend/: Revert the hack to llsupport/regalloc.  Implement a version that uses the hopefully "correct and official" API.08:47
kenaan12arigo default 1138058e92147d 15/pypy/jit/backend/x86/: merge heads08:47
stakkars_ (~tismer@89.204.137.106) left irc: Quit: schnarch08:48
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Remote host closed the connection08:52
khs (~khs@2001:700:300:2120:725a:b6ff:fee5:a44) joined #pypy.08:52
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.08:52
kenaan12arigo default 119f32a507b5be 15/pypy/jit/backend/x86/assembler.py: A better interface.  Possibly a translation fix.08:54
amauryfijal: I manually replaced all the NUL chars with '-' in C code, and pypy-c works08:58
fijalamaury: voodoo08:58
amaurythe issue seems to be the way large (=negative) hash values are printed08:58
amauryfijal: I found this by computing the hash value myself08:59
amauryi.e externally with CPython08:59
mat^2 (~mathias@212.130.113.35) joined #pypy.09:05
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.09:06
arigatoamaury: any clue where it comes from?09:08
arigatoamaury: e.g. does it work if you try to compute on the prompt: '%dL' % large-negative-value?09:10
amauryyes09:12
Action: fijal fights with RPython and reusing code09:13
arigatoand if you do that with the JIT?  e.g. for i in range(10000): the same thing09:13
amaurythat's what I did09:17
aleksi (~aleksi@85.235.191.82) joined #pypy.09:17
mat^2 (~mathias@212.130.113.35) left irc: 09:19
arigatocan you paste in a bug report the details, e.g. which pypy-c etc.?09:24
arigatoit's not impossible that it's related to the new (and previously buggy) getinteriorfield in the jit09:25
arigato"and previously buggy" = "I think it was buggy at some point, but maybe that bug wasn't one, but then again maybe there's another bug..."09:26
fijalsetinteriorfield was buggy09:26
fijaland it never get emitted on trunk09:27
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.09:27
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) joined #pypy.09:27
cfbolzhello09:27
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.09:27
cfbolzarigato: ping?09:27
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.09:27
witulskihello²09:27
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.09:27
fijalcfbolz: hi carl09:27
cfbolzusing tagged integers (and the jit), executing richards becomes twice as fast09:28
arigatocfbolz: hi09:28
cfbolzand pystone 30% faster09:28
arigatofijal: ah, oh09:28
mat^2 (~mathias@212.130.113.35) joined #pypy.09:28
arigatocfbolz: uuuuh09:28
arigatoexcellent I suppose09:28
cfbolzalthough I am comparing against a slightly older pypy09:29
cfbolzshould try a nightly09:29
cfbolz(I'm sure the effect will now disappear)09:29
fijal?09:29
fijalannoying :/09:29
fijalor good?09:29
cfbolzI don't know, it's definitely surprising09:29
mat^2 (~mathias@212.130.113.35) left irc: Client Quit09:30
fijalyes09:30
cfbolzhahaha09:30
cfbolzof course the effect disappears when I use a nightly :-)09:31
cfbolzgood09:31
mat^2 (~mathias@212.130.113.35) joined #pypy.09:31
amaury (amaury_@nat/google/x-votcdcxvaddsvnbl) left irc: Ping timeout: 240 seconds09:32
lac (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Quit: No Ping reply in 180 seconds.09:33
lac (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.09:33
arigato:-\09:34
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 240 seconds09:34
cfbolzarigato: that's good, it means we got 2x faster over some period09:35
kushal (~kdas@114.143.166.239) joined #pypy.09:35
kushal (~kdas@114.143.166.239) left irc: Changing host09:35
kushal (~kdas@fedora/kushal) joined #pypy.09:35
arigatoyes, on richards09:35
cfbolzyes09:35
fijalhttp://speed.pypy.org/09:36
fijalso we're only 4.4x faster than python 2.709:36
arigatodamn, the 5x number is against python 2.609:36
arigatowhat was unladen swallow's number against?  :-)09:36
cfbolzarigato: 3.1?09:37
cfbolzah, no09:37
cfbolzthe other way round09:37
arigatoer?09:37
cfbolzI thought it was a fork of 3.109:37
arigatook, but what other way around?09:38
cfbolzbut it is one of 2.x and the PEP promised to adapt it to 3.109:38
fijalcfbolz: no no09:38
fijalcfbolz: only later09:38
arigatoah09:38
fijalarigato: also we uploaded translation figures09:39
arigatoah, you mean that we're slower than 5x because of the numbers for translation09:39
arigato(well "because of" or not)09:39
fijaland sympy and json_bench09:40
fijalyes, we have more of less-awesome benchmarks these days09:40
arigatogood09:40
cfbolzyes09:40
fijalhttp://speed.pypy.org/comparison/?exe=1%2BL%2Bdefault%2C2%2B472&ben=1%2C34%2C27%2C2%2C25%2C3%2C4%2C5%2C22%2C6%2C39%2C7%2C8%2C23%2C24%2C9%2C10%2C11%2C12%2C13%2C14%2C15%2C35%2C36%2C37%2C38%2C16%2C28%2C30%2C32%2C29%2C33%2C31%2C17%2C18%2C19%2C20&env=1&hor=false&bas=2%2B472&chart=normal+bars09:40
arigatobtw, anyone feel like killing "translation-stackcheck"?09:40
Action: fijal kills it09:40
cfbolzit's *crucial* that that's fast!09:41
cfbolzok, I should definitely think more before doing quick-shot benchmarks09:41
fijalcfbolz: it's crucial it doesn't jump 100% between days09:41
arigatoI guess its speed varies a lot depending on whether there is a full GCs occurring during this short period of time or not09:41
arigatoor maybe 1 vs 2 full GCs, I don't know09:42
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.09:42
cfbolzI benchmarked default, as opposed to int-tag-untag-as-operations09:42
kenaan12cfbolz int-tag-untag-as-operations 115dea661f90f6 15/pypy/jit/: make all optimizeopt tests pass09:43
kenaan12cfbolz int-tag-untag-as-operations 11be97182b1a31 15/pypy/jit/metainterp/test/: those are also not needed09:43
kenaan12cfbolz int-tag-untag-as-operations 1176ff9be0b630 15/: merge default09:43
fijalcfbolz: haha09:44
kenaan12fijal benchmarks 1139ae6234d7ac 15/benchmarks.py: don't report stackcheck numbers09:44
arigatoyay, now on the int-tag-untag-as-operations branch I can make the GC as bad as I want and richards is not going to show it09:45
cfbolzarigato: possible, yes09:46
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 244 seconds09:46
kenaan12fijal benchmarks 11f04d6d63ba60 15/benchmarks.py: fix a test09:47
arigato (~arigo@89.204.154.65) left irc: Quit: See you09:47
ojii (~ojii@40-34.60-188.cust.bluewin.ch) left irc: Quit: Leaving09:48
fijalcfbolz: we haven't improved on translation since we started measuring09:49
fijalwhich is end of July09:49
cfbolzfijal: surprise?09:51
fijalnot really, no09:51
fijalbut maybe we should one day09:51
fijalI know how to improve on ai btw09:52
cfbolzah?09:52
fijalconstructs like09:53
fijaltuple(<genexpr>)09:53
kenaan12fijal numpy-multidim 11345d2c256ce7 15/pypy/module/micronumpy/: getsetitem for single items09:53
fijalare slow09:53
fijalwe probably need a jitdriver there09:53
fijalfor space.unpackiterable essentially09:53
cfbolzfun09:53
cfbolzI think we might want a general solution for this, at some point09:54
fijallist(<genexptr>)09:54
fijalyes09:54
fijalthis is a reason for *a lot of things* to be slow09:54
fijalbecause we often end up with looking up next on stuff per iteration09:54
fijaland also entry bridge is compiled inefficiently09:54
fijalas I can see now those are two places only so far - which is unpackiterable and it's copy in init__List09:55
dolo (~dolo@124-168-203-130.dyn.iinet.net.au) joined #pypy.09:56
Sho_ (~EHS1@kde/hein) left irc: Quit: Stop leaking memory like it's going out of fashion.09:57
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.09:57
amaury (~amaury_@74.125.57.34) joined #pypy.10:07
fijalAlex_Gaynor: ping10:08
fijalcfbolz: anyway a simple change genexpr->list comprehension speeded it up 5x10:12
mat^2 (~mathias@212.130.113.35) left irc: 10:13
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 256 seconds10:14
lucian (~lucian@93-97-174-115.zone5.bethere.co.uk) joined #pypy.10:19
mat^2 (~mathias@212.130.113.35) joined #pypy.10:19
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.10:28
kenaan12fijal numpy-multidim 1121fa3eefa93b 15/pypy/module/micronumpy/: few small fixes, start working on NDimSlice10:30
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 245 seconds10:33
cfbolzof course switching to the branch changes nothing at all10:38
cfbolzbecause only erased ints are optimized, not tagged ints10:39
Action: fijal realizes he didn't work on new features in a bit10:47
G2P1 (~G2P@fw-asn1.ornis.com) joined #pypy.10:48
G2P (~G2P@fw-asn1.ornis.com) left irc: Ping timeout: 240 seconds10:50
stakkars_ (~tismer@i59F742EF.versanet.de) joined #pypy.10:53
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) joined #pypy.11:04
antocuni (~antocuni@host200-47-dynamic.116-80-r.retail.telecomitalia.it) left irc: Ping timeout: 260 seconds11:37
ronnywhere do we put tests for modified parts of the stdlib?11:41
fijalvoidspace: any reason why you mention ppa now?11:42
voidspacefijal: just going through my preparations for the Ubuntu Developer Summit11:43
voidspacefijal: I fly for Orlando on Sunday11:43
fijalah11:44
fijalcool!11:44
fijalwhat you gonna present?11:44
voidspacefijal: saw the pypy discussions in my preparations folder11:44
voidspacefijal: I'm not presenting at all11:44
voidspacefijal: my team is sprinting there11:44
voidspacefijal: but I'm hoping to discuss pypy in ubuntu with barry and matthias11:44
fijalcool11:46
fijalpypy in canonical or pypy in ubuntu?11:46
voidspaceubuntu :-)11:48
fijalI got a good json speedups11:49
fijalI'll post stuff on my blog11:49
voidspace*great*11:50
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) left irc: Quit: Computer has gone to sleep.11:52
fijalso json in pypy is now faster than the C extension11:52
fijalfor encoding11:52
lacfijal: yippee!11:52
fijallac: we have it on the nightly11:52
k_bx (~k_bx@195.20.130.1) left irc: Ping timeout: 245 seconds11:53
fijalhttp://speed.pypy.org/timeline/?exe=1%2C5&base=2%2B472&ben=json_bench&env=1&revs=200&equid=off11:53
ronnyhmm11:56
gutworth (~benjamin@dhcp-128-36-89-186.central.yale.edu) joined #pypy.11:56
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.11:59
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.12:01
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.12:03
stakkars_lac: Hi Laura12:08
stakkars_lac: I'm still undecided about the sprint. There are other concerns. Are you there?12:10
fijalstakkars_: hi12:12
lacstakkars_: I am here now12:17
Rhy0lite (dje@nat/ibm/x-yhcjgbcravrreemm) joined #pypy.12:18
ojii (~ojii@adsl-84-226-255-220.adslplus.ch) joined #pypy.12:21
antocuni (~antocuni@host200-47-dynamic.116-80-r.retail.telecomitalia.it) joined #pypy.12:25
ojii (~ojii@adsl-84-226-255-220.adslplus.ch) left irc: Read error: Operation timed out12:26
ojii (~ojii@84.226.127.220) joined #pypy.12:28
kushal (~kdas@fedora/kushal) left irc: Quit: This computer has gone to sleep12:30
caution (~caution@unaffiliated/caution) joined #pypy.12:35
cautionpypy is slow, I'm guessing because of the lack of cPickle, how can I load the C version of pickle?12:36
fijalcaution: you can't12:36
santagada (~leonardo@177.18.63.32) joined #pypy.12:36
fijalcaution: you mean pickle is slow?12:36
cautionpickle is fast in regular python12:37
cautionslow in pypy12:37
cautionnothing I can do?12:37
fijalnot really12:38
fijalshort of implementing cPickle in pypy12:38
ronnyagain, where do we keep test for the modified stdlib?12:40
cautionthink I should use marshal instead?12:40
fijalronny: in modified12:40
tlynn (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) left irc: Ping timeout: 240 seconds12:45
kenaan12fijal extradoc 11bb06f91061f3 15/blog/draft/faster-json.rst: Add a blog post12:49
fijalanyone feel like reviewing?12:49
kushal (~kdas@fedora/kushal) joined #pypy.12:51
antocunifijal: I can12:56
fijalantocuni: go ahead12:56
lizardo (~lizardo@189.2.128.130) joined #pypy.12:58
ronnyis there any way to run a single selected applevel cpython test?13:00
k_bx (~k_bx@94.244.19.62) joined #pypy.13:00
fijalronny: no13:03
kushal (~kdas@fedora/kushal) left irc: Quit: This computer has gone to sleep13:03
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.13:04
kenaan12antocuni extradoc 114e7631f04f96 15/blog/draft/faster-json.rst: some english fixes, and rephrase a couple of sentences13:06
antocunifijal: I think you should explain a bit better what the lines in the table refers to13:06
antocuniand possibly explain in the beginning that in cpython there is json.py vs _json.c, etc.13:06
fijalisn't that explained in the beginning?13:07
fijalantocuni: here you go, bold for you13:08
antocuniwell, you say only at the end that in cpython is python+C, while in pypy is python+python13:09
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 258 seconds13:09
antocunialso, I suppose that "Cpython 2.7 no C extension" runs the very same code as "PyPy 1.5" and "PyPy 1.6", but you don't say it13:09
antocuniwhile PyPy trunk runs a different version13:09
kenaan12fijal extradoc 1116b98b486879 15/blog/draft/faster-json.rst: make some stuff bold13:09
fijalNote that CPython by13:09
fijaldefault uses the optimized C extension, while PyPy uses the pure Python one13:09
kenaan12fijal extradoc 11300ff59a5bbd 15/blog/draft/faster-json.rst: merge13:09
fijalhow I don't say that?13:09
antocuninote that that sentence is mine :)13:10
antocuniI'm just saying that you should explain better13:10
fijalno it's not13:11
fijalwell, it was there before13:11
fijalfeel free to improve if you think it's not enough13:11
antocuniyou shouldn't ask for reviews if you don't accept comments :-)13:12
fijalI do13:12
fijalI however don't see what can be clearer13:12
fijalso feel free to improve13:12
fijalI claim it's right there13:12
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) joined #pypy.13:23
derdon (~derdon@pD9E1C7F0.dip.t-dialin.net) joined #pypy.13:28
ojii (~ojii@84.226.127.220) left irc: Remote host closed the connection13:29
fijalantocuni: still reviewing?13:29
antocunifijal: no sorry, I got busy with something else13:30
fijalok13:30
ojii (~ojii@adsl-84-226-255-220.adslplus.ch) joined #pypy.13:31
fijalojii: hey13:34
fijalantocuni: that means you still want to review but not now or you don't feel like it?13:34
antocunino, feel free to publish it13:35
fijalok13:36
kushal (~kdas@fedora/kushal) joined #pypy.13:41
EnCuKou (~encukou@193.167.65.94) left irc: Quit: Mizím.13:44
ojiiyeh fijal 13:45
ojiidid i seriously just randomly typed 'hey' reversed?13:48
ojii#notenoughsleep13:49
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.13:50
CIA-2503bgola 07roundup * 10#923/Regular expressions with optional look-ahead operations failing to match groups: 13:56
CIA-25[chatting] More information on this, I simplified the test case and did some debug. The13:56
CIA-25regexps bytecode are the same on both PyPy and CPython.13:56
CIA-25... * 14https://bugs.pypy.org/issue92313:56
matehat (~matehat@184.151.127.230) joined #pypy.13:59
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.14:00
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.14:00
kenaan12fijal numpy-multidim 116fd1b6212a20 15/pypy/module/micronumpy/: few small fixes, start working on NDimSlice14:04
kenaan12fijal numpy-multidim 11c50cad7db70b 15/pypy/module/micronumpy/test/test_numarray.py: fix the test. previous commit by accident - start working on slices14:04
kenaan12fijal numpy-multidim 11301f1937556a 15/pypy/module/micronumpy/: Work on casting. Seems to fly14:04
kenaan12fijal numpy-multidim 114fb18345b1ab 15/pypy/module/micronumpy/test/test_numarray.py: more tests14:09
kenaan12fijal numpy-multidim 11e6db0d0d0445 15/pypy/module/micronumpy/test/test_numarray.py: even more tests, you can never be sure14:10
fijalAlex_Gaynor: wuhu14:11
ronnydamn it14:11
fijalseems to fly at least for simple stuff14:11
fijalronny: ?14:11
ronnyi cant even run the single file for the heapq test14:11
ronny(it does in the middle, and errors on my test for the heapq check14:12
santagada (~leonardo@177.18.63.32) left irc: Ping timeout: 240 seconds14:13
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 255 seconds14:13
Action: fijal broke *all* the zjit tests though14:13
matehat (~matehat@184.151.127.230) left irc: Quit: Leaving...14:13
ronnyis there any reasonable way to run specific tests on a fast box?14:16
fijalronny: what are you doing?14:19
fijalronny: you use pypy-c for those tests right?14:19
antocuni_ (~antocuni@host193-122-dynamic.0-79-r.retail.telecomitalia.it) joined #pypy.14:20
ronnyfijal: yes, i suppose im invoking something wrong, it runs them in py.py14:21
ronnyfijal: if i just do a pypy -m test.test_heapq, stuff is fine14:22
fijalyeah, then why not do that?14:23
antocuni (~antocuni@host200-47-dynamic.116-80-r.retail.telecomitalia.it) left irc: Ping timeout: 260 seconds14:23
kenaan12RonnyPfannschmidt default 113fec6b7c733c 15/lib-python/modified-2.7/: copy heapq to modified-2.7 and correctly deal with nlargest/nsmallest getting n < 0, fixes issue92414:25
santagada (~leonardo@201.47.210.238.dynamic.adsl.gvt.net.br) joined #pypy.14:26
Rhy0litefijal: ping14:29
tomh (u501@gateway/web/irccloud.com/x-ghkrqcijcpmjsvvr) left irc: Read error: Connection reset by peer14:30
fijalRhy0lite: ponfg14:32
fijalpong even14:32
Rhy0liteTravis Oliphant will visit IBM Watson next week14:33
fijalcool14:33
Rhy0liteanything you want me to discuss?14:33
fijalnot sure if you can convince him about anything14:33
Rhy0lite:-)14:33
Rhy0litedo you want to join in a phone conversation with him?14:33
Rhy0litebasically, given that he will be captive here for a day14:34
Rhy0liteif I can help with the conversation, let me know14:34
fijalRhy0lite: see PM14:34
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.14:36
Ademan (~dan@adsl-71-141-242-47.dsl.snfc21.pacbell.net) left irc: Quit: leaving14:37
hobu (~hobu@prometheus.hobu.net) joined #pypy.14:37
kushal (~kdas@fedora/kushal) left irc: Quit: Leaving14:45
fijalRhy0lite: feel like having a look at my blog post: https://bitbucket.org/pypy/extradoc/src/300ff59a5bbd/blog/draft/faster-json.rst14:46
dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) joined #pypy.14:50
lucianfijal: wow, that's really fast. awesome14:53
fijallucian: it can go faster14:54
fijalthere are few more tweaks14:54
lucianfijal: :)14:58
kenaan12fijal extradoc 115ce3464a789c 15/blog/draft/faster-json.rst: fixes by rhyolite14:59
mvt (~mvantelli@87.213.45.85) left irc: Quit: Leaving15:02
mcdonc (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds15:02
kenaan12edelsohn extradoc 11ae2dbf57dd1b 15/blog/draft/faster-json.rst: More native English speaker phrases.15:03
kenaan12edelsohn extradoc 11b4c00aeae648 15/blog/draft/faster-json.rst: Another typo.15:05
ronnymeh15:07
kenaan12edelsohn extradoc 11a9917028d281 15/blog/draft/faster-json.rst: I'm on a roll.15:07
ronnyseems like it is impossible to have a sane way to handle test collection for subitems of the cpython testsuite15:08
aleksi (~aleksi@85.235.191.82) left irc: Remote host closed the connection15:09
stakkars_ (~tismer@i59F742EF.versanet.de) left irc: Quit: schnarch15:10
k_bx (~k_bx@94.244.19.62) left irc: Ping timeout: 260 seconds15:12
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.15:12
k_bx (~k_bx@94.244.19.62) joined #pypy.15:13
twomashi (~Adium@190.2.44.65) joined #pypy.15:16
kenaan12fijal extradoc 11f79ab024b2c1 15/blog/draft/faster-json.rst: mention generators15:17
kenaan12fijal extradoc 117e2623ff21cf 15/blog/draft/faster-json.rst: expand jit is good section15:17
canta (~canta@77-20-123-240-dynip.superkabel.de) joined #pypy.15:17
kenaan12cfbolz int-tag-untag-as-operations 11d3b9577c4c08 15/pypy/rpython/lltypesystem/rtagged.py: make tagged instances use the new operations as well15:17
twomashiHi all. I'd like to use pypy to speed up some data processing scripts, but i'm worried about it's JSON performance15:19
cfbolztwomashi: try a nightly, it has been much improved15:19
twomashiis there a fast JSON library for pypy, or a fast serialization format compatible with cPython?15:19
twomashiwhich one, simplejson?15:19
fijaltwomashi: no, a nightly pypy comes with a much faster json15:20
antocuni_twomashi: fijal is about to publish a blog post about improved json performance :-)15:20
twomashiok, I'll give it a shot!15:20
Action: fijal just called a function that has # This is currently not possible to be called from anywhere.15:20
tomh (u501@gateway/web/irccloud.com/x-rbnhjcjzilzubfsf) joined #pypy.15:21
twomashiantocuni_: that's great, bad JSON performance would be a major gap for pypy in web apps15:22
fijaltwomashi: yop15:22
fijalthat's why improved it :)15:22
kenaan12fijal extradoc 115c778250e982 15/blog/draft/faster-json.rst: add another para15:26
kenaan12fijal numpy-multidim 1161f36db28f06 15/pypy/module/micronumpy/: setitem with slice - part one15:27
ronnydo we have anything to help with speedy reading of binary file formats?15:33
fijalnope15:34
elliottc1ble (~me@ell.io) joined #pypy.15:34
fijalronny: if you're looking for a nice work, improving bytearray would be cool15:35
ronnyfijal: right now i need something that sanely can turn me a tar stream into reasonable data15:35
fijalbytearray is ideal15:36
ronnytarfile of the stdlib is pretty much broken15:36
fijalexcept you have to improve it15:36
kenaan12fijal extradoc 11f55956fe1956 15/blog/draft/faster-json.rst: some more fixes from rhyolite15:37
ronnyfijal: i not sure i understand15:37
fijalwhich part?15:37
elliottc1ble (~me@ell.io) left irc: Client Quit15:38
jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Ping timeout: 240 seconds15:38
mcdonc (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds15:38
ronnyfijal: how exactly would bytearray help with my problem?15:39
ronnyirs pretty much reading and binary structure parsing15:39
fijalstruct is fast15:39
ronnyam i understanding it right, that the intention is to read the headers in a right-sized bytearray, then run struct on those?15:40
fijalyes15:40
fijalbut not as of now15:40
fijalsince bytearray is dead slow15:40
ronnywould there be any noticable difference to bytes/str?15:41
fijalyes15:41
fijalbytearray is mutable15:41
fijalso you can make file.readinto() fast15:41
fijaland things like this15:42
fijalbytes and bytearray are the same thing no?15:42
ronnyhmk15:43
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.15:43
ronnyi dont have the slightest idea how to make them fast tho15:43
cfbolzfijal: XXX: explain in one line what the benchmark does?15:43
cfbolzfijal: is still in the post15:43
fijaluh15:43
twomashii've just done a very quick test of nightly build to decode alot of JSON data, and im not seeing any improvement over a build from ~1 month ago using json.loads15:44
fijalremoved the XXX15:44
twomashithe new build i got was pypy-c-jit-latest-linux6415:44
fijalit's silly, the code explains it much better I think15:44
fijaltwomashi: the improvements only cover encoder :)15:45
fijaltwomashi: I can improve the decoder as well I guess15:45
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.15:45
fijalhttp://morepypy.blogspot.com/2011/10/speeding-up-json-encoding-in-pypy.html15:45
fijaltwomashi: it's relatively straightforward, do you feel like contributing?15:46
kkris (~kris@93-82-46-138.adsl.highway.telekom.at) joined #pypy.15:46
fijalronny: is there a way to get py.test to print me a traceback if I press Control C?15:46
twomashifijal: I'd love to, if i can...15:46
fijaltwomashi: sure you can15:46
twomashinever did anything on compilation though...15:46
fijalit's all pure python15:47
fijalyou don't even have to compile anything15:47
fijalhttp://morepypy.blogspot.com/2011/10/speeding-up-json-encoding-in-pypy.html15:47
fijalreading the blogpost is a good start15:48
fijaland providing a reasonable benchmark is a second step15:48
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.15:48
ronnyfijal: i think i'll be working on that one, atm there isnt15:49
fijalronny: :/15:49
twomashiso the work is digging into the pure python json package and finding what the bottlenecks are under pypy15:50
twomashiI'll give it a shot15:50
fijaltwomashi: yes15:50
fijalAlex_Gaynor: ping15:50
twomashiok i'll let you know if i come up with anything. Thanks!15:51
fijalfeel free to have questions15:51
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 240 seconds15:52
mat^2 (~mathias@212.130.113.35) left irc: Ping timeout: 240 seconds15:56
lucian (~lucian@93-97-174-115.zone5.bethere.co.uk) left irc: Ping timeout: 260 seconds15:58
fijalwe should do a pypy release16:01
fijalI'm waiting for justin's branch16:02
Trundle (~andy@python/site-packages/trundle) left irc: Remote host closed the connection16:02
k_bxSo PyPy does have a __pypy__.builders.StringBuilder :-)16:04
asmeurer_ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy.16:04
asmeurer_ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Read error: Connection reset by peer16:06
asmeurer__ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy.16:06
fijalk_bx: yop16:06
whitelynx (~whitelynx@li117-47.members.linode.com) joined #pypy.16:06
G2P1 (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving.16:06
khs (~khs@2001:700:300:2120:725a:b6ff:fee5:a44) left irc: Quit: Leaving16:07
Alex_Gaynorfijal: pong16:10
fijalAlex_Gaynor: lots of numpy tests fail with -A16:11
fijalI'm unhappy16:11
Alex_Gaynorfijal: last time I looked it was only some str formatting ones16:11
fijalbool seems striking16:11
Alex_GaynorAH yes, it's we don't have numpy.bool_ type16:11
fijalcan you double check now?16:11
Alex_GaynorI'll fix it eventually16:11
fijalok16:11
Alex_GaynorIt's on my TODO, after complex16:11
fijalcan you provide skip for tests that don't conform somehow?16:11
fijalwould be super useful16:11
Alex_Gaynorcomplex of course required interiorfield16:11
nirbheek_ (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Read error: Operation timed out16:12
fijalis copy doing really a copy?16:14
fijaland why it's unclear from tests?16:14
Alex_Gaynorfijal: dunno, also "RMS is nuts" is totally true16:16
Alex_Gaynorfijal: when he came to my school he got into a shouting match with a student about creative commons music or something like that16:16
fijalawesome16:17
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) left irc: Read error: Operation timed out16:18
fijalwell, at least he does not use browsers16:19
Action: fijal ponders why16:19
nirbheek_ (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.16:20
fijalon the other hand, if he does not use browsers, why he does not have an ssh on 443?16:20
Alex_Gaynorwho gives a damn ;)16:20
Alex_Gaynorok time to write an optimizeopt test so I can merge this branc16:20
k_bx (~k_bx@94.244.19.62) left irc: Ping timeout: 245 seconds16:25
Alex_Gaynorfijal: what's the API for interiorfielddescrof?  what args does it take16:28
fijalARRAY and a fieldname16:29
Alex_Gaynorthanks16:29
kenaan12alex_gaynor virtual-dicts 115e267e0a16b0 15/: merged default16:30
kenaan12alex_gaynor virtual-dicts 114aa0b3d16e34 15/pypy/jit/metainterp/optimizeopt/test/: optimizeopt test for this16:30
Alex_Gaynorfijal: ok to merge this branch?16:30
fijalI guess so16:30
Alex_Gaynorsweet16:30
hobuI have an issue in some pypy ctypes code that works in cpython land, but does not in pypy. I think it boils down to https://github.com/pypy/pypy/blob/master/lib_pypy/_ctypes/basics.py#L21 and where.__dict__['_base'] ending up as None. Any ideas where I look in my code that could be causing this to happen? 16:31
ronnyoO16:32
ronnywhy do we have a copy on github?16:32
kenaan12alex_gaynor virtual-dicts 119e7581b504d9 15/: close for merge16:32
kenaan12alex_gaynor default 1196310036438c 15/pypy/: merged virtual-dicts branch, this branch makes it possible for array of structs to be virtualized, and allows...16:32
Action: Alex_Gaynor punches buildbots16:33
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1810 [12alex]16:34
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/676 [12alex]16:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/544 [12alex]16:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1072 [12alex]16:34
fijalwuhu!16:34
santagadahobu: why not post your code here?16:43
hobusorry.  It's this cast, right here. https://github.com/Toblerity/rtree/blob/master/rtree/index.py#L67516:44
hobup_mins is a <_ctypes.pointer.LP_LP_c_double object at 0x0000000102ebfcc8>16:44
hobumins is a <_ctypes.array.c_double_Array_2 object at 0x0000000102ebfd00>16:45
hobuwhen we try to cast, we get https://gist.github.com/132012816:46
hobu(using latest 1.6 pypy release)16:46
kenaan12hager ppc-jit-backend 11064b64295d89 15/pypy/jit/backend/ppc/ppcgen/opassembler.py: Ok, there actually IS an SLWI instr. on PPC. Improved generated code in emit_unicodegetitem and emit_unicod...16:46
kenaan12hager ppc-jit-backend 1159fae2045cad 15/pypy/jit/backend/ppc/ppcgen/opassembler.py: merge16:46
kenaan12hager ppc-jit-backend 11b6e8bd8c4973 15/pypy/jit/backend/test/runner_test.py: Skip test_array_of_structs if floats are not supported.16:46
kenaan12hager ppc-jit-backend 110c1ba83db460 15/pypy/jit/backend/ppc/ppcgen/: Implemented DEBUG_MERGE_POINT and JIT_DEBUG.16:46
lac_ (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.16:47
santagadahobu: if you can the first thing is to try a nightly16:47
santagadaronny: there is even a pull request there16:47
fijalsantagada: unlikely to be fixed16:47
Action: hobu fetches16:47
santagadaI know, but it doesn't hurt and is fast16:47
lac (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Ping timeout: 240 seconds16:47
santagadafijal: did you see https://github.com/pypy/pypy/pulls there is apull for numpy16:49
hobusantagada: same result16:49
hobuwith pypy-c-jit-48400-adab424acda7-osx6416:49
santagadahobu: great, now I think the best course would be to make a bug report with all this data :)16:50
hobuI'd like to fix it myself, if someone could give me a clue as to what might be going on16:50
hobupresumably this is the result of some reference going out of scope that would be held in scope cpython, right?16:53
santagadaI don't know, but it is a good guess16:55
kenaan12hager ppc-jit-backend 11cffb9538d307 15/pypy/jit/backend/ppc/ppcgen/: CAST_INT_TO_PTR, CAST_PTR_TO_INT.16:56
teknico (~quassel@88-149-208-230.dynamic.ngi.it) left irc: Remote host closed the connection16:56
fijalhobu: no16:56
fijalhobu: reference going out of scope is not an issue in python16:57
fijallocals keep there reference alive16:57
kenaan12fijal numpy-multidim 111065599494c0 15/pypy/module/micronumpy/test/test_numarray.py: another test16:58
kenaan12fijal numpy-multidim 11647e3d21832c 15/pypy/module/micronumpy/test/test_zjit.py: skip test_zjit for now16:58
kenaan12fijal numpy-multidim 1179e8268eceeb 15/pypy/module/micronumpy/: fix until all tests pass, except test_zjit16:58
fschwidom (~fschwidom@46.114.93.141) joined #pypy.16:59
Alex_Gaynorfijal: I wonder if maybe test_zjit should really just be test_pypy_c tests, because we're mocking out like half of the python VM at this point16:59
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.16:59
fijalAlex_Gaynor: well, the main issue is that we can test it before translations17:01
fijalthat is kinda cool17:01
Alex_Gaynorfijal: true, but it's a pain, you have to  lie to the annotator all over the place, it breaks when you change internal APIs, etc.17:02
fijalwill be super-useful at the point where we want to say benchmark register allocator17:02
fijalsure17:02
fijalwelcome to TDD I fear17:02
fijalit's still super-useful17:02
fijalmultidim arrays kinda work17:02
Action: fijal writes more tests17:03
Alex_Gaynordo you support weird slicing, like x[:,3]17:03
fijalyes17:03
kenaan12hager ppc-jit-backend 1100a70b5f77a9 15/pypy/jit/backend/ppc/ppcgen/: GETFIELD_RAW, SETFIELD_RAW, GETFIELD_RAW_PURE, GETFIELD_GC_PURE.17:03
fijalnot sure about the negative slices yet17:03
fijallook at tests, this is what's supported :)17:04
mitchellh (~mitchellh@c-69-181-107-107.hsd1.ca.comcast.net) joined #pypy.17:04
MjrTom (MjrTom@azureus/MjrTom) left irc: Quit: Planet Earth is blue and there's nothing I can do17:05
MjrTom (MjrTom@azureus/MjrTom) joined #pypy.17:05
Alex_Gaynorfijal: what about the other thingy, x[None,3] or somethign which turns a 1D array into 2D or something like that17:05
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) left irc: Quit: etrepum17:05
fijalprobably not17:06
fijalAlex_Gaynor: write tests17:06
zain__ (~textual@c-67-160-201-63.hsd1.ca.comcast.net) joined #pypy.17:06
Alex_GaynorI don't know what the behavior is17:06
felipejcruz (~felipecru@177.98.0.102) joined #pypy.17:07
fijalthat's why I want -A to work17:07
dgl (~dgl@109.86.165.231) joined #pypy.17:07
Alex_Gaynorgive me a week or two, I'll make it happen17:07
daniloaf (~daniloaf@177.42.133.132) joined #pypy.17:09
daniloaf (~daniloaf@177.42.133.132) left irc: Client Quit17:09
mcdonc (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds17:10
Nick change: TkTech|Train -> TkTech17:11
amaury (~amaury_@74.125.57.34) left irc: Ping timeout: 240 seconds17:15
mtigas (~Adium@users.spokesman.com) joined #pypy.17:17
rekamso (~textual@67.51.82.66) joined #pypy.17:18
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 255 seconds17:21
jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.17:24
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy17:25
kenaan12arigo stm 116ed608d0af8e 15/pypy/translator/stm/: A subclass of llinterp that detects non-STM-friendly instructions.17:26
kenaan12arigo stm 11acb04280b949 15/pypy/translator/stm/: Keep track of the stm mode in which the LLSTMFrame is. Only allow operations that make sense in the current mode.17:26
kenaan12arigo stm 113de5c2a4796d 15/pypy/translator/stm/: A graph transformer, so far just renaming '{get,set}field' into 'stm_{get,set}field'.17:26
kenaan12arigo stm 11070bf17d0fbb 15/pypy/translator/stm/: We cannot return out of the frame that started a regular transaction.17:26
kenaan12arigo stm 11caa7e690cfc7 15/pypy/translator/stm/: - Refactor to support getfield of immutables.   That's enough to raise simple exceptions.  - Complain when an exception...17:26
kenaan12arigo stm 11d382548b28e1 15/pypy/translator/stm/llstminterp.py: Simplify.17:26
kenaan12arigo stm 11df4fdd533f50 15/pypy/translator/stm/: transform.py should not touch immutable getfields17:26
fijalAlex_Gaynor: pff, I don't have a week or two17:26
Alex_Gaynorfijal: change the test to be assert foo is numpy.True_ and numpy.False_ and add numpy.True_ and False_ to be aliases for now?17:27
fijalAlex_Gaynor: busy with multidim, but ok17:29
fijalwhy is numpy.True_ something else in the first place?17:29
kenaan12fijal numpy-multidim 11d73fb983a4bb 15/pypy/module/micronumpy/: creation from sequences17:29
Alex_Gaynorfijal: because numpy.array([True])[0] returns a subclass of bool17:29
fijalpfff17:29
fijalnice17:29
Alex_Gaynorall numpy things use their own types17:29
Alex_Gaynorfijal: ok let me optimize ones() first17:30
fijalmeh17:31
fijalI wonder how the creator can be fast now :/17:31
fijalgood17:31
fijalslices seem to work nicely17:31
kenaan12fijal numpy-multidim 1157661ae58454 15/pypy/module/micronumpy/test/test_numarray.py: creation from sequences17:31
Alex_Gaynorfijal: wait until we merge list-strategies and do special cases for things?17:31
k_bx (~k_bx@94.244.19.62) joined #pypy.17:36
tito (~tito@pdpc/supporter/bronze/tito) joined #pypy.17:36
etrepum (~bob@accessnat4.mochimedia.net) joined #pypy.17:36
fijaletrepum: hi17:36
etrepumfijal: hey17:36
titohey guys. Pypy target llvm, emscripten use llvm to output javascript. So Pypy -> javascript is doable with that approach ?17:37
fijaletrepum: http://morepypy.blogspot.com/2011/10/speeding-up-json-encoding-in-pypy.html17:37
fijalyou might be interested in this17:37
Alex_Gaynorfijal: http://paste.pocoo.org/show/499131/17:37
Alex_Gaynortito: pypy doesn't traget llvm17:37
fijaltito: yop, work though17:37
etrepumfijal: I saw that, looks similar to what the C extension does17:37
fijaletrepum: good I guess17:37
Alex_Gaynoretrepum: except we do it faster than C ;)17:37
titohttp://llvm.org/ProjectsWithLLVM/#pypy > "At the time of this writing, PyPy currently targets LLVM and C"17:37
JaRoel|4D (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection17:37
etrepumfijal: I haven't had a chance to try and reintegrate that code into simplejson yet17:37
etrepumAlex_Gaynor: well, maybe, for that one benchmark17:37
exarkuntito: The llvm backend isn't very well maintained anymore.17:37
Alex_Gaynortito: apparently the time of writing was 2 years ago at least17:38
exarkuntito: That's the fun of "At the time of this writing" :)17:38
Alex_Gaynoretrepum: so we need mroe benchmarks17:38
etrepumAlex_Gaynor: simplejson still calls quite a bit into Python even if you're using the extension17:38
Alex_Gaynoretrepum: exactly, and that's how we win :)17:38
exarkuntito: PyPy also used to have a JavaScript backend, so if you're going to rescue something old and unmaintained, you could just rescue that and leave out llvm.17:38
etrepumAlex_Gaynor: yeah17:38
titoso, if i would want to bring back python to javascript, what's the best approach ? output llvm and use emscripten, or implement javascript output directly ?17:38
kenaan12alex_gaynor default 11e54c4fc626c7 15/pypy/module/micronumpy/: optimize numpy.ones a bit17:39
titoexarkun, yes, i was talking with fijal on that in June :)17:39
exarkunIf you're very familiar with llvm and emscripten and think they work well enough to have a Python runtime on top of them now, then you could try to revive the llvm backend.17:39
exarkunBut if you're not very familiar with them, then I expect going straight to JavaScript would be easier.17:39
ousadotito: there is pyjamas17:40
exarkunFewer moving pieces, fewer surprises.17:40
titopyjamas no thanks. =)17:40
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1810 [12alex]17:41
ousadotito: well.. you said "bring back python to javascript" 17:41
lac_ (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Read error: Connection reset by peer17:42
jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Read error: Connection reset by peer17:42
etrepum_ (~bob@accessnat4.mochimedia.net) joined #pypy.17:42
titoi did a shortcut, but pyjamas is not fiting to my needs, while pypy look the best approach17:42
exarkunousado: pyjamas isn't really Python17:42
ousadoI know I know17:43
jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.17:43
titoso, for bring back javascript to pypy, it's 1. diy 2. found funds for fijal ? :)17:43
lac (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.17:43
etrepum__ (~bob@accessnat4.mochimedia.net) joined #pypy.17:44
etrepum_ (~bob@accessnat4.mochimedia.net) left irc: Read error: No route to host17:44
fijaletrepum: er17:45
etrepum (~bob@accessnat4.mochimedia.net) left irc: Ping timeout: 248 seconds17:45
Nick change: etrepum__ -> etrepum17:45
fijaletrepum__: I kinda claim we can do better for many benchmarks :)17:45
etrepumfijal: yeah, I believe that's possible, I'd just have to try it with some more to see it myself17:45
fijalwe have a better GC17:46
fijalfor starters17:46
fijaltito: it's a bit unclear17:46
fijallike are you willing to have a 10M download for a python interpreter?17:46
titofijal, in the first time, yes, but they are lot of ways of reducing that17:47
fijalok17:47
fijalthen revive the JS backend17:47
titopypy is able to understsand high level python, while llvm -> emscripten woudln't17:47
fijalit's a bit of work to finish though17:48
kenaan12alex_gaynor default 11de9715d6219f 15/pypy/jit/metainterp/optimizeopt/test/: fix this test17:48
fijalit won't be much better than emscripten for starters17:48
fijala bit depends what are your goals17:48
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.17:48
mat^2 (~mathias@212.130.113.35) joined #pypy.17:48
titohttp://kivy.org -> same code running anywhere17:48
exarkunDid all the projects about just putting a Python runtime into a browser die?17:49
titowell, even if they die, what i'm seeing is something that people want, and so, would be possible to bring it no ?17:51
ousadotito: you're better off using a sane language that can be translated to JS without having to simulate too much of a dynamic runtime17:51
kenaan12fijal numpy-multidim 11e7d27a90d510 15/pypy/module/micronumpy/: nonzero support and more tests for slices17:51
fijalexarkun: I guess security is an issue17:51
titoas all other project, it's just a matter of interest / fund17:51
fijalousado: that's very debatable17:51
fijalousado: and it's not really a topic here, pyjamas is not python17:51
fijalthat's it17:51
titoyes17:51
fijalyou can be happy with it, you can be not17:51
exarkunfijal: Issues aside, I'm just curious about what people are working on17:52
exarkunI remember a couple projects people started to do this, but I forget what they were called.17:52
fijalyeah, people tried17:52
ousadoI'm not talking about pyjamas.. anymore that is17:52
fijalhey, even I embedded pypy into webkit17:52
titoi don't target a python app to be as small and fast as if it was written directly in javascript17:53
titowhat i'm targeting is to maintain one python code over all platform. and it's not for public, but if you think about buisness app, they can run on desktop or web.17:54
tito20Mb downloading in cache is ok for that approach.17:54
titoyou're downloading much more when lurking on youtube so...17:55
fijalok17:55
kenaan12fijal numpy-multidim 11fbb3d36afef8 15/pypy/module/micronumpy/test/test_numarray.py: make the test test what I wanted to test17:55
fijaltito: yeah, well, finding some funds would make it happen :)17:55
fijalas we're discussing options17:55
fijalyou can just embed pypy/cpython into the browser17:55
fijaland ship a custom browser17:56
fijalif you can do it17:56
Alex_Gaynorno one would use that17:56
fijaltito: but I am generally available for hire ;-)17:56
fijalAlex_Gaynor: depends what your target is17:56
fijalif you're shipping an application17:56
Alex_Gaynorfijal: have you done any work on broadcasting for ufuncs yet?17:56
fzzzy (~donovan@nat/mozilla/x-nualopzjotsxpcuq) joined #pypy.17:57
titoembed pypy/cpython into browser, nop. but if somebody is able to make it an alternative to javascript for everyone, why not :) But it's a long term goal ^^17:57
fijalAlex_Gaynor: nope17:57
fijaltito: I don't think anyone would ship pypy with a browser due to binary size17:58
titofijal, i just hope to get fund fast now.17:58
fijaltito: so JS backend sounds like an option17:58
titoyep17:58
fijalAlex_Gaynor: I'll get stuff done as fast as I can write tests :)17:59
Alex_Gaynorfijal: +1 for tests :)17:59
fijalseems like a lunch break anyway17:59
fijaltest_zjit is fucked18:00
fijalwe need to unroll all the shape encoding loops18:00
Alex_Gaynor@jit.look_inside_iff18:00
Alex_Gaynor:)18:00
fijalno no18:00
fijaljust unroll_safe18:00
twomashi (~Adium@190.2.44.65) left irc: Quit: Leaving.18:01
tomh (u501@gateway/web/irccloud.com/x-rbnhjcjzilzubfsf) left irc: Remote host closed the connection18:01
Alex_Gaynor (u1246@gateway/web/irccloud.com/x-ecpcksyavcturcto) left irc: Remote host closed the connection18:01
alexsuraci (u1962@gateway/web/irccloud.com/x-npbnghqxrnvkbiab) left irc: Remote host closed the connection18:01
bfirsh (u1308@gateway/web/irccloud.com/x-vmxjlfrnzdgglvvz) left irc: Remote host closed the connection18:01
bfirsh (u1308@gateway/web/irccloud.com/x-gvqrvcwsvublfhee) joined #pypy.18:03
fijal"I know that the PyPy guys read this subreddit, so I have some quick questions"18:04
twomashi (~Adium@190.2.44.65) joined #pypy.18:06
fijalpfffffff18:07
alexsuraci (u1962@gateway/web/irccloud.com/x-npxtfptmiuuxjgoa) joined #pypy.18:08
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Textual IRC Client: http://www.textualapp.com/18:08
fijalwhere is alex when you need one...18:10
Alex_Gaynor (u1246@gateway/web/irccloud.com/x-duxsckocffxwnqfz) joined #pypy.18:11
#pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services.18:11
ojii (~ojii@adsl-84-226-255-220.adslplus.ch) left irc: Quit: Leaving18:12
fijalAlex_Gaynor: pfff18:12
fijalAlex_Gaynor: look at my reddit comment18:12
Alex_GaynorI see18:12
fijalit's just that sub has a loop without a jit driver18:12
kenaan12fijal numpy-multidim 112944b13d49dd 15/pypy/module/micronumpy/test/test_numarray.py: more tests18:13
Alex_GaynorI odn't think that's teh issue18:13
fijalwe should do something more automatically18:13
fijalyes, it is18:13
Alex_Gaynoryou aren't looping over any python thing18:13
fijalthere is an optimization18:13
fijaldoes not matter18:13
Alex_Gaynorif you don't have a callable for sub, what's the issue?18:13
fijalyou do have a callable for sub18:13
fijalin case of json18:13
fijalhowever you don't call it18:13
fijalso far so good18:13
fijalhowever I believe you don't compile the regex18:14
fijalbecause it does not contain a loop18:14
dgl (~dgl@109.86.165.231) left irc: Quit: Computer has gone to sleep.18:15
Alex_Gaynoruh, ask armin I guess18:15
fijalAlex_Gaynor: regarding test_zjit, might be cooler to use the language we have in compile.py18:16
Alex_GaynorI feel like we had this conversation already once though18:16
fijalinstead of constructing them by hand18:16
fijaland that would address some issus18:16
Alex_Gaynorfijal: pff, definitely a good idea IMO, what way we only have to update one thing18:16
fijallike annotations18:16
Action: fijal gets some food and does that18:16
Alex_GaynorI made them share the same fake space a while ago, that's good I suppose18:16
fijalcause I don't feel like fixing all the signatures18:17
fijalI think we have an issue about regex or a point in planning/jit.txt18:17
Alex_Gaynorfijal: are you cahnging teh dtype API at all? (aka should I wait on comlex branch?)18:17
fijalnope18:17
fijaljust arrays always have shape18:17
Alex_Gaynork, cool18:17
JaRoel|4D (~jaroel|4d@2001:980:7133:1:e926:34dd:a9f:5969) joined #pypy.18:19
JaRoel|4D (~jaroel|4d@2001:980:7133:1:e926:34dd:a9f:5969) left irc: Remote host closed the connection18:19
JaRoel|4D (~jaroel|4d@sink.jaroel.nl) joined #pypy.18:19
kenaan12alex_gaynor extradoc 11386468696bbc 15/planning/jit.txt: this is done18:20
Alex_Gaynorfijal: FYI before you even try, using string builder in sub is a mess18:20
fijalthat's fine18:21
bgolai found a fix for the issue I reported yesterday with regexps (Issue923). i wrote a test for it and the tests are passing, should I just open a pull request? 18:21
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 252 seconds18:21
fijalbgola: sounds like a plan?18:21
bgolak18:22
Alex_Gaynorbgola: sounds great, thanks!18:22
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Quit: Leaving18:30
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.18:36
\toothrot (~mux@unaffiliated/htoothrot) joined #pypy.18:37
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.18:39
htoothrot (~mux@unaffiliated/htoothrot) left irc: Ping timeout: 276 seconds18:40
lmoura (~lmoura@186.215.206.130) left irc: Ping timeout: 240 seconds18:41
mwhudson (~mwh@linaro/mwhudson) joined #pypy.18:47
antocuni_ (~antocuni@host193-122-dynamic.0-79-r.retail.telecomitalia.it) left irc: Ping timeout: 260 seconds18:47
fschwidom (~fschwidom@46.114.93.141) left irc: Remote host closed the connection18:53
Rhy0litefijal: you only just discovered this about RMS?18:55
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/676 [12alex]18:56
ericflo (~ericflo@75.103.8.110) joined #pypy.18:57
Rhy0litefijal: At RMS's request, GCC is changing Red Hat Linux to Red Hat GNU/Linux18:57
Alex_Gaynorwhat a waste of paper work18:57
Alex_Gaynorfijal: FYI stepping away for a little bit, let me know if I have any TODOs now that I can work on complex again :)18:58
Action: Alex_Gaynor wants to get tickets to see ken jennings tonight18:59
Action: fijal is having utter fun writing a parser19:00
fijalpython is awesome!19:00
timonatorfijal: are you using some library for it?19:01
fijaltimonator: nope, just fun-by-hand19:01
timonatormhm19:01
etrepum (~bob@accessnat4.mochimedia.net) left irc: Remote host closed the connection19:04
etrepum (~bob@accessnat4.mochimedia.net) joined #pypy.19:04
jokeyfijal: it truly is :)19:08
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.19:13
santagadafijal: parser for what?19:13
fijalsantagada: for fun19:16
tomh (u501@gateway/web/irccloud.com/x-kizayexssboeecpg) joined #pypy.19:16
santagadafijal: I wanted to know what language/file format19:16
fijalyou'll see in a sec19:17
ronnyfijal: ok, actually --fulltrace might help you19:19
santagada (~leonardo@201.47.210.238.dynamic.adsl.gvt.net.br) left irc: Ping timeout: 240 seconds19:22
mitchellh (~mitchellh@c-69-181-107-107.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep19:25
Rhy0lite (dje@nat/ibm/x-yhcjgbcravrreemm) left irc: Quit: Leaving19:26
mitchellh (~mitchellh@c-69-181-107-107.hsd1.ca.comcast.net) joined #pypy.19:29
kenaan12alex_gaynor default 11993b01fd53d4 15/pypy/objspace/std/test/test_stdobjspace.py: fix a test that was still trying to use W_TypeObject.interplevel_cls19:32
stakkars_ (~tismer@i59F6B62A.versanet.de) joined #pypy.19:40
kkris (~kris@93-82-46-138.adsl.highway.telekom.at) left irc: Remote host closed the connection19:54
etrepum_ (~bob@accessnat4.mochimedia.net) joined #pypy.19:54
etrepum (~bob@accessnat4.mochimedia.net) left irc: Ping timeout: 260 seconds19:56
Nick change: etrepum_ -> etrepum19:56
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/544 [12alex]19:57
nickik (~nick@217.150.231.165) joined #pypy.20:03
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1072 [12alex]20:05
lizardo (~lizardo@189.2.128.130) left irc: Quit: Leaving20:07
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.20:08
CIA-2503ltratt 07roundup * 10#925/RPython arrays sometimes getting the wrong type?: 20:09
CIA-25[new] This patch to the RPython Converge VM was meant to be a simple code tidy-up:20:09
CIA-25https://github.com/ltratt/converge/commit/65cc1dc822c19df0 ... * 14https://bugs.pypy.org/issue92520:09
ltratt (~ltratt@84.93.138.38) joined #pypy.20:09
fijalnice20:09
Alex_Gaynorfijal: this RPython error?20:11
fijalhttp://paste.pocoo.org/show/499207/20:11
Alex_Gaynorpff20:12
Alex_Gaynorwhat ahve you done20:12
fijalI'm trying to make test_zjit nicer20:13
fijalbut now they're seeing too much I think20:13
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 248 seconds20:14
fijalit seems like a very weird annotator bug20:14
fijalspace.fromcache doesn't work20:14
fijalpffff20:20
fijalfeel like fighting with the weird lookuperror?20:20
Alex_Gaynornot really20:20
kenaan12fijal numpy-minilang 11ced3b41aaacb 15/pypy/module/micronumpy/: progress on supporting strange language for test_zjit20:21
kenaan12fijal numpy-minilang 11263d430ab0d3 15/pypy/module/micronumpy/: a bit of rpythonization20:21
Alex_Gaynoris this a branch from default or multidim?20:21
exarkunwon't hg tell you?20:22
Alex_Gaynorprobably if I give it magic voodoo20:23
fijalfrom default20:24
stakkars__ (~tismer@i59F6B62A.versanet.de) joined #pypy.20:25
fijalAlex_Gaynor: look at test_compile20:27
Alex_Gaynoron numpy-minilang?20:27
fijalyeah, commiting20:27
kenaan12fijal numpy-minilang 11349ca005d2ef 15/pypy/module/micronumpy/test/test_compile.py: oops, forgot to add a test20:28
Action: fijal ponders if he feels like fighting with this error20:28
stakkars_ (~tismer@i59F6B62A.versanet.de) left irc: Ping timeout: 258 seconds20:28
Alex_Gaynorfijal: what am I looking at here?20:28
fijalthe language20:29
Alex_Gaynorseems like a lot of work just to avoid needing to translate to run these tests20:29
fijalbut a lot of fun!20:30
mwhudsonthat suspected spam message on pypy-dev is ... different20:30
Alex_Gaynorfijal: pff, if you want to have fun writing a parser, go improve JSON decoding20:30
fijalAlex_Gaynor: people said the same about oparser20:30
fijal"a lot of work to just write more tests..."20:30
Alex_Gaynorwell opparser is useful20:30
fijalbelieve me it's worth it20:30
Alex_Gaynor;)20:30
fijalyes :)20:30
Alex_GaynorI'm just not so sure this is useful.  ATM we have good infrastructure for most of numpy stuff, so we don't do much performance stuff with it20:31
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 258 seconds20:33
fijalyes yes yes...20:34
Alex_Gaynorbtw -> is a silly operator for indexing20:34
fijaldon't you have stuff to do?20:35
fijalfix my lookup error20:35
Alex_Gaynoryes, and no20:35
Alex_Gaynori'm going to work on complex stuff now20:36
fzzzy_ (~donovan@nat/mozilla/x-gwaloaddmwsnkeri) joined #pypy.20:36
fzzzy (~donovan@nat/mozilla/x-nualopzjotsxpcuq) left irc: Read error: Connection reset by peer20:36
Nick change: fzzzy_ -> fzzzy20:36
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.20:37
intchanter (~intchante@c-76-27-100-38.hsd1.ut.comcast.net) left irc: Ping timeout: 260 seconds20:37
intchanter (~intchante@c-76-27-100-38.hsd1.ut.comcast.net) joined #pypy.20:39
lmoura (~lmoura@187.58.105.14) joined #pypy.20:41
kenaan12fijal numpy-minilang 11c729d657e525 15/pypy/module/micronumpy/: ok, not mix two spaces, leads to confusing errors20:42
ltratt (~ltratt@84.93.138.38) left irc: Quit: using sirc version 2.211+KSIRC/1.3.1220:45
lmoura (~lmoura@187.58.105.14) left irc: Read error: Connection reset by peer20:50
daniloaf (~daniloaf@187.114.217.96) joined #pypy.20:54
canta (~canta@77-20-123-240-dynip.superkabel.de) left irc: Quit: Leaving.20:55
fijal<* struct object_vtable { subclassrange_min=<pypy.rpython.normalizecalls.TotalOrderSymbolic object at 0x3f16f50>, subclassrange_max=<pypy.rpython.normalizecalls.TotalOrderSymbolic object at 0x3f16f90>, rtti=..., name=..., instantiate=... }>20:58
fijal(Pdb++) [graph for graph in graphs if 'compute' in str(graph)][0].startblock.ex\20:58
fijalits[0].args[1].value._obj.typeptr.name20:58
fijal<* array [ 'A', 's', 's', 'e', 'r', 't', 'i', 'o', 'n', 'E', 'r', 'r', 'o', 'r', '\x00' ]>20:58
Alex_Gaynorhaving fun?20:59
derdon (~derdon@pD9E1C7F0.dip.t-dialin.net) left irc: Remote host closed the connection21:01
lmoura (~lmoura@187.58.100.139) joined #pypy.21:06
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.21:09
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 244 seconds21:10
antocuni_ (~antocuni@host193-122-dynamic.0-79-r.retail.telecomitalia.it) joined #pypy.21:15
felipejcruz (~felipecru@177.98.0.102) left irc: Ping timeout: 252 seconds21:21
etrepum (~bob@accessnat4.mochimedia.net) left irc: Quit: etrepum21:21
fijalAlex_Gaynor: looks like it21:21
kenaan12fijal numpy-minilang 11f1d8e024a3ed 15/pypy/module/micronumpy/: pass the test_zjit finally21:21
Alex_Gaynorfijal: looks like what?21:22
fijallike having fun21:24
__pv (pav@xen.ihme.org) joined #pypy.21:24
rekamso (~textual@67.51.82.66) left irc: Ping timeout: 252 seconds21:24
Alex_Gaynorheh21:24
felipejcruz (~felipecru@177.98.0.102) joined #pypy.21:27
mitchellh (~mitchellh@c-69-181-107-107.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep21:35
mitchellh (~mitchellh@c-69-181-107-107.hsd1.ca.comcast.net) joined #pypy.21:36
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 248 seconds21:36
nirbheek_ (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 260 seconds21:39
felipejcruz (felipecru@177.98.0.102) left #pypy.21:43
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Ping timeout: 244 seconds21:43
fijalHAHAHA21:45
Action: fijal found a bug in pypy's strip21:45
fijalat least on RPython level21:45
Alex_Gaynorfijal: patches welcome?21:47
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/11821:49
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/69421:49
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/92121:49
fijalnow I wonder how hard would it be to reuse the ll_meta_interp graphs21:50
fijalbecause we don't have to regenerate them21:50
kenaan12fijal numpy-minilang 1141172dd9df6f 15/pypy/rpython/: a test and a fix. took a while :/21:50
kenaan12fijal numpy-minilang 11c75ffa281f05 15/pypy/module/micronumpy/: convert tests and some rpython fixes21:50
ronan (~ronan@host86-151-196-170.range86-151.btcentralplus.com) left irc: Ping timeout: 240 seconds21:53
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.21:54
tlynn (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) joined #pypy.21:55
eropple|Air (~ed@pool-74-104-187-201.bstnma.fios.verizon.net) joined #pypy.21:59
eropple|Air (~ed@pool-74-104-187-201.bstnma.fios.verizon.net) left irc: Client Quit22:01
eropple|Air (~ed@pool-74-104-187-201.bstnma.fios.verizon.net) joined #pypy.22:01
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) left irc: Quit: Leaving...22:03
kenaan12fijal numpy-minilang 11a9aa7bc83108 15/pypy/: reuse the meta interped graph22:10
Unhelpful (~quassel@rockbox/developer/Unhelpful) left irc: Ping timeout: 240 seconds22:10
fijalgood22:11
fijalAlex_Gaynor: I claim we'll use this left and right to debug what's going on22:11
kenaan12fijal numpy-minilang 117861ff043f50 15/pypy/module/micronumpy/: cleanup22:11
fijalespecially when we'll want to do SSE and other crap22:11
fijalit's "forward thinking"22:11
fijalnot "overengineeering"22:11
fijaland anyway22:13
fijalprobably enough for today22:13
nickik (~nick@217.150.231.165) left irc: Quit: Verlassend22:18
tilgovi (~randall@75.101.111.78) joined #pypy.22:22
tilgovi (~randall@75.101.111.78) left irc: Changing host22:22
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.22:22
_pyre_ (~stian@137.241.189.109.customer.cdi.no) joined #pypy.22:22
_pyre_Hey, noticed that speed.pypy.org was fixed. But speed the last few days have been heavily reduced by the looks of it. From 5x to 4.4x :(22:22
Unhelpful (~quassel@rockbox/developer/Unhelpful) joined #pypy.22:24
fijal_pyre_: it's not reduced22:28
fijalwe have new benchmarks22:28
fijalthat are below 5x22:28
fijalmostly22:28
fijalalso we upgraded to 2.7 for cpython22:28
fijalbut that's less of an impact22:28
Ademan (~yaaic@34.sub-97-159-164.myvzw.com) joined #pypy.22:28
_pyre_ow, ok. But wouldn't it be nice to also upgrade the results used on the older versions22:28
Unhelpful (~quassel@rockbox/developer/Unhelpful) left irc: Ping timeout: 240 seconds22:29
Ademanare there any low/mid tasks we want done before the sprint? I want to make myself useful this weekend.22:29
_pyre_Or atleast mension it was the result using diffrent benchmarks.22:29
fijal_pyre_: what do you mean?22:39
k_bx (~k_bx@94.244.19.62) left irc: Ping timeout: 260 seconds22:39
_pyre_I mean, the benchmark for 1.6, 1.5 etc22:39
fijalyes maybe22:39
fijalthat would put them on the main at least22:39
fijalI didn't do that however :)22:39
fijal_pyre_: I wouldn't be too attached to the number btw22:40
_pyre_I mostly just check in after merging in a branch to see what gains on it. It's nice to know when you want to make code for pypy as well22:42
_pyre_Big changes in the benchmarks or just some minor stuff?22:43
fijalwe uploaded sympy, chameleon and translation benchmarks22:46
fijalI'm a bit unsure if they're reflected in the total when I look at it22:46
fijalmaybe it's just 2.6 -> 2.7 move22:46
Ademanblegh chameleon...22:48
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.22:51
Unhelpful (~quassel@rockbox/developer/Unhelpful) joined #pypy.22:52
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 240 seconds22:55
tlynn (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) left irc: Ping timeout: 240 seconds22:59
stakkars_ (~tismer@i59F6B62A.versanet.de) joined #pypy.23:07
twomashi (~Adium@190.2.44.65) left irc: Quit: Leaving.23:08
ronan (~ronan@host86-151-196-170.range86-151.btcentralplus.com) joined #pypy.23:10
stakkars__ (~tismer@i59F6B62A.versanet.de) left irc: Ping timeout: 260 seconds23:10
etrepum (~bob@accessnat4.mochimedia.net) joined #pypy.23:11
_habnabit (~habnabit@python/site-packages/habnabit) left irc: Quit: ZNC - http://znc.sourceforge.net23:12
stakkars_ (~tismer@i59F6B62A.versanet.de) left irc: Quit: schnarch23:18
fzzzy (~donovan@nat/mozilla/x-gwaloaddmwsnkeri) left irc: Quit: fzzzy23:19
whitelynx (~whitelynx@li117-47.members.linode.com) left irc: Quit: Ex-Chat23:28
dmsk (~dmsk@62.61.145.89.generic-hostname.arrownet.dk) joined #pypy.23:29
antocuni_ (~antocuni@host193-122-dynamic.0-79-r.retail.telecomitalia.it) left irc: Ping timeout: 260 seconds23:29
_habnabit (~habnabit@python/site-packages/habnabit) joined #pypy.23:34
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.23:34
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/69423:35
mitchellh (~mitchellh@c-69-181-107-107.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep23:36
CIA-2503bgola 07roundup * 10#923/Regular expressions with optional look-ahead operations failing to match groups: I just opened a pull request that fixes this issue. https://bitbucket.org/pypy/pypy/pull-request/13/fixes-issue-923 * 14https://bugs.pypy.org/issue92323:36
mitchellh (~mitchellh@c-69-181-107-107.hsd1.ca.comcast.net) joined #pypy.23:39
antocuni_ (~antocuni@host193-122-dynamic.0-79-r.retail.telecomitalia.it) joined #pypy.23:44
dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) left irc: Quit: Leaving23:47
mtigas (~Adium@users.spokesman.com) left irc: Quit: Leaving.23:53
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection23:56
stakkars (~tismer@p5DDB5E98.dip.t-dialin.net) left irc: Read error: Connection reset by peer23:57
stakkars (~tismer@p5DDB5E98.dip.t-dialin.net) joined #pypy.23:58
--- Fri Oct 28 201100:00

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