#pypy IRC log for Friday, 2011-12-16

voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.00:06
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host00:06
voidspace (~anonymous@python/psf/voidspace) joined #pypy.00:06
nedbatCan someone help me understand this error:  Not Implemented: sandboxing for external function 'pypy__float2longlong'    I get this when using json inside the sandbox. 00:06
Alex_Gaynorthere's some external function float2longlong, which converts puts a floats bits in a longlong storage00:09
Alex_Gaynorthe sandbox doesn't either a) know it is safe00:09
Alex_Gaynoror b) have a safe impl of it00:09
Alex_GaynorI think it's safe, not sure how to tell teh sandbox about it though00:10
nedbatAlex_Gaynor: thx, is it ringing any bells why this worked before I pulled in the changes from the last few months?  Also, why would json need such an operation?00:10
Alex_GaynorI'm guessing it's used internally by something, I really have no idea though00:11
nedbatAlex_Gaynor: ok, just looking for ideas.00:11
Alex_Gaynorsorry, fijal might know00:11
nedbatAlex_Gaynor: I'm figuring on finding him in the morning...00:11
_pyre_ (~stian@137.241.189.109.customer.cdi.no) joined #pypy.00:11
Alex_Gaynormailing list is reccomended then00:12
Alex_Gaynorwaiting for the stars of US/Europe time to align isn't a great way forward IME00:12
nedbatAlex_Gaynor: I've had pretty good luck with it, but my duty cycle is lower than yours in the first place! :)00:12
tilgovi (~randall@173-13-150-22-sfba.hfc.comcastbusiness.net) joined #pypy.00:12
tilgovi (~randall@173-13-150-22-sfba.hfc.comcastbusiness.net) left irc: Changing host00:12
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.00:12
_pyre_Neither cpyext (it hangs out of nowhere, but runs fine in the pypy console :s), nor ctypes like me (perhaps because I'm pretty new to ctypes)00:14
nedbatAlex_Gaynor: actually, what has worked for me: send an email to the list, the fijal pings me on IRC in the morning with the answer!00:16
Alex_Gaynor:)00:17
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) joined #pypy.00:17
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 245 seconds00:18
etrepum (~bob@c-67-180-192-170.hsd1.ca.comcast.net) left irc: Ping timeout: 240 seconds00:21
nedbathmmm, just importing json causes the error.00:22
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds00:22
etrepum (~bob@108-212-65-164.lightspeed.sntcca.sbcglobal.net) joined #pypy.00:26
etrepum (~bob@108-212-65-164.lightspeed.sntcca.sbcglobal.net) left irc: Quit: etrepum00:34
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection00:37
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Ping timeout: 252 seconds00:39
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) left irc: Ping timeout: 244 seconds00:43
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) left irc: Remote host closed the connection00:45
jterrace (~jterrace@aegis.CS.Princeton.EDU) left irc: Quit: Leaving.00:46
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1470 [12amaury, py3k]00:50
rekamso (~textual@216.190.27.14) left irc: Ping timeout: 240 seconds01:12
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/74601:20
jterrace (~Adium@dynamic-oit-vapornet-b-153.Princeton.EDU) joined #pypy.01:23
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) joined #pypy.01:32
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/116101:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/64401:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/30601:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/29201:34
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/188901:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/63301:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/147101:34
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/75101:34
bbot213Exception: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/29201:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/82801:34
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 245 seconds01:41
mat^2 (~mathias@212.130.113.35) joined #pypy.01:41
e1ven (~E1ven@SQ7/ProjectLead/E1ven) joined #pypy.01:42
jterrace (~Adium@dynamic-oit-vapornet-b-153.Princeton.EDU) left irc: Quit: Leaving.01:46
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace01:57
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Ping timeout: 252 seconds01:59
nedbat (~nedbat@python/psf/nedbat) joined #pypy.02:25
tilgovi (~randall@nat-sonicnet.noisebridge.net) joined #pypy.02:28
tilgovi (~randall@nat-sonicnet.noisebridge.net) left irc: Changing host02:28
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.02:28
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/97902:32
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) left irc: Quit: quit02:32
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.02:51
dracman (~draco@212.255.36.109) left irc: Ping timeout: 252 seconds02:53
Nick change: PiotrSikora_ -> PiotrSikora02:59
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) joined #pypy.03:24
anish_ (~anish@123.201.199.129) joined #pypy.03:42
anish_ (~anish@123.201.199.129) left irc: Client Quit03:44
anish_ (~anish@123.201.199.129) joined #pypy.03:48
anish_ (~anish@123.201.199.129) left irc: Ping timeout: 240 seconds03:56
DanKluev (~root@84-205-ppp.vntc.ru) joined #pypy.03:56
anish_ (~anish@123.201.199.129) joined #pypy.04:06
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/304:11
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 244 seconds04:15
kenaan12arigo arm-backend-2 11c8a475c4557b 15/pypy/jit/backend/arm/runner.py: Kill duplicate function04:32
kenaan12arigo arm-backend-2 112f5bb75ff7c8 15/pypy/jit/backend/arm/assembler.py: Fix fail_boxes_count, which is one more than the highest index written. Add an assert checking that the point...04:32
kenaan12arigo arm-backend-2 11a770817aed83 15/pypy/jit/backend/arm/assembler.py: Sorry, nonsense.04:41
kushal (~kdas@fedora/kushal) joined #pypy.04:42
_pyre_ (~stian@137.241.189.109.customer.cdi.no) left irc: Quit: Leaving.04:42
kushal (~kdas@fedora/kushal) left irc: Quit: Leaving04:49
Arach (~arach@2.95.83.234) left irc: Ping timeout: 252 seconds04:52
kenaan12arigo default 11d9b372cf25b0 15/pypy/rlib/longlong2float.py: Fix for sandbox.04:52
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds04:55
kenaan12arigo default 119307d48e5da0 15/pypy/jit/backend/x86/test/test_jump.py: Fix test: "width" => "get_width()"04:55
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) left irc: Quit: asmeurer__04:57
kenaan12arigo buildbot 11b5e0f9f4bc83 15/bot2/pypybuildbot/master.py: Don't run this server nightly as long as it hogs the same global lock at tannit and takes more than 5 hours to run...05:04
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) left irc: Quit: buildmaster reconfigured: bot disconnecting05:09
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) joined #pypy.05:10
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/17205:10
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/30705:10
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1890 [12arigo]05:12
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/645 [12arigo]05:12
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1162 [12arigo]05:12
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/634 [12arigo]05:12
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/829 [12arigo]05:12
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/752 [12arigo]05:12
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1472 [12arigo]05:12
DanKluev (~root@84-205-ppp.vntc.ru) left irc: Ping timeout: 240 seconds05:13
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.05:26
DanKluev (~root@109-109-211-093-xdsl.vntc.ru) joined #pypy.05:29
rekamso (~textual@c-71-197-238-154.hsd1.wa.comcast.net) joined #pypy.05:43
dgl (~dgl@109.86.165.231) joined #pypy.06:14
dgl (~dgl@109.86.165.231) left irc: Client Quit06:14
mat^2 (~mathias@212.130.113.35) left irc: Ping timeout: 248 seconds06:21
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1472 [12arigo]06:21
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/829 [12arigo]06:23
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/645 [12arigo]06:35
durin42 (~durin@adium/durin42) left irc: Ping timeout: 255 seconds06:39
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1162 [12arigo]06:41
durin42 (~durin@adium/durin42) joined #pypy.06:41
dgl (~dgl@109.86.165.231) joined #pypy.06:46
Nick change: Gulaway -> Gulopine06:57
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Quit: ZNC - http://znc.in07:11
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/634 [12arigo]07:12
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.07:12
fijal (~fijal@197.174.15.134) joined #pypy.07:16
fijalhi07:17
Nick change: Gulopine -> Gulaway07:23
dracman (~draco@212.255.47.152) joined #pypy.07:23
mvt (~mvantelli@87.213.45.85) joined #pypy.07:24
squiddy (~squiddy@f053084242.adsl.alicedsl.de) joined #pypy.07:31
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.07:36
mwhudson (~mwh@linaro/mwhudson) left irc: Ping timeout: 245 seconds07:38
JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy.07:39
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 248 seconds07:40
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1890 [12arigo]08:12
fijalFAIL08:14
rekamso (~textual@c-71-197-238-154.hsd1.wa.comcast.net) left irc: Quit: Computer has gone to sleep.08:18
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy.08:19
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds08:24
jml (~jml@mumak.net) left irc: Quit: Coyote finally caught me08:36
jml (~jml@mumak.net) joined #pypy.08:40
jml (~jml@mumak.net) left irc: Changing host08:40
jml (~jml@canonical/launchpad/jml) joined #pypy.08:40
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.08:42
CIA-1303dongying 07roundup * 10#965/xml.etree.ElementTree says 'unknown encoding' of a regular encoding: 08:49
CIA-13[new] I've been trying to parse xml string using python, codes following:08:49
CIA-13#-*- coding: utf-8 -*-08:49
CIA-13import xml.etree.ElementTree as xmlet ... * 14https://bugs.pypy.org/issue96508:49
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) joined #pypy.09:07
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) left irc: Changing host09:07
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.09:07
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection09:08
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/752 [12arigo]09:10
amaury_ (~amaury_@74.125.57.34) joined #pypy.09:10
mattip (~chatzilla@bzq-79-183-120-145.red.bezeqint.net) joined #pypy.09:15
mattipfijal: I'm messing around with learning buildslave at the moment, I won't be getting back to broadcast till ~Monday09:16
Arach (~arach@128.75.184.8) joined #pypy.09:16
fijalmattip: ok, cool09:17
ojii (~ojii@84.226.127.220) joined #pypy.09:17
fijalmattip: what are you doing with buildslave?09:17
mattipTrying to contribute a buildslave for windows 32/6409:17
fijalcool09:17
fijalvery cool actually :)09:18
mattipOne that doens't pop up dialog boxes :)09:18
fijalindeed :)09:18
fijalAlex_Gaynor: hey, you promised a memcopy in case arrays are contiguous09:19
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy09:19
Alex_Gaynorfijal: you promised an is_contigious method on arrays/slices, is this actually a bottleneck?09:20
fijalno, but it prevents me from trying a different method09:20
Alex_Gaynora different method for what?09:20
fijalfor laplace09:20
mattipI did actually write an is_contiguous on the numpypy-is_contiguous branch09:21
fijalAlex_Gaynor: see, why not review that?09:21
Alex_Gaynorah, ok I'll take a look09:21
fijalAlex_Gaynor: I would like to do something *Exactly* like C09:22
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.09:22
Alex_Gaynor?09:22
fijalwhich would be to copy array instead of allocating a new one each iteration09:22
Alex_Gaynorarray.copy()09:22
Alex_Gaynorit already uses memcpy if it's  aconcrete array09:22
fijalthat creates a new one09:22
fijalthat's precisely what I'm trying to avoid09:22
fijalI want to do an inplace copy09:22
fijallike u[:] = old[:]09:22
fijalor u[:] = old09:23
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.09:24
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.09:24
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.09:24
fijalAlex_Gaynor: feel free to wait for refactor-signature merge btw09:24
Alex_Gaynorsure09:24
fijalanyway, I'm now 3 yaks down the road09:25
Alex_Gaynoryes, line up the yaks nicely and shave them09:25
Alex_Gaynorhow's signature refactor coming anyways?09:26
fijalit depends what's the feature list09:26
Alex_Gaynorall current stuff + sharing09:27
fijalif I want to merge it now, it's just a matter of reintroducing broadcasting and flatiter09:27
fijalsharing is done09:27
fijalonly for array iterators, not for views09:27
fijalbut we can experiment with that09:27
Alex_Gaynorso iterator sharing, but not array sharing09:27
fijalyes09:27
Alex_Gaynoronce everything is merge ready I can do that, if you want09:27
fijalwell, it's not yet ready09:28
Alex_Gaynorsure09:28
fijaland I got to running tests with pypy -> figuring out it's tracing of super-long functions over and over -> trying to improve hooks to see which functions09:28
fijalso I'm quite a few yaks away09:28
fijalAlex_Gaynor: if you feel like helping, you can cleanup immutable_fields09:29
fijaltranslation explodes because name is apparently mutable09:29
Alex_Gaynorwhat's wrong with them?09:29
fijal(I don't think it is)09:29
Alex_Gaynorname?09:29
fijalImmutableConflictError: field 'name' is defined mutable in class <InstanceRepr for pypy.module.micronumpy.interp_ufuncs.W_Ufunc>, but listed in _immutable_fields_ in subclass <InstanceRepr for pypy.module.micronumpy.interp_ufuncs.W_Ufunc1>09:29
Alex_GaynorI'll take a look09:30
Alex_Gaynorfijal: oh, of course, you have it on attrs in the baseclass, but not immutable fields09:30
Alex_Gaynorit needs to be marked immutable on the base class with it09:30
Alex_Gaynoryou can't have an attribute that isn't listed in immutable on a base class, but is on a subclass09:31
fijaloh09:32
fijalAlex_Gaynor: if I allocate a new array each iteration09:37
fijal(in C)09:37
fijalI get 2.1s vs 3s on pypy's numeric_309:37
fijalso it's not bad09:37
Alex_Gaynorhow much is it with no allocation?09:37
fijal1.4s09:37
Alex_Gaynorso sharing/asm generation is more of the difference09:38
Action: fijal is confused09:38
fijalwhat are you saying?09:38
Alex_Gaynorjust that generating better assembler actually makes more of a difference than the allocation here09:39
Eventh (evenwiik@caracal.stud.ntnu.no) joined #pypy.09:39
fijalstill09:40
fijalnot 2x difference09:40
fijalthat is fine, our assembler is quite crap09:40
fijalI'm fine with saying "we're 30% slower than a handcoded C solution"09:40
Alex_GaynorI'm happier saying "we win" though :D09:40
Alex_Gaynorbut yes, even 30% is not too bad09:40
Alex_Gaynorsince we know we can improve quite a bit09:41
fijaland this is a bit silly example09:41
fijalon a more complex one the difference will get smaller09:41
Alex_Gaynorwell, it's not our example, which is why I think this is great09:42
anish_ (~anish@123.201.199.129) left irc: Ping timeout: 245 seconds09:43
kenaan12fijal better-jit-hooks 11a6fd97f41c91 15/pypy/rlib/jit.py: I'm sure this is not very intentional09:46
fijalAlex_Gaynor: neat idea09:49
Alex_Gaynor?09:49
fijalif you have different view iterators09:49
fijalthey can't be shared in general, because they traverse different space, right?09:49
Alex_Gaynordefine different09:49
Alex_Gaynordifferent base array09:49
fijalno09:49
fijallike09:49
Alex_Gaynoror different shape/strides09:49
fijalu[1:]09:50
fijalu[:-1]09:50
fijalfor example09:50
fijalthey can't have different shape09:50
Alex_Gaynorthere you can share arrya ptr09:50
fijalbecause it would explode09:50
Alex_Gaynorbut not iterator09:50
fijalright09:50
fijalbut still, the iterator would overflow dimensions at the same point09:50
fijalso you can't share a concrete offset09:50
fijalbut you can share guards09:50
Alex_Gaynorif you are clever, yes.09:50
Alex_Gaynorlets get array and iterator first09:51
Alex_Gaynorguard is harder I think09:51
fijalno09:51
Alex_Gaynorwell, not totally, but still09:51
Alex_Gaynorother stuff is more valuable anyways, I think09:51
fijal[citation needed]09:52
fijalguards are expensive09:52
anish_ (~anish@123.201.199.27) joined #pypy.09:52
Alex_Gaynorof couse09:52
voidspace (~voidspace@python/psf/voidspace) joined #pypy.09:52
fijalvoidspace: hi09:53
amaury_ (~amaury_@74.125.57.34) left irc: Read error: Operation timed out10:03
jaredw (~jared@122-59-91-107.jetstream.xtra.co.nz) joined #pypy.10:03
mattipstupid question from someone who doesn't really understand x86 registers10:05
mattiponce all the pointers are in place,10:05
mattipshouldn't the iterator just increment an address by the next offset?10:06
mattipThen the iterators could be shared if the underlying storage structure is the same, no?10:06
fijalyes, but that's not how RPython really works10:08
fijalI mean it's possible and even probably not very hard10:08
fijalbut we need to teach JIT some primitives10:09
mattipok, thanks.10:10
fijalit's possible but also relatively boring10:11
fijalI also think it's too early10:11
fijalbecause it breaks some abstractions10:11
fijallike you will no longer be able to use array API and whatnot10:11
Alex_GaynorDavid told me ptr addition isn't reccomended for x86 anym10:11
Alex_Gaynorreo10:11
Alex_Gaynoryou are supposed to use indexes10:12
ronan (~ronan@host86-142-249-87.range86-142.btcentralplus.com) left irc: Ping timeout: 240 seconds10:12
bbot2_ (~bbot2@bzq-79-183-120-145.red.bezeqint.net) joined #pypy.10:12
bbot2_ (~bbot2@bzq-79-183-120-145.red.bezeqint.net) left irc: Remote host closed the connection10:12
fijalit was in 1996 ;-)10:14
kenaan12fijal refactor-signature 1106599e469d83 15/pypy/module/micronumpy/interp_ufuncs.py: mark this as immutable as well10:18
bbot2_ (~bbot2@bzq-79-183-120-145.red.bezeqint.net) joined #pypy.10:18
amaury_ (~amaury_@74.125.57.34) joined #pypy.10:23
mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.10:24
mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host10:24
mfoord (~anonymous@python/psf/voidspace) joined #pypy.10:24
mattipthen I don't feel so bad, the last time I had to really DO anything in assembler was 1981 or so.10:24
anish_ (~anish@123.201.199.27) left irc: Quit: Leaving10:27
bbot2_Started: 15http://localhost:8099/builders/pypy-c-jit-linux-x86-32/builds/0 [12theboss]10:28
fijaltheboss?10:31
bbot2_4Failure: 15http://localhost:8099/builders/pypy-c-jit-linux-x86-32/builds/0 [12theboss]10:33
Action: mattip blushing10:39
mattipI didn't know a local master sends messages here. Where else is it broadcasting my foolish attempts to?10:40
Eventh (evenwiik@caracal.stud.ntnu.no) left irc: Quit: Lost terminal10:40
fijalwait a sec10:41
fijalwhy do you need a local master?10:41
mattipI won't.10:41
fijalok10:42
fijaljust testing?10:42
mattipBut I wanted to test out the system, learning and all.10:42
apanda (~apanda@59.178.223.182) joined #pypy.10:42
fijalif you copied pypy config then it'll report here10:42
fijalit's fine btw10:42
JStoker (jstoker@unaffiliated/jstoker) left irc: Ping timeout: 244 seconds10:42
apanda (apanda@59.178.223.182) left #pypy.10:42
voidspace (~voidspace@python/psf/voidspace) left irc: Quit: Uhm... gotta go10:43
Nick change: mfoord -> voidspace10:43
mattipdepends on your definition of fine. Mine includes "paying for making a mistake"10:45
fijalit's fine that bots spams here :)10:46
fijalthat's what I wanted to say10:46
mattipOK, thnks.10:46
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) left irc: Quit: asmeurer__10:51
arigato (~arigo@89.204.138.157) joined #pypy.10:52
fijalhi armin10:53
khs (~khs@151.43.202.84.customer.cdi.no) left irc: Quit: Leaving10:55
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.10:55
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) left irc: Remote host closed the connection10:56
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.10:56
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/17210:56
arigatohi11:03
bbot2_ (~bbot2@bzq-79-183-120-145.red.bezeqint.net) left irc: Quit: buildmaster reconfigured: bot disconnecting11:03
fijalarigato: did you make decaying counter default?11:04
arigatofijal: that one is "your fault" I think:11:04
arigatohttp://buildbot.pypy.org/summary/longrepr?testname=test_random_effects_on_stacklet_switch&builder=own-linux-x86-32&build=1890&mod=jit.codewriter.test.test_call11:04
antocuniarigato: why did you remove the new benchmark from the nightly scheduler?11:04
arigatofijal: yes11:04
fijalarigato: it I think did beak quite some benchmarks11:04
antocunithey acquire a different lock than the tannit one, if I did things correctly11:04
arigatoantocuni: because it took tannit's lock for 5h3011:04
arigatoyou didn't so far :-)11:04
arigatolook at tonight's waterfall11:05
antocuniyes, I'm doing11:05
arigatofijal: yes, I know11:05
fijalarigato: ok11:05
fijalyes, indeed the test_random effects looks like my fault11:05
fijallet me fix it11:05
antocuniand I see that at around 3:00 "run benchmarks on top of pypy-c" was running on both tannit and speed-python11:05
arigatohow do you know?11:06
antocunibecause there is no gap?11:06
arigatoantocuni: the waterfall is not really useful: a "run benchmarks" box can also mean "waiting to run benchmarks"11:06
antocuniah11:07
antocuniI think that in some older version of buildbot it really displayed a gap11:07
fijaluh11:07
arigatoantocuni: :-(11:08
antocuniarigato: so, how do YOU know that it kept the lock? :-)11:08
jaredw (~jared@122-59-91-107.jetstream.xtra.co.nz) left irc: Ping timeout: 240 seconds11:08
arigatoantocuni: well, because at 7am this morning it didn't finish on tannit11:09
fijalhum11:09
fijaldo you know hotpy is still alive?11:09
mikefcthere was a blog post in the last month or so i thought.11:09
arigatoantocuni: another way to see it is that all other runners that should start at 3:00 where waiting11:10
fijalmikefc: impressive11:10
kenaan12fijal refactor-signature 118ac7503b0bf3 15/pypy/module/micronumpy/: Refactor a bit - now ConcreteArray is something with strides, everything else has no strides11:10
fijalmikefc: I wonder what is he aiming to do with it11:10
mikefcfijal: he was quite heavily asking for money to merge hotpy2 in with cpython11:10
antocuniarigato: true11:10
fijalmikefc: would they let him in the first place?11:10
antocuniarigato: actually, I didn't restart the master after I checked in the updated code for locks11:11
antocuniI just did a reconfigure11:11
antocunimaybe it's because of that11:11
arigatoah, bah11:11
arigatoyes, likely11:11
Action: antocuni never really understood what can be reconfigured and what need to be restarted11:11
arigatoI tried a reconfigure after commenting out the line11:11
arigatoand that failed completely11:11
mikefcThe HotPy(2) project aims to bring the optimisations developed and assessed in the original HotPy project to CPython.11:11
mikefcIt is expected that the resulting HotPy(2) interpreter will be about three times faster than the current CPython interpreter for pure Python code.11:11
mikefcThe HotPy(2) project is currently looking for funding.11:11
mikefchttps://sites.google.com/site/makingcpythonfast/11:11
arigatoantocuni: I think a very good approximation is: reconfig does not work, don't use11:12
antocunieh :-)11:12
Action: antocuni --> lunch11:12
fijal10 man months of work11:13
fijalpretty agressive schedule, but then again, there are no promises in terms of speedups11:14
fijalarigato: how much effort was psyco?11:15
kenaan12arigo default 11179263e7dd58 15/pypy/module/pypyjit/test_pypy_c/test_generators.py: Fix: again only look at the "loop" part, ignoring the "preamble" part.11:15
arigatofijal: note that some other benchmarks seem to have improved by the counters-decay branch...11:16
mikefcdid the original hotpy ever get out of the "it's my thesis" stage?11:16
fijalarigato: yes, I know11:16
fijalbut some slowed down 2x11:16
arigatofijal: a bit no clue how raytrace manages to loose that much11:16
fijalmikefc: the hotpy(2) repo is empty11:16
mikefcfijal: he's waiting for funding :)11:17
fijalmikefc: believing him that 10 months can give you a decent speedups in Python including a JIT would mean that either we're all dummies11:17
fijalor pypy's approach is broken11:17
arigatofijal: bah? it looks like the default number of "decay_halflife" is just so that it gets bad results11:19
fijalarigato: pfff11:19
fijalpfff11:19
arigatoby tweaking the value a bit, hop, it's fast again11:19
fijalas in everything else works much better?11:19
fijal:/11:19
arigatowell it makes no difference for raytrace11:19
arigatoas expected11:19
arigatoI suppose it's hitting just a bad point11:20
arigatothen it ends up compiling what should be the main loop as a bridge, or something like that11:20
arigatospambayes slows down for some other reason that I cannot figure out11:24
fijalspambayes is heavily regex driven11:29
verte (~verte@python/site-packages/verte) joined #pypy.11:29
fijaland not all of those regexes are compiled I think11:29
fijalnice, my branch segfaults :/11:30
amaury_ (~amaury_@74.125.57.34) left irc: Ping timeout: 240 seconds11:31
RiaanFV (~riaanfv@41-133-33-124.dsl.mweb.co.za) joined #pypy.11:34
arigatospambayes has a much longer warm-up time11:36
hakanardo__ (~ardo@putten.maths.lth.se) left irc: Read error: Connection reset by peer11:36
arigatoeven with decay_halflife=011:36
verte (~verte@python/site-packages/verte) left irc: Remote host closed the connection11:39
verte (~verte@python/site-packages/verte) joined #pypy.11:39
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/30711:42
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Ping timeout: 240 seconds11:43
arigato (~arigo@89.204.138.157) left irc: Quit: See you11:43
fijaloh11:43
fijalwell, I forgot to implement a feature :)11:43
bgola (~bgola@c9518082.virtua.com.br) left irc: Ping timeout: 244 seconds11:45
RiaanFV (riaanfv@41-133-33-124.dsl.mweb.co.za) left #pypy ("Leaving").11:47
_pyre_ (~stian@137.241.189.109.customer.cdi.no) joined #pypy.11:48
bgola (~bgola@c9518082.virtua.com.br) joined #pypy.12:00
JStoker (jstoker@unaffiliated/jstoker) joined #pypy.12:00
verte (~verte@python/site-packages/verte) left irc: Quit: ~~~ Crash in JIT!12:05
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) joined #pypy.12:05
h0h0 (c010b8ae@gateway/web/freenode/ip.192.16.184.174) joined #pypy.12:11
fijalmeh :/12:11
fijalhpk: is your java doing anything interesting on tannit?12:11
fijalor just screwing my benchmarks?12:11
fijalhpk: and can I make it stop?12:13
nedbat (~nedbat@python/psf/nedbat) joined #pypy.12:13
khs (~khs@dhcp-102-205.idi.ntnu.no) joined #pypy.12:14
fijalAlex_Gaynor: ok, so sharing iterators gives us next to nothing :)12:36
derdon (~derdon@p5DE8BB92.dip.t-dialin.net) joined #pypy.12:50
nedbatbranch management advice?  I have a few changes on my sandbox branch, and need the fix in https://bitbucket.org/pypy/pypy/changeset/d9b372cf25b0 .  Should I a) merge my branch to trunk, b) merge that fix to my branch, c) make the same changes on my branch?  I'd like to be working with stable code if I could.12:51
fijal (~fijal@197.174.15.134) left irc: Ping timeout: 276 seconds12:53
jnoller (~jnoller@vpn.nasuni.com) joined #pypy.12:57
_pyre_ (~stian@137.241.189.109.customer.cdi.no) left irc: Quit: Leaving.13:00
mattipany pypy buildbot experts?13:04
mattipI have a local buildbot master on one box13:04
mattipand a slave on another.13:04
mattipBut the slave is reported with "Not used by any Builders"13:05
mattipHow do I tell the master that I want it to build win32 / win64 targets?13:05
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) left irc: Quit: buildmaster reconfigured: bot disconnecting13:09
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) joined #pypy.13:09
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/4 [12antocuni]13:10
mattipnevermind: we_are_debugging13:14
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.13:23
derdon (~derdon@p5DE8BB92.dip.t-dialin.net) left irc: Remote host closed the connection13:27
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.13:32
JStoker (jstoker@unaffiliated/jstoker) left irc: Ping timeout: 252 seconds13:34
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.13:37
JStoker (jstoker@unaffiliated/jstoker) joined #pypy.13:45
arigato (~arigo@89.204.155.23) joined #pypy.13:49
e1ven (~E1ven@SQ7/ProjectLead/E1ven) left irc: Read error: Connection reset by peer14:01
mattip (~chatzilla@bzq-79-183-120-145.red.bezeqint.net) left irc: Quit: ChatZilla 0.9.87 [Firefox 8.0/20111115183158]14:08
ojiiwow pypy just filled up my 16GB of RAM when I tried to translate it...14:12
whitelynx (~whitelynx@63.241.75.144) joined #pypy.14:16
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) joined #pypy.14:22
nedbat (~nedbat@python/psf/nedbat) left irc: Read error: Connection reset by peer14:31
nedbat (~nedbat@python/psf/nedbat) joined #pypy.14:31
WildChild (~wildchild@modemcable165.184-57-74.mc.videotron.ca) left irc: Remote host closed the connection14:32
nirbheekojii, that's a first14:33
nirbheekafair14:33
azanella (~azanella@189.6.80.207) joined #pypy.14:51
ojiinirbheek, that shouldn't happen, should it?14:57
h0h0 (c010b8ae@gateway/web/freenode/ip.192.16.184.174) left irc: Quit: Page closed15:05
ronnyojii: when did it fill it up? afair there is a certain gcc version with a bug that can cause that15:13
exarkungcc 4.2, I think.15:13
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace15:16
ojiiI have gcc gcc (Ubuntu/Linaro 4.6.1-9ubuntu3) 4.6.15:20
ojiironny, not sure15:20
ojiihad to kill it15:20
ojiisince it almost killed my box15:20
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Quit: Leaving15:26
rguilleberthttp://mail.python.org/pipermail/python-dev/2011-December/115004.html15:31
rguillebertisn't it already implemented on PyPy ?15:34
exarkunIt kinda sounds like it, yea.15:34
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.15:34
Action: exarkun didn't read the code15:34
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/4 [12antocuni]15:34
kkris (~kris@80-123-45-172.adsl.highway.telekom.at) joined #pypy.15:37
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) joined #pypy.15:42
Kaskuka (~John@osbk-4d08b9ca.pool.mediaWays.net) left irc: Ping timeout: 240 seconds15:42
Shanita (~John@osbk-4db060c4.pool.mediaWays.net) joined #pypy.15:42
fijal (~fijal@41.54.125.202) joined #pypy.15:55
dgl (~dgl@109.86.165.231) left irc: Read error: Connection reset by peer15:55
fijalhi15:55
fijalnedbat: you can merge default -> your branch any time you want15:57
nedbatfijal: that's the route I chose.15:57
G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving.16:00
mvt (~mvantelli@87.213.45.85) left irc: Quit: Leaving16:03
bgola (~bgola@c9518082.virtua.com.br) left irc: Ping timeout: 255 seconds16:10
bgola (~bgola@c9518082.virtua.com.br) joined #pypy.16:10
jterrace (~jterrace@aegis.CS.Princeton.EDU) joined #pypy.16:13
JStoker (jstoker@unaffiliated/jstoker) left irc: Ping timeout: 240 seconds16:16
bgola (~bgola@c9518082.virtua.com.br) left irc: Ping timeout: 252 seconds16:18
lmoura (lmoura@gateway/indt/session) left irc: Ping timeout: 240 seconds16:19
bgola (~bgola@c9518082.virtua.com.br) joined #pypy.16:20
voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.16:22
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host16:22
voidspace (~anonymous@python/psf/voidspace) joined #pypy.16:22
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 244 seconds16:25
jterracefijal: merging fromstring to numpypy got pycollada tests up to 31/46 (from 23/46)16:26
jterracethings failing now: string dtype, concatenate, vdot, isnan, cumsum, asmatrix16:26
fijalgood, getting somewhere16:26
jterraceyep16:26
fijalwhat's next?16:26
jterracewell i thought concatenate16:26
jterracebut the way i did it in app-space is not the right way16:26
jterraceit really should be at interp level16:26
jterraceand we need some sort of axis iterator16:27
fijalaxis iterator would be cool16:28
fijalit'll solve a dot as well16:28
fijalfeel free to experiment with something16:28
jterraceyeah i will play around on a branch16:28
fijalmattip would want one as well16:28
jterraceyeah i know16:28
JStoker (jstoker@unaffiliated/jstoker) joined #pypy.16:33
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Remote host closed the connection16:35
JaRoel (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection16:36
aurojit (~apanda@59.178.215.53) joined #pypy.16:37
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.16:41
derdon (~derdon@p5DE89F38.dip.t-dialin.net) joined #pypy.16:42
fijal (~fijal@41.54.125.202) left irc: Read error: Connection reset by peer16:42
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.16:45
bgola (~bgola@c9518082.virtua.com.br) left irc: Ping timeout: 248 seconds16:48
ltratt (~ltratt@87.113.171.48) joined #pypy.16:49
ltrattHi all. Does RFFI support an external C function calling back to RPython functions - I'm sure it does, but I'm struggling to find an example of this.16:50
fijal (~fijal@197.104.192.172) joined #pypy.16:53
bgola (~bgola@c9518082.virtua.com.br) joined #pypy.16:56
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) joined #pypy.16:56
dgl (~dgl@109.86.165.231) joined #pypy.16:59
Elsvent (Elsvent@118-170-154-55.dynamic.hinet.net) joined #pypy.17:03
Elsvent (Elsvent@118-170-154-55.dynamic.hinet.net) left #pypy.17:03
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.17:13
pedronis_ (~pedronis@73-53.195-178.cust.bluewin.ch) joined #pypy.17:19
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) left irc: Ping timeout: 245 seconds17:20
gehho2 (~gehho2@X190178.ppp.dion.ne.jp) left irc: Ping timeout: 244 seconds17:25
ltratt (~ltratt@87.113.171.48) left irc: Quit: using sirc version 2.211+KSIRC/1.3.1217:27
lmoura (lmoura@gateway/indt/session) joined #pypy.17:28
ojii (~ojii@84.226.127.220) left irc: Quit: Leaving17:36
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.17:40
h0h0 (~k@f38016.upc-f.chello.nl) joined #pypy.17:42
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 255 seconds17:56
mtigas (~mtigas@users.spokesman.com) joined #pypy.18:00
dmalcolm (david@nat/redhat/x-iielqcpvxnvwcchr) joined #pypy.18:01
fijalAlex_Gaynor: btw, I know how to make VirtualViews nice18:01
Alex_Gaynoryou mean a slice of a virtual array?18:01
fijalyes18:01
fijalessentially what you do is you keep the chunks18:02
fijaland you only calculate strides for concrete arrays down the chain18:02
fijalso if you have say (a + b)[::2]18:03
fijalyou keep (0,0,2)18:03
fijalor whatever representation is as an argument to create_slice18:03
Alex_Gaynorcool18:03
fijaland then at force you propagate it down18:03
fijalso you have a good iterator for everyone and 0 overhead18:03
fijalcool stuff, surfing definitely helps with coding :)18:03
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds18:10
fijalAlex_Gaynor: iterator sharing does  not win much18:10
Alex_Gaynorfijal: do array sharing (should be easyish), that should help regalloc a ton18:10
Action: fijal does18:10
fijalyes sir :)18:10
fijalI should do broadcasting & flatiter at some point18:10
fijalbut that means I probably won't do view sharing18:10
fijaler18:11
fijalview iterator sharing18:11
fijalAlex_Gaynor: also scalar no longer calculates strides18:13
Alex_Gaynorthat makes quite some sense :)18:13
fijalAlex_Gaynor: do you have any opinions about the approach?18:20
fijallook at signature.py18:20
fijalgenerally at compile time we construct barebone signature18:21
fijaland if it matches (eq) we return an already compiled one18:21
jterracex.flat = 318:21
jterracedoesn't work btw, since it's read only18:21
Alex_Gaynorfijal: makes sense18:22
fijaljterrace: what that would do...?18:22
jterracex = numpy.array([[1,2],[3,4]])18:23
jterracex.flat = 718:23
jterracex == [[7,7],[7,7]]18:23
fijalnice18:23
fijalwell18:23
fijalx.flat[:] = 7 works as well?18:23
jterracebut you get a read-only error in numpypy18:24
jterracesure, but it works in cpython18:24
fijalit's relatively easy to implement as an obscure special case18:24
fijaljust change GetSetProperty to something that can also set the flat18:24
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace18:24
rekamso (~textual@63-228-127-70.dia.static.qwest.net) joined #pypy.18:28
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds18:34
DasIch_ (~dasich@p3E991D99.dip.t-dialin.net) joined #pypy.18:36
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.18:39
DasIch (~dasich@p4FFDED50.dip.t-dialin.net) left irc: Ping timeout: 240 seconds18:39
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection18:44
Circlefusion (~cf@74-138-199-109.dhcp.insightbb.com) left irc: Read error: Connection reset by peer18:44
Circlefusion (~cf@74-138-199-109.dhcp.insightbb.com) joined #pypy.18:49
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Ping timeout: 248 seconds18:51
lleonardo (~leonardo@190.94.99.125) joined #pypy.18:54
Nick change: lleonardo -> lcubo18:55
Shanita (~John@osbk-4db060c4.pool.mediaWays.net) left irc: Read error: Connection reset by peer18:58
__main__ (~main@adsl-99-173-15-158.dsl.pltn13.sbcglobal.net) left irc: Read error: Connection reset by peer18:59
kkris (~kris@80-123-45-172.adsl.highway.telekom.at) left irc: Ping timeout: 245 seconds19:02
Shanita (~John@osbk-4db060c4.pool.mediaWays.net) joined #pypy.19:05
fijalwhy pointers are not hashable :/19:08
fijalare addresses hashable?19:08
arigatothe problem is that prebuilt dicts containing pointers or addresses are never going to work19:09
__main__ (~main@adsl-99-173-15-158.dsl.pltn13.sbcglobal.net) joined #pypy.19:09
fijalright19:10
fijalok, just casted to address19:10
fijaleasy enough19:10
arigatoyes, if you're careful19:10
fijalwhy?19:11
fijalI'm not going to build anything prebuilt made out of them19:11
arigatomake sure you don't store addresses that may be freed19:12
fijalright19:12
arigatoor worse yet, freed and reused for something different19:12
fijalif I use them just as dictionary keys it's fine?19:12
fijalno19:12
fijalyes, precisely19:12
fijalanyway, no, I don't think it's possible19:12
fijalI can implement this as a list of pointers and just compare to all19:13
fijalas well19:13
arigatodoesn't change anything?19:13
arigatoyou're talking about non-gc pointers, right?19:13
fijalyes19:13
fijalbut they have to be alive19:14
fijalthis is a short-lived cache for assigning numbers19:14
fijalwell, if I keep them in a list they can stay as pointers obviously :)19:14
arigatothey won't be kept more alive if stored in a list or as addresses in a dict, of course19:14
fijalbut that's besides the point, I'm just being argumentative19:14
fijalall this stuff is definitely alive19:14
jaredw (~jared@122-59-91-107.jetstream.xtra.co.nz) joined #pypy.19:15
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds19:21
jnoller (~jnoller@vpn.nasuni.com) left irc: 19:25
Alex_Gaynorfijal: replace is TERRIBLE19:31
Alex_Gaynorfixing fixign fixing19:31
Shanita (~John@osbk-4db060c4.pool.mediaWays.net) left irc: Read error: Connection reset by peer19:32
gehho2 (~gehho2@ZO060028.ppp.dion.ne.jp) joined #pypy.19:32
fijal (~fijal@197.104.192.172) left irc: Read error: Connection reset by peer19:33
fijal (~fijal@41.52.35.237) joined #pypy.19:34
Kaskuka (~John@osbk-4db060c4.pool.mediaWays.net) joined #pypy.19:36
ericflo (~ericflo@75.103.8.110) joined #pypy.19:36
Alex_Gaynorfijal: http://paste.pocoo.org/show/521852/ look ok?19:44
fijalyop19:46
kkris (~kris@80-123-34-180.adsl.highway.telekom.at) joined #pypy.19:46
kenaan12alex_gaynor default 1110601f705a55 15/pypy/objspace/std/stringobject.py: make str.replace() not do tons of copying19:47
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/635 [12alex]19:47
aurojit (~apanda@59.178.215.53) left irc: Ping timeout: 240 seconds19:52
fijalAlex_Gaynor: http://paste.pocoo.org/show/521858/19:53
fijalarigato: do I have to do something special to keep adresses in a dict?19:53
Alex_Gaynorfijal: fix hash apparently19:53
arigatofijal: no, it either works or doesn't19:53
arigato(don't remember)19:54
fijalarigato: hehe :)19:54
fijalarigato: ok19:54
arigatoyes, it works19:54
arigatosee get_name_from_address() in pyjitpl.py for example19:54
fijalso why mine does not?19:57
fijalthose are equal19:57
fijalbut not hash-equal19:57
Alex_Gaynorfijal: if you look at the code in pyjitpl it doesn't do dict lookups for untranslated19:59
apanda (~apanda@59.178.212.226) joined #pypy.20:01
fijalAlex_Gaynor: cleanup will be necessary :)20:07
fijalcode duplication in signature20:07
Alex_Gaynorfijal: cleanup sounds good20:08
fijalnow that I have an idea what a view is :)20:08
fijalview is just a non-contiguous array20:08
fijalso we can remove tons of code20:08
Alex_Gaynorpossibly non-contigious20:08
fijalyes20:09
Alex_Gaynorbut yes, I see what you mean20:09
fijalso in general, arrays that don't own a datastore20:09
Alex_Gaynoryes20:09
fijalwe have tons of code that does not do much these days20:09
Alex_Gaynorif you give a basearray strides20:09
fijalbecause the assumption was implicit20:09
fijalbut we did not know :)20:09
Alex_Gaynoreverything becomes easy20:09
Alex_GaynorI think NumPyPy is some sort of zen buddhism quest to harmonize all the subclasses20:10
fijal (~fijal@41.52.35.237) left irc: Read error: Connection reset by peer20:10
fijal (~fijal@197.106.31.38) joined #pypy.20:14
fijalback20:14
fijalAlex_Gaynor: now I need a jit optimization20:14
Alex_Gaynorwhat do you need?20:14
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.20:14
fijaltwo consecutive getinteriorfields_raw should really be collapsed into one20:14
fijallazy fields :)20:14
kenaan12fijal refactor-signature 1182d0ce07b964 15/pypy/module/micronumpy/: sharing arrays20:15
Alex_Gaynorwe don't have lazy fields for any raw stuff20:15
fijalwell, then we should20:15
fijalI think for now it's good enough to say "any raw store flushes all the fields"20:15
Alex_GaynorI think so, we don't have any aliasing rules20:15
Alex_GaynorThough numpypy does obey the C rule of a char* can alias anything20:16
fijalI would be very skeptical, we might not20:16
fijalso it's just about "everything can alias everything"20:16
fijalfeel like implementing it?20:16
Alex_Gaynorany raw memory can alias any raw memory20:16
Alex_Gaynorsounds like a good starting point20:16
Alex_GaynorATM, no.20:16
fijal:(20:16
fijalotherwise the sharing arrays is pointless IMO20:17
Alex_Gaynorno it's not20:17
Alex_Gaynorbecause you only- use a single register20:17
Alex_Gaynor(and traces are more readable)20:17
fijalyou still have consecutive reads20:17
Alex_GaynorATM u + u + u + u will use 4 registers20:17
fijalyes, well20:17
fijalit'll use 820:17
Alex_Gaynorright20:17
fijal4 for u and 4 for values20:17
fijaland now it'll use 520:18
fijalI want 2 :)20:18
Alex_Gaynorpoint is, tons of registers20:18
fijalok, whenever you feel like it then20:18
Alex_GaynorI should finish complex20:18
fijalarigato: the test failure is obscure, but you're including the .c file, was that intentional?20:18
Alex_GaynorI guess I'll do that today, after I get replace compiled20:18
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.20:19
fijalcool20:19
arigato#include-ing?20:19
fijalarigato: yes20:19
arigato(yay, /me made sense out of a no-context-at-all question from fijal :-)20:20
fijalit includes both stacklet.h and stacklet.c20:20
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) left irc: Ping timeout: 244 seconds20:20
fijalsorry sorry20:20
fijalit took you years to learn my mind, but anyway, good progress :)20:20
arigatowe20:20
arigatoyes, that was intentionally an obscure way to let this work both non-translated and translated20:21
fijalokey20:21
fijalarigato: ok, my point is I don't think I changed much besides also hashing on it20:23
arigatowell, you changed something20:23
lahwranfijal: I'm curious; is pypy what you do for a living?20:23
fijaland it broke something else :)20:23
fijalarigato: yes, ok, I'll investigate a bit more20:23
fijallahwran: I'm trying :)20:23
arigatoif you're lost, you need to play with the hg history to make sure which revision did it, etc. etc.20:24
fijallahwran: why?20:24
fijalarigato: yes ok20:24
fijalarigato: I'm sure it's mine checkin that broke it20:24
lahwranbecause you seem to spend a lot of time on it and you appear very knowledgeable about it20:24
arigato(and find out that it was me all along, changes are :-)20:24
arigatoah ok20:24
lahwranparticularly due to the former, I thought that maybe you might be doing it for a living20:24
fijalbut it might be something else that was just triggered20:24
fijallahwran: armin spends more time and is more knowledgable, just not so vocal :)20:24
fijallahwran: I'm trying pretty hard to do pypy for a living20:24
arigatoactually I was20:25
arigatoat the university20:25
lahwranah20:25
fijalarigato: well, me as well :)20:25
arigatoyes (just not at a university :-)20:26
jaredw (~jared@122-59-91-107.jetstream.xtra.co.nz) left irc: Remote host closed the connection20:27
fijalright20:27
fijalI would if I had a degree :)20:28
fijallahwran: any other reasons why you ask?20:28
fijal(like do you want to hire me to do something in pypy :)20:28
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Quit: Leaving20:29
lahwranheh no20:29
fijal:)20:33
amaury_bah, expat encodings are broken20:37
amaury_bad design20:37
khs (~khs@dhcp-102-205.idi.ntnu.no) left irc: Ping timeout: 252 seconds20:38
lcubohi. I am new here. 20:42
mwhudson (~mwh@linaro/mwhudson) joined #pypy.20:42
fijallcubo: hi20:43
lcubosuppose that you want to start understand the code. which part do you think is the most important to view?20:43
fijaldepends what you're after :)20:44
lcuboor where I can start20:44
fijalwhat interests you?20:44
lcuboI like to makes things faster, but that pretty general I think20:44
amaury_I'd start with the description of Python objects20:44
fijalor the JIT20:44
amaury_hum20:45
fijalyes?20:45
fijallcubo: do you like assembler?20:45
lcubofijal I like but I dont code assembler since university (like three years ago)20:45
fijalthen you can work on say a better register allocator?20:46
fijalit's quite a bit of work though20:46
fijallcubo: if you like making things faster, I would suggest to read our blog series about writing interpreters in RPython20:49
fijalhave you read it by chance?20:49
fijalhttp://morepypy.blogspot.com/2011/04/tutorial-writing-interpreter-with-pypy.html20:49
fijalthis and Part 2 as well20:49
Alex_Gaynorfijal: http://paste.pocoo.org/show/521876/ confusing measurements20:49
fijallcubo: this should give you some idea what's going on20:49
fijalAlex_Gaynor: where did you take that from? :)20:50
Alex_Gaynorfijal: it's just a random example of time not showing teh full story20:50
fijalright20:50
Alex_Gaynorfijal: http://paste.pocoo.org/show/521877/ is kind of a giveaway20:50
fijalheh :)20:51
fijalwhy is our replace so slow?20:51
Alex_Gaynorfijal: http://paste.pocoo.org/show/521881/20:51
fijaland anyway, why is VanL reporting 8x?20:51
Alex_Gaynorfijal: it gets worse if you increase iterations20:51
Alex_Gaynor(I cut it down to work easier)20:52
fijalah20:52
fijalisn't replace using a trick?20:52
fijalif refcount == 1 replace in-place?20:52
Alex_Gaynoruh, I don't think so20:52
Alex_Gaynoranyway our replace is fine20:52
Alex_Gaynorlook at times in the last past20:52
lcubo_ (~leonardo@190.94.99.125) joined #pypy.20:52
lcubo (~leonardo@190.94.99.125) left irc: Read error: Operation timed out20:52
fijaloh, so it's add?20:52
Alex_Gaynoryes20:53
amaury_in CPython, I'm sure there are tricks like this for .upper() &co20:53
Alex_GaynorI don know how CPython add is fast here though20:53
fijaldo you happen to have a build with StringConcat object?20:53
Alex_GaynorI added `a = gtxt` so it can do inplace20:53
Alex_Gaynorno, I don't20:53
khs (~khs@dhcp-058217.wlan.ntnu.no) joined #pypy.20:53
fijalthat would not give you anything though20:53
fijalit might be that refcounting is our enemy here20:53
fijalanyway20:53
Action: fijal off to watch a movie20:53
fijalAlex_Gaynor: implement caches for getinteriorfield_raw20:53
Alex_Gaynorheh, ok20:53
fijal(just get btw)20:53
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/635 [12alex]20:54
lcubo_register allocator you said? I will investigate that. 20:56
fijal (~fijal@197.106.31.38) left irc: Ping timeout: 252 seconds20:58
apanda (~apanda@59.178.212.226) left irc: Ping timeout: 252 seconds20:58
apanda (~apanda@59.178.207.67) joined #pypy.21:05
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection21:10
jterrace (~jterrace@aegis.CS.Princeton.EDU) left irc: Quit: Leaving.21:16
arigato (~arigo@89.204.155.23) left irc: Quit: See you21:22
fmilo (~fmilo@12.219.160.90) left irc: Quit: Ex-Chat21:28
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) joined #pypy.21:30
rguilleberthttp://mrjoes.github.com/2011/12/15/sockjs-bench.html nice results21:31
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds21:32
lcubo_ (~leonardo@190.94.99.125) left irc: Ping timeout: 248 seconds21:45
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.21:47
rekamso"At first, I attempted to write one in Go, but it was not able to keep up with single instance of the sockjs-tornado running on PyPy." oops :)21:57
mikefcisn't Go supposed to have all sorts of super parallel thing-lets?21:58
ronnymaybe go is a practical joke22:00
fijal (~fijal@41.51.247.36) joined #pypy.22:00
fijaldmalcolm: you promised a gcc plugin :)22:00
fijaldmalcolm: also, it's a bit sad that fedora ships ~10% slower pypy22:00
dmalcolmfijal: I looked, but realized that what I really wanted is the reliable approach, and shadowstack appears to be that22:02
fijalbut it's slower22:02
dmalcolmthat way I don't get grief from the other side, for not using the distro compilation flags22:02
Alex_Gaynorfijal: we should go to david's GCC thing and tell them to add stack roots22:02
dmalcolmI get complaints either way22:02
fijaldmalcolm: chat tomorrow22:02
dmalcolmwill probably be feeding/changing diapers tomorrow :)22:03
h0h0 (~k@f38016.upc-f.chello.nl) left irc: Quit: leaving22:06
fijal (~fijal@41.51.247.36) left irc: Ping timeout: 252 seconds22:07
JStoker (jstoker@unaffiliated/jstoker) left irc: Ping timeout: 244 seconds22:07
rekamsoare there plans for PyPy sprints at Pycon?22:07
Alex_Gaynorrekamso: for sure22:08
rguillebertmikefc, I think it is but the compiler wasn't producing very fast code the last time I checked22:10
dmalcolmAlex_Gaynor: re the gcc plugin approach: the issue that gave me The Fear (TM) was: at what point in optimization passes should they be added?  I have a (trivial) script that can emit all local variables and their types, but at -O3 or whatever, locating all of them after the optimizer has done its thing seems prone to error22:12
dmalcolmthe caveat here is that I've mostly only been dealing with the really early passes of GCC; that's my comfort zone with GCC's insides22:13
dmalcolm(register allocation etc is still mostly a black box to me)22:13
dmalcolm(within gcc that is)22:13
ronnyhmm22:16
khs (~khs@dhcp-058217.wlan.ntnu.no) left irc: Quit: Leaving22:16
ronnywonder how hard a clang plugin for that would be22:17
stakkars_ (~tismer@i59F6BB3B.versanet.de) joined #pypy.22:18
JStoker (jstoker@unaffiliated/jstoker) joined #pypy.22:23
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/74722:34
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/17322:34
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/98022:34
aurojit (~apanda@59.178.214.32) joined #pypy.22:38
kkris (~kris@80-123-34-180.adsl.highway.telekom.at) left irc: Quit: Leaving.22:39
apanda (~apanda@59.178.207.67) left irc: Ping timeout: 245 seconds22:40
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.22:41
nedbat (~nedbat@python/psf/nedbat) left irc: Quit: Leaving22:42
stakkars__ (~tismer@i59F6BB3B.versanet.de) joined #pypy.22:51
izibi (~izibi@unaffiliated/izibi) joined #pypy.22:51
stakkars_ (~tismer@i59F6BB3B.versanet.de) left irc: Ping timeout: 252 seconds22:55
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Ping timeout: 276 seconds22:56
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) left irc: Ping timeout: 252 seconds23:01
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) joined #pypy.23:06
maxyz (~maxy@186.18.94.35) left irc: Ping timeout: 245 seconds23:11
maxyz (~maxy@186.18.94.35) joined #pypy.23:12
stakkars__ (~tismer@i59F6BB3B.versanet.de) left irc: Quit: schnarch23:15
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) left irc: Ping timeout: 252 seconds23:16
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 240 seconds23:22
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) left irc: Quit: quit23:23
whitelynx (~whitelynx@63.241.75.144) left irc: Quit: Ex-Chat23:29
dmalcolm (david@nat/redhat/x-iielqcpvxnvwcchr) left irc: Quit: Leaving23:43
derdon (~derdon@p5DE89F38.dip.t-dialin.net) left irc: Remote host closed the connection23:46
nlogax (~nlogax@unaffiliated/nlogax) joined #pypy.23:54
--- Sat Dec 17 201100:00

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