#pypy IRC log for Friday, 2012-01-20

azanella (~azanella@189.6.80.131) left irc: Ping timeout: 260 seconds00:01
Transformer (~Transform@ool-4a59e397.dyn.optonline.net) joined #pypy.00:01
icrazyhack (~horieyui@183.93.198.19) joined #pypy.00:02
Transformer (Transform@ool-4a59e397.dyn.optonline.net) left #pypy.00:02
antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) left irc: Ping timeout: 252 seconds00:06
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 252 seconds00:13
rguillebert (~rguillebe@89-40.5-85.cust.bluewin.ch) left irc: Quit: Quitte00:14
jamesaxli mannaged to get Mariadb and flask work with PYPY :)00:21
icrazyhack (~horieyui@183.93.198.19) left irc: Ping timeout: 252 seconds00:22
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace00:22
icrazyhack (~horieyui@183.60.101.84) joined #pypy.00:23
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Read error: Connection reset by peer00:23
timotimore what wes was saying: isn't that how numpypy already works?00:28
dripton (~dripton@pool-72-86-45-238.clppva.fios.verizon.net) left irc: Quit: Konversation terminated!00:28
timotimoonly compute the values at the end, combine all the ufuncs applied to them without making temporary arrays in between?00:28
dripton (~dripton@pool-72-86-45-238.clppva.fios.verizon.net) joined #pypy.00:30
Alex_Gaynortimotimo: yes, I just replied00:31
dripton (~dripton@pool-72-86-45-238.clppva.fios.verizon.net) left irc: Client Quit00:31
timotimogo you!00:31
timotimoAlex_Gaynor: would i come over as a prick if i asked you to please create less quoting pyramids?00:32
nedbat (~nedbat@python/psf/nedbat) joined #pypy.00:32
Alex_Gaynorno, but I won't do it anyways :)  make your mail client do the heavy work for you00:32
timotimohmm, maybe there's a thunderbird extension for that00:33
kvda (~kvda@124-169-134-221.dyn.iinet.net.au) joined #pypy.00:36
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.00:41
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.00:45
icrazyhack (~horieyui@183.60.101.84) left irc: Ping timeout: 252 seconds00:48
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) joined #pypy.00:52
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.00:54
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) joined #pypy.00:56
voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.00:57
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host00:57
voidspace (~anonymous@python/psf/voidspace) joined #pypy.00:57
icrazyhack (~horieyui@221.181.65.124) joined #pypy.00:58
__main__ (~main@adsl-99-173-15-158.dsl.pltn13.sbcglobal.net) left irc: Read error: Connection reset by peer01:09
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Ping timeout: 276 seconds01:10
icrazyhack (~horieyui@221.181.65.124) left irc: Ping timeout: 255 seconds01:15
__main__ (~main@adsl-99-173-15-158.dsl.pltn13.sbcglobal.net) joined #pypy.01:18
icrazyhack (~horieyui@221.192.139.107) joined #pypy.01:25
icrazyhack (~horieyui@221.192.139.107) left irc: Client Quit01:26
Tobu (~Tobu@unaffiliated/tobu) left irc: Read error: Operation timed out01:27
nettok (~quassel@190.148.221.40) joined #pypy.01:36
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/21001:38
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/3701:38
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace01:47
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy.01:47
Nick change: csenger -> csenger|zzz01:48
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/78101:49
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.01:54
squiddy (~squiddy@f053085115.adsl.alicedsl.de) left irc: Quit: Leaving01:55
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/193202:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/68302:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/33302:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/34302:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/68102:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/120002:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/86402:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/151702:00
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/79402:00
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) joined #pypy.02:15
ericflo (~ericflo@75.103.8.110) left irc: Quit: ericflo02:17
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection02:24
davisagli (~davisagli@davisagli.com) left irc: Excess Flood03:01
davisagli (~davisagli@davisagli.com) joined #pypy.03:02
Shanita (~John@osbk-4db15131.pool.mediaWays.net) joined #pypy.03:03
Kaskuka (~John@osbk-4db16df5.pool.mediaWays.net) left irc: Ping timeout: 276 seconds03:04
csenger|` (~user@dslb-084-058-172-023.pools.arcor-ip.net) joined #pypy.03:06
papercrane (~papercran@75.101.111.82) left irc: Ping timeout: 245 seconds03:07
csenger|zzz (~user@dslb-084-058-140-232.pools.arcor-ip.net) left irc: Ping timeout: 255 seconds03:10
csenger|` (~user@dslb-084-058-172-023.pools.arcor-ip.net) left irc: Client Quit03:11
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/102003:24
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/33303:25
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 252 seconds03:28
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.03:44
dracman (~draco@212.255.41.87) left irc: Ping timeout: 252 seconds03:45
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds03:48
dracman (~draco@212.255.37.212) joined #pypy.03:52
WildChild (~wildchild@modemcable123.12-202-24.mc.videotron.ca) left irc: Ping timeout: 240 seconds03:53
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.03:54
WildChild (~wildchild@modemcable123.12-202-24.mc.videotron.ca) joined #pypy.03:58
tilgovi_ (~randall@couchdb/developer/tilgovi) joined #pypy.04:03
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 260 seconds04:04
Nick change: tilgovi_ -> tilgovi04:05
bbot23Success: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/193204:26
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 252 seconds04:28
anish_ (anish@nat/redhat/x-pgigpwizgyzxivca) joined #pypy.04:38
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 276 seconds04:44
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.04:53
DasIch (~DasIch@p4FFDD0E2.dip.t-dialin.net) left irc: Quit: DasIch05:00
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Ping timeout: 245 seconds05:09
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!05:13
scitesy (~scitesy@d118-75-236-90.nap.wideopenwest.com) joined #pypy.05:14
nettok (~quassel@190.148.221.40) left irc: Remote host closed the connection05:22
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/151705:31
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/86405:32
scitesy (~scitesy@d118-75-236-90.nap.wideopenwest.com) left irc: Quit: Leaving05:33
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 252 seconds05:50
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/120005:51
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/68105:54
lambacck (~chris@d24-150-124-118.home.cgocable.net) left irc: Ping timeout: 240 seconds05:54
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/68305:56
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.05:59
DasIch (~DasIch@p4FFDD0E2.dip.t-dialin.net) joined #pypy.06:06
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) joined #pypy.06:06
DasIch (~DasIch@p4FFDD0E2.dip.t-dialin.net) left irc: Quit: DasIch06:17
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Ping timeout: 252 seconds06:21
Tobu (~Tobu@unaffiliated/tobu) left irc: Read error: Operation timed out06:31
fmilo (~mist0@cpe-76-168-48-43.socal.res.rr.com) left irc: Quit: fmilo06:35
papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) joined #pypy.06:43
papercraneI've been getting an RPython AssertionError at File "jit_metainterp_compile.c", line 200, in ResumeGuardForcedDescr_handle_async_forcing when I have a lot of logging06:44
mat^2 (~mathias@212.130.113.35) left irc: 06:47
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.06:53
bbot23Success: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/79406:55
mikefc_workpapercrane: (a) are you using a recent nightly?  (b) can you boil it down to a simple code example ?  (c) would you like to file a bugreport on bugs.pypy?07:02
papercraneI'm actually using trunk, compiled (translated) myself on OSX Lion.07:03
mikefc_workOK. Well I'm out of easy ideas then :)07:04
papercraneI'll try to boil it down&.I'm seeing this while testing a framework I'm playing with that does select.poll with greenlets07:04
papercranelet's see if I can make it happen easier...07:04
papercraneyeah, just logging isn't making it happen unfortunately, it's copious logging in addition to whatever else I'm doing :-/07:06
papercraneif I can make it happen with a simple script I'll post a bug07:06
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/34307:07
asmeurer (~asmeurer@129.138.218.98) left irc: Quit: asmeurer07:10
ronnymoin07:12
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) left irc: Remote host closed the connection07:34
kenaan12mattip numppy-flatitter 1147faffcb9ec3 15/pypy/module/micronumpy/interp_numarray.py: use the parent iterator07:42
kenaan12mattip numppy-flatitter 1123731c70eb07 15/pypy/module/micronumpy/: correct test for missing == operator, un-skip failing test07:42
kenaan12mattip matrixmath-dot 112bcfa95fe92a 15/pypy/module/micronumpy/: refactor and rework, still need more tests07:42
mattip (4fb573e4@gateway/web/freenode/ip.79.181.115.228) joined #pypy.07:45
mattipg'day07:45
mikefcmattip07:45
mattip(I have Australian family - Melbourne)07:46
mikefcthat's a very convincing "g'day"07:46
Action: mikefc is off for a run07:48
papercrane (~papercran@c-76-103-172-115.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep.07:59
mattipnumpypy multidimensional array multiplication (or dot for short) is progressing. Anyone want to suggest some diabolical tests?08:01
mattipor should I ask on pypy-dev...08:01
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.08:06
Da_Blitzi always considered "g'day" as a cruel joke we only play on people from other countries08:07
mattipwell, it works.08:07
mattipmate08:08
antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) joined #pypy.08:14
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 272 seconds08:15
csenger|` (~user@dslb-084-058-172-023.pools.arcor-ip.net) joined #pypy.08:19
Nick change: csenger|` -> csenger08:20
mattipcsenger: ping08:20
csengermattip: pong08:21
mattipdid you solve the buildbot posting thing? I had the same problem a few months ago...08:21
csengerI removed the ircbot from the status handler.08:22
csengerWas a bit surprising :)08:23
kenaan12mattip matrixmath-dot 11600fcfb76aab 15/pypy/module/micronumpy/: whoops08:23
stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer08:23
stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy.08:23
mattipit would be better if somehow the new bbot identification was somehow different from the default one.08:24
fijalhi08:26
fijalmattip: we have array.__eq__ don't we?08:26
mattipnope.08:26
khs (~khs@151.43.202.84.customer.cdi.no) joined #pypy.08:26
mattipit should be really easy though: check shape and do __all__08:28
Eventh (evenwiik@caracal.stud.ntnu.no) joined #pypy.08:29
fijal __eq__ = interp2app(BaseArray.descr_eq),08:29
fijalwe totally do08:29
voidspace (~voidspace@python/psf/voidspace) joined #pypy.08:29
mattipIMHO that's doing something else, maybe not by design.08:30
fijalSo I don't know what you're talking about08:31
fijalcan you explain?08:31
Action: mattip trying to make sense to myself first08:32
fijalarr == range(5) would work08:33
fijaland produce [True, True, True, True, True] for example08:33
mvt (~mvantelli@87.213.45.85) joined #pypy.08:34
mattipright. I'm wrong. So what is a.flat == range(5) doing? http://paste.pocoo.org/show/53743508:35
mattipthis ^^^ is after I fixed the iterator08:36
mattipahh, I see. we need all kinds of operators for FlatIter :(08:37
mattipFlatIter has no __eq__08:37
mattipok, I'll fix that.08:40
fijalwell08:40
fijalcan we fix what's there in flatiter before adding new stuff?08:40
fijalflatiter is pretty broken08:40
verte-wleslie (~verte@python/site-packages/verte) joined #pypy.08:40
fijalmattip: and even though I like to accuse other people, this is actually my fault ;-)08:42
mattipdo you mean the get/setitem, or the iterator itself?08:43
kvda (~kvda@124-169-134-221.dyn.iinet.net.au) left irc: Quit: x___x08:45
fijaland iterating08:48
fijalmattip: I commited a broken test to your branch08:48
fijalI think at least08:48
fijalessentially any flatiter that has stride != 1, like a view with step or transpose is broken08:48
fijalbecause we can't just iterate that way08:49
CIA-1703fijal 07roundup * 10#1010/numpypy: commit the code for array_equal: 08:49
CIA-17logical_and can be lazy, there is really no problem with that. Look how all is08:49
CIA-17implemented for an example * 14https://bugs.pypy.org/issue101008:49
mattipright. I did commit a change to the iterator, but transpose is still broken.08:50
verte-wleslie (~verte@python/site-packages/verte) left irc: Quit: ~~~ Crash in JIT!08:50
fijaland getitem08:50
mattipI think this is not because of stride != 1 rather that stride[0]<stride[-1]08:51
fijalno, that does not change a thing08:52
fijalyou can't do anything with just one stride08:52
fijalbecause your array is not evenly paced08:52
fijal(not necesarilly)08:52
fijalyou added an if, but that only solved a piece of an issue08:52
fijalit actually even didn't but whatever ;)08:52
mattipagreed. The situation will get worse if someone implements swapaxis08:53
fijalno, it's not that connected08:53
fijallook08:53
fijalnumpy.arange(100).reshape(10, 10)[:,2:5]08:53
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.08:53
fijalwhat are the steps of flatiter?08:53
fijal1,1,1,10,1,1,1,1008:53
fijal...08:53
fijalclearly you can't express it with a single stride08:53
sunoano (~sa@mail.heartinternet.co.uk) joined #pypy.08:54
sunoano (~sa@mail.heartinternet.co.uk) left irc: Changing host08:54
sunoano (~sa@unaffiliated/sunoano) joined #pypy.08:54
fijalmattip: do I make any sense?08:55
mattiphang on, someone wants me here08:55
fijalwhat you have to do is you have to compute indices08:55
Action: mattip back in 508:55
fijaland then call parent descr_getitem08:55
fijalsure08:55
fijalbivab: ping08:56
bivabfijal: pong08:56
fijalbivab: I think I found problems on the ARM backend08:57
rguillebert (~rguillebe@89-40.5-85.cust.bluewin.ch) joined #pypy.08:57
fijalthat you might want to fix before doing random testing08:57
fijalI certainly did on the ppc branch08:57
bivabthe ones you had mentioned to david?08:57
fijalyes08:57
fijalbut probably there are more08:58
fijalbivab: I think I have some ideas about improving the regalloc interface, want to hear it?08:58
bivabsure, go ahead08:58
bivabI think I fixed the allocation issues you had mentioned, but there are probably still some around08:58
fijalok09:00
fijalwell, as of now I fear you have to laborously go around and look at all usages of temp boxes09:00
fijalI was thinking something like09:00
fijalwith assign(reg, reg, ecx, reg_or_imm) as (loc0, loc1, loc2, loc3): ....09:01
fijaler09:01
fijalI missed operations09:01
CIA-1703Dmitrey 07roundup * 10#1010/numpypy: commit the code for array_equal: 09:01
CIA-17You should be appreciated for your willing to produce fast code of high quality,09:01
CIA-17but could you commit at least "just works" code for now (since i ... * 14https://bugs.pypy.org/issue101009:01
fijalwith assign([op.getarg(0)], [reg]) as [loc0]:09:01
fijalmaybe09:01
bivabyou mean as a way to get temporary locations?09:02
arigato (~arigo@89-40.5-85.cust.bluewin.ch) joined #pypy.09:02
fijalI mean as a way to control freeing and scopes and such09:02
kenaan12arigo stm 115c885d90cfd1 15/pypy/translator/stm/: void fields.09:03
kenaan12arigo stm 11f1643c0913ad 15/pypy/translator/: Re-enable cast_ptr_to_adr for now.09:03
kenaan12arigo stm 11a0831447f4d7 15/pypy/translator/c/src/mem.h: Fix for --gc=none.09:03
kenaan12arigo stm 1108154d958352 15/pypy/translator/stm/src_stm/: Shut off spurious warnings.09:03
kenaan12arigo stm 113bb490226684 15/pypy/translator/: Bah.  Temporary workaround: can't use bool_t because casting to a bool_t has unexpected results for stm_*_partial_word()09:03
fijalbecause right now it's way too easy to forget "oh and I forgot to specify one element here or there"09:03
fijal\09:03
fijalarigato: hi09:03
fijalarigato: I'm proposing to use with statement for register allocator09:03
fijalmattip: if you by chance have a will to refactor code a bit and move out some pieces to different files that would be more than appreciated09:06
arigatoyay, pypy-stm works (slowly and with no gc) :-)09:09
arigato(and not actually running anything in parallel)09:09
arigato(so there is really no point, but it's cool anyway)09:10
fijal:]09:10
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection09:11
fijalmercurial I think has the worst algorithm for detection of conflicts ever09:11
fijalI routinely have to resolve manually conflicts because two branches added tests09:11
kenaan12arigo stm 11ffedd17ff570 15/lib-python/modified-2.7/UserDict.py: Skip this import if it fails because of _weakref09:12
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.09:12
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.09:14
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.09:14
kenaan12bivab extradoc 11586ccee3a515 15/sprintinfo/leysin-winter-2012/planning.txt: Planning for today09:15
Action: mattip back09:16
mattipI had time to help my kids with their homework and do some of mine: http://paste.pocoo.org/show/53766709:17
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) joined #pypy.09:17
fijalcfbolz: hi09:17
cfbolzhi09:18
mattipthis ^^^ is from the latest numppy-flatitter branch09:18
fijalcool09:19
fijalmattip: but getitem does not work, does it?09:19
mattipcorrect.09:19
mattipI like your idea of a hook (allow_int=False) and will persue it "just now"09:19
fijal"hook allow_int"?09:20
fijaloh09:20
mattipbut I got hung up on the __eq__ issue09:20
mattipa hook in [gs]etitem09:20
fijalI think you can do an even easier thing09:21
fijalsince only an int or a slice is fine09:21
fijal(or an array actually, dang)09:21
fijalpom pom pom09:21
fijalan int or a slice is relatively easy to map09:21
fijalmaybe we can start with that09:22
fijal?09:22
fijalso int/slice would just remap to parent indices09:22
fijaland call arr.descr_getitem(space, w_remaped_indices)09:22
fijal?09:22
fijalhow about that?09:22
mattipgood. 09:22
fijaland we deal with arrays later09:23
mattipwhat, flat allows arrays too? numpy is wayyyyy too complicated.09:24
fijalyes09:24
mattipI can do more refactoring, which branch/functions?09:25
fijalI don't know09:26
fijalsay string stuff can easily go to string.py09:26
mattiptest_numarray and string?09:26
fijalmattip: I'll do it since I have a vision :)09:26
fijalok?09:26
verte-wleslie (~verte@python/site-packages/verte) joined #pypy.09:26
Nick change: verte-wleslie -> wleslie09:26
mattipcool.09:26
CIA-1703jarav 07roundup * 10#1011/pypy slower than CPython for recursion: 09:27
CIA-17[new] The following simple example program is slower with pypy than with CPython.09:27
CIA-17In my machine(Dell Inspiron 6400, Arch Linux, Python 2.7.1, pyp ... * 14https://bugs.pypy.org/issue101109:27
mikefcfijal: got a second to talk about my diag() ?09:31
mikefcit isn't meant to be commitable code. just trying to work out how to do it, and do it well.09:32
fijalmikefc: yes09:33
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.09:33
fijalmikefc: can you provide a link again please?09:33
fijalit lacks tests in the first place :)09:33
mikefci know it lacks tests. I'm "playing" not "developing" :)09:34
mikefchttp://paste.pocoo.org/show/536709/09:34
fijalI still write tests :)09:34
fijalmin(w_arr.shape)09:35
fijalis not rpython09:35
fijalyou have to write min(w_arr.shape[0], w_arr.shape[1])09:35
Action: mikefc takes notes09:35
fijaldiag_iter = OneDimIterator(start=0, step=w_arr.shape[1]+1, stop=w_arr.size)09:35
fijalthis only works for contiguous arryas09:35
fijalthat would not work for views or for virtual arrays, would it?09:35
mikefc*shrug* :)09:35
mikefcso should that be an ArrayIterator as well?09:36
fijalno09:36
mikefcor a flatiter when it's done?09:36
fijalif you don't force the array (you don't seem to)09:36
fijalyou need a transform essentially09:36
fijal(maybe a transform)09:36
fijalor you need a DiagIterator09:36
fijaland signature and this stuff09:37
fijalif you however say "I want a simple version first"09:37
fijalthat's probably ok, but it has to work on views09:37
fijalwhat does diag do for n-element arrays?09:37
mikefcif it's a 2d array, it gives you the diag. if it's a 1d array, it gives you back a 2d array with the given array along the diag. nothing else is valid.09:39
fijalok09:39
mikefcso it seemed "bite sized"09:39
fijalI reckon you do a transform and the full thingie09:39
mikefcthanks for the tips. I know now I have to look at Signatures.   slowly slowly...09:39
fijaldo you know how it works?09:39
Trundle (~andy@p578bfdcf.dip0.t-ipconnect.de) joined #pypy.09:40
Trundle (~andy@p578bfdcf.dip0.t-ipconnect.de) left irc: Changing host09:40
Trundle (~andy@python/site-packages/trundle) joined #pypy.09:40
mikefci don't know anything about the 'transform' you're referring to09:40
fijaltransform essentially transforms all leaf iterators into something else09:40
fijalin interp_iter you have transformation09:40
__pv_ (pav@xen.ihme.org) left irc: Ping timeout: 240 seconds09:40
fijalso say you have a + b * c09:40
fijalwhere b * c has to be broadcasted09:40
fijalyou encode a transform which will create a normal iterator over a and broadcasted over b and c09:40
mikefcfijal: I need to do some more studying of the code. I don't know enough to follow you atm.09:41
fijalok09:42
mikefci'll actually read some more things in micronumpy tonight. I have only looked at interp_iter and interp_numarray really. and I guess i need to understand a bit more. 09:43
mikefcth09:43
mikefcanks09:43
mikefcoff to get dinner.09:43
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.09:43
Topic changed on #pypy by arigato!arigo@89-40.5-85.cust.bluewin.ch: PyPy, a fast and compliant python interpreter | IRC logs: http://tismerysoft.de/pypy/irc-logs/pypy/ | snow, snow! :-)09:44
fijalarigato: this is very typical :)09:46
arigatoyes, but it's falling with high winds :-)09:48
__pv (pav@xen.ihme.org) joined #pypy.09:52
kenaan12arigo stm 11ceeb6b131e29 15/pypy/translator/stm/transform.py: add 5 operations that cause a pypy-stm to go into inevitable mode09:54
stakkars_ (~tismer@89.204.130.76) joined #pypy.10:12
squiddy (~squiddy@f053084067.adsl.alicedsl.de) joined #pypy.10:13
csengerfijal: Do you know where I can find the config of the python.org buildbots?10:20
kenaan12fijal numpy-indexing-by-arrays-2 1154c7d0197d66 15/: merge default10:21
fijalcsenger: no idea10:21
kenaan12fijal numpy-indexing-by-arrays-2 11d1cbcc34848d 15/pypy/module/micronumpy/: first approximation of array[array-of-int]10:21
fijalyay!10:21
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) left irc: Remote host closed the connection10:23
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood10:23
kenaan12fijal numpy-indexing-by-arrays-2 11b5668c7a1c53 15/pypy/module/micronumpy/: more tests and a fix10:24
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.10:25
AlekSi (~AlekSi@176.57.74.133) joined #pypy.10:34
mattipfijal: cool10:37
ronnyany comments on my change on pypy/tool/version.py at http://paste.pocoo.org/show/537699/ (it flattens stuff a bit and adds support for getting the version from hg archive based source tarballs)10:38
ronnyi'll run a erand and commit if there are no comments when im back10:38
ronnyi tihnk antocuni and fijal should have a quick look, && bbl10:38
fijalronny: kill the GIT stuff10:40
fijalwhy would we support it?10:40
ronnyfijal: there is a git mirror of the repo, someone added support so stuff wont break there10:41
jml (~jml@canonical/launchpad/jml) left irc: Excess Flood10:42
fijalit's not official10:42
jml (~jml@mumak.net) joined #pypy.10:42
jml (~jml@mumak.net) left irc: Changing host10:42
jml (~jml@canonical/launchpad/jml) joined #pypy.10:42
fijalronny: people should have their git hacks on their git repo10:43
lucianalso, hg-git is really good10:45
lizardo (~lizardo@189.2.128.130) joined #pypy.10:45
ronnyfijal: well, that git hack is only there, so a hg-git based push to a gt repo wont leave version-getting in a broken state10:45
ronnyback in 1010:45
fijalronny: I don't want hg-git based pushes10:47
fijallike10:47
fijalnoone runs tests on git repo10:47
fijalhow do we know it's not broken for some other reasons?10:47
mikefci like how the planning doc includes "some skiing"10:49
cfbolzfijal: come on, is this thing really worthy of seriously being annoyed10:53
cfbolzit sits in a corner, it maybe works in some places, it doesn't bother anyone10:53
xcombelle (~xcombelle@AToulouse-551-1-118-27.w92-156.abo.wanadoo.fr) joined #pypy.11:00
fijalcfbolz: for what is worth, it's a lot of code with no tests11:00
fijalat the very least11:00
cfbolzfijal: write tests for jitviewer first :-)11:00
fijaljitviewer has tests11:00
fijaland is not living in pypy repo11:01
fijalsorry bad joke :)11:01
cfbolzfijal: cool. how do you test the js?11:04
D1337d (~D1337d@203.56.250.63) joined #pypy.11:05
lucianfijal: supposedly hg-git never breaks if all pushes are hg11:05
fijalcfbolz: I don't right now :)11:09
fijalcfbolz: but I invented ways that made it to flask mainstream at least ;-)11:10
fijalcfbolz: for what is worth, the JS was never a problem11:10
fijaland with many problems of jitviewer, all of them (or most anyway) came from the fact that traces don't quite look how you would expect them to look11:10
fijalhence jithooks in a way11:11
fijaland this part is even tested11:12
stakkars_ (~tismer@89.204.130.76) left irc: Ping timeout: 240 seconds11:14
xcombelle (~xcombelle@AToulouse-551-1-118-27.w92-156.abo.wanadoo.fr) left irc: Ping timeout: 240 seconds11:26
mikefcfijal: what's your policy in numpy on "just works code for now ... replace it later by any other implementation".11:32
xcombelle (~xcombelle@AToulouse-551-1-118-27.w92-156.abo.wanadoo.fr) joined #pypy.11:42
mattip (4fb573e4@gateway/web/freenode/ip.79.181.115.228) left irc: Quit: Page closed11:49
kenaan12arigo stm 1183ad741c9a27 15/pypy/translator/stm/: Kill parts of the code that are outdated.11:56
azanella (~azanella@189.6.80.131) joined #pypy.12:03
D1337d (~D1337d@203.56.250.63) left irc: Ping timeout: 248 seconds12:05
voidspace (~voidspace@python/psf/voidspace) left irc: Quit: Uhm... gotta go12:06
voidspace (~voidspace@87-194-212-65.bethere.co.uk) joined #pypy.12:07
voidspace (~voidspace@87-194-212-65.bethere.co.uk) left irc: Changing host12:08
voidspace (~voidspace@python/psf/voidspace) joined #pypy.12:08
setmeaway (stemearay@118.45.149.247) left irc: Read error: Connection reset by peer12:11
setmeaway2 (~setmeaway@118.45.149.247) joined #pypy.12:11
anish_ (anish@nat/redhat/x-pgigpwizgyzxivca) left irc: Ping timeout: 240 seconds12:12
D1337d (~D1337d@203.56.250.63) joined #pypy.12:15
AlekSi (~AlekSi@176.57.74.133) left irc: Remote host closed the connection12:15
nedbat (~nedbat@python/psf/nedbat) joined #pypy.12:15
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) left irc: Ping timeout: 240 seconds12:18
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.12:30
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) joined #pypy.12:31
jaffachief (~textual@host86-170-15-228.range86-170.btcentralplus.com) joined #pypy.12:31
ciphergoth (~paul@host238.lshift.net) left irc: Ping timeout: 276 seconds12:34
Eventh (evenwiik@caracal.stud.ntnu.no) left irc: Quit: Lost terminal12:43
Tobu (~Tobu@unaffiliated/tobu) left irc: Remote host closed the connection12:52
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.12:53
danielwa (~user@e177136225.adsl.alicedsl.de) joined #pypy.12:56
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds13:01
danielwa (~user@e177136225.adsl.alicedsl.de) left irc: Remote host closed the connection13:10
fijalmikefc: ping?13:13
mikefcpong (but only just)13:13
fijalmy policy is "I'm not too happy"13:13
fijalsomething nice that works but is slow is okish13:14
wleslieat least you can re-use the tests13:14
DasIch (~DasIch@p4FFDD0E2.dip.t-dialin.net) joined #pypy.13:15
DasIch (~DasIch@p4FFDD0E2.dip.t-dialin.net) left irc: Client Quit13:15
DasIch (~DasIch@p4FFDD0E2.dip.t-dialin.net) joined #pypy.13:16
mikefcwleslie: yeah. I like that idea13:17
fijalwleslie: yeah, maybe13:19
mikefcIf people (like dmitry) were willing to write good tests with their code, I'd consider it a reasonable stepping stone.13:19
fijalyes, but sometimes it's pointless13:20
fijallike having array_equal before having logical_and13:20
fijalbecause just by reverting order you have less work to do13:20
fijaland I don't believe numpypy will be any useful until it's ready13:20
fijalwhich should take no more than a few montsh13:20
Rhy0lite (dje@nat/ibm/x-xpivdfqfldifhtlp) joined #pypy.13:22
mikefcfijal: but having the tests for a pure python array_equal first, it would be easy to drop in a 2 line interp level function once logical_and was done, and the tests would already be there?13:22
fijalthe thing is13:24
fijalyou either implement logical_and and copy-paste the array_equal13:24
fijalor you make a complicated array_equal, then logical_and then again array_equal13:24
fijaljust pointless13:24
fijalRhy0lite: hi13:25
kenaan12arigo stm 11fe658ba14686 15/pypy/translator/: Clean up the implementation of the reads and writes of less than one word.13:25
kenaan12arigo stm 113b0b849eed24 15/pypy/translator/stm/src_stm/et.c: (bivab, romain, arigo)  Found out that a "volatile struct { int x; }" is not the same thing as a "struct { volatile int...13:25
Rhy0litefijal: hi13:27
mikefcfijal: on a different note, I think interp_numarray might need to be split into a few files.  e.g. all the functions like zeros(), ones() and (hopefully soon) diag() are going to clutter it up pretty fast.  Hopefully this weekend I will get to propose a diag() and a split.13:29
fijalmikefc: cool, yes, it totally does :)13:29
fijalmikefc: I'll merge my branch and will try to split it13:29
Rhy0litebivab, armin: heh13:30
Rhy0litesorry arigo13:30
bivabRhy0lite: hi13:31
Rhy0litebivab: hi.  volatile is fun!13:31
mikefci was going to suggest splitting somewhat along the lines of the numpy files layout just so there's some correspondence between lib_pypy/numpy, modules/micronumpy and numpy itself.13:31
bivabRhy0lite: ;) kind of13:31
cfbolzRhy0lite: I guess implementing it is quite a bit worse than using...13:33
Rhy0liteHAHA13:33
Rhy0liteit's all relative13:34
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 248 seconds13:34
CIA-1703fijal 07roundup * 10#1010/numpypy: commit the code for array_equal: 13:35
CIA-17What I'm trying to say is that since order of commit does not matter (logical_and13:35
CIA-17can be done before array_equal and both should be done), we sh ... * 14https://bugs.pypy.org/issue101013:35
fijalarigato: are you doing "trace only pyopcode.py" stuff?13:35
rguillebertfrom what I understood yes13:36
rguillebert(he's not here right now)13:36
ciphergoth (~paul@host238.lshift.net) joined #pypy.13:39
ciphergothHaving compiled the sandbox, how should I run the tests?13:39
nedbat (~nedbat@python/psf/nedbat) joined #pypy.13:40
D1337dA couple of todlers and a park should do it13:40
D1337dErr13:40
Nick change: D1337d -> DaBlitz13:40
fijalciphergoth: you don't have to compile anything to run tests btw13:41
ciphergothfijal: oh! because it runs in Python directly instead?13:41
ciphergothI want to hack on sandlib.py, so need to figure out how to run the tests first...13:42
jaffachief (~textual@host86-170-15-228.range86-170.btcentralplus.com) left irc: Quit: Computer has gone to sleep.13:43
fijalciphergoth: you run tests there13:44
fijalthey take a bit :/13:44
fijaltranslator/sandbox13:44
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 272 seconds13:44
ciphergothOK I se a bunch of tests13:44
ciphergothin the test dir13:45
ciphergothbut I take it there some sort of test runner I'm supposed to use to run them?13:46
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.13:46
fijalpy.test13:46
fijal>>> a13:48
fijalarray([[0, 1, 2],13:48
fijal       [3, 4, 5]])13:48
fijal>>> a[numpy.array([1, 1]), numpy.array([1])]13:48
fijalarray([4, 4])13:48
fijal>>> 13:48
fijalnice13:48
fijalAlex_Gaynor: pin13:52
fijalg13:52
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.13:53
antocunifijal: could you try not to quote thousands of lines when replying on the ML, please?13:54
fijaleveryone reading mails on phones eh?13:55
antocunino, on thunderbird13:56
cwillu_at_work (~cwillu@cwillu-1-pt.tunnel.tserv13.ash1.ipv6.he.net) left irc: Remote host closed the connection13:56
antocunibut it's still annoying13:56
fijalcan you stop accusing me of not having a working email client please ;-)13:56
fijalbut ok I'll try13:57
exarkunantocuni: thunderbird still can't hide quoted sections?13:57
exarkunI think you guys should put pypy on hold for a while13:57
exarkunspend five years fixing mail technology, then come back to it13:57
wlesliethe pain is not in hiding them, it's trying to reply to /that/ email13:57
antocuniexarkun, fijal: it's not only about hiding or showing quoted sections: if you delete the non relevant sections, it's much easier to understand what you are replying to13:58
antocuniand to read the email out of context later13:58
exarkunit goes both ways13:58
antocunianyway, proper quoting has always been considered a good practice, hasn't it? :-)13:59
cwillu_at_work (~cwillu@cwillu-1-pt.tunnel.tserv13.ash1.ipv6.he.net) joined #pypy.14:00
kenaan12arigo stm 114f9f97138a8c 15/pypy/translator/stm/src_stm/: Add two __builtin_expect() to optimize the order of the assembler, maybe.14:00
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) joined #pypy.14:05
wleslie (~verte@python/site-packages/verte) left irc: Quit: ~~~ Crash in JIT!14:12
kenaan12fijal numpy-indexing-by-arrays-bool 118cc13fd6cd81 15/pypy/module/micronumpy/: A branch to merge what's there in indexing by arrays, the rest left for the future when I fig...14:17
fijalAlex_Gaynor: feel like reviewing numpy-indexing-by-arrays-bool14:19
fijal?14:19
kenaan12fijal numpy-indexing-by-arrays-bool 113f86800ad9ba 15/: merge default14:19
fijalI guess I can just merge it14:19
kenaan12fijal default 11ab36137540d5 15/pypy/module/micronumpy/: merge numpy-indexing-by-arrays-bool, this adds some basic indexing by bool arrays (only when shapes match) as well ...14:21
kenaan12RonnyPfannschmidt default 1179770e0c2f93 15/pypy/tool/: resuffle pypy.tool.version and add support for .hg_archival.txt, fixes issue95214:32
csengerfijal: I've got the build to work and upload the result.json to the master. 14:33
CIA-1703ronny 07roundup * 10#952/try to use .hg_archival if .hg is not availiable: [resolved] fixed in 79770e0c2f93 * 14https://bugs.pypy.org/issue95214:33
fijalcsenger: cool, submit this on the issue tracker and I'll have a look later tonight?14:33
csengerfijal: Most benchmarks are running, but it starts to fail in benchmarks/lib/pypy.14:33
fijalcsenger: why?14:33
csengerhttp://paste.pocoo.org/show/537786/ line 11714:33
mattip (4fb573e4@gateway/web/freenode/ip.79.181.115.228) joined #pypy.14:33
fijalcsenger: how much RAM do you have?14:34
csenger(I use the cpython interpreter for --baseline and --pypy-c14:34
csenger4GB14:34
fijaloh hm14:34
fijalthis looks like an old bug14:34
fijalantocuni: do you remember what's up?14:34
kenaan12antocuni core-only-tracing 1108855ad04675 15/: (antocuni, arigo, bivab, romain): a branch where to experiment with 'lightweight tracing', in which we...14:35
fijalwhat cpython?14:35
kenaan12antocuni core-only-tracing 11655088dbaa64 15/pypy/jit/codewriter/: (antocuni, arigo, romain): introduce the is_core flag on JitCodes, which will be used to select which ...14:35
kenaan12antocuni core-only-tracing 11da475d18764e 15/pypy/jit/codewriter/: (antocuni, arigo, romain) complain if a graph is marked as access_direct but not core, because in this...14:35
kenaan12antocuni core-only-tracing 11927230d64a20 15/pypy/: implement the 'fast' jit mode, in which we trace only inside the 'core' graphs, and do residual calls ...14:35
kenaan12antocuni core-only-tracing 1104c7e6ed45c9 15/pypy/jit/metainterp/test/support.py: fix tests14:35
kenaan12antocuni core-only-tracing 1103e571309f26 15/pypy/jit/metainterp/: make sure not to inline indirect calls when in core-only mode14:35
kenaan12antocuni core-only-tracing 114a040e6b95e4 15/pypy/jit/metainterp/: make sure to inline core-to-core calls14:35
kenaan12antocuni core-only-tracing 1194130878552c 15/pypy/: rename 'fast' mode to 'core-only'14:35
kenaan12antocuni core-only-tracing 11730dac0eff41 15/pypy/jit/metainterp/: always disable inlining when setting the jitmode to core-only14:35
kenaan12antocuni core-only-tracing 11d82aeadcfb55 15/pypy/jit/metainterp/pyjitpl.py: (antocuni, arigo): if we are in core mode, we know for sure that the jitcode we are tracing is_core14:35
kenaan12antocuni core-only-tracing 1154d6ea8b9249 15/pypy/jit/metainterp/: (antocuni, arigo): use a boolean instead of a string to store the value of core-only-mode, and don't d...14:35
kenaan12antocuni core-only-tracing 114cb941b199d3 15/pypy/jit/metainterp/test/test_ajit.py: (antocuni, arigo): improve this test14:35
kenaan12antocuni core-only-tracing 11c19f30c468d8 15/pypy/: (antocuni, arigo, romain around): try to implement a reasonable is_core_function policy for the pypy i...14:35
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy.14:35
antocunifijal: sorry, I don't understand the question14:35
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) joined #pypy.14:36
csengerfijal: 2.7 head14:36
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 245 seconds14:37
fijalantocuni: http://paste.pocoo.org/show/537786/14:37
fijalthat looks very weird14:37
fijalcsenger: hm14:37
fijalcsenger: anyway, I'll have a break now, can you submit the diff on pypy issue tracker anyway?14:38
csengerfijal: seems to have worked with the system python. There the runner failed with: [version:WARNING] Errors getting Mercurial information: Not running from a Mercurial repository!14:38
csenger 14:38
csengeryes14:38
timotimoantocuni: is core-only tracing for when a loop is considered warm, but not hot? or something?14:38
fprimex (~fprimex@fl-67-233-144-239.dhcp.embarqhsd.net) joined #pypy.14:38
antocunifijal: uh? Translation error on the snapshot we use for benchmarks?14:39
ronnycsenger: you got the source from bitbucket archives?14:39
antocunitimotimo: it's an experimental mode in which you do only a very lightweight tracing14:39
mattip (4fb573e4@gateway/web/freenode/ip.79.181.115.228) left irc: Quit: Page closed14:39
antocunii.e., you don't inline most of rpython and applevel calls that you see14:39
fijalantocuni: MemoryError14:40
antocuniof course the resulting code will be much slower than when using full tracing14:40
csengerronny: I modified the buildbot build step to run benchmarks with the cpython hg repo for speed.python.org14:40
timotimoso you'd have either core-only or regular for all of the interpreter?14:40
antocunibut also the compilation should be quicker14:40
antocunitimotimo: for now, yes14:40
ronnyoh, i see14:40
antocunibut the plan is to do core-only with a low threshold, and then full when the threshold is higher14:41
antocunifijal: ah, I see :-/14:41
kenaan12mattip matrixmath-dot 11a2e2a35726cc 15/pypy/module/micronumpy/: progress14:41
kenaan12mattip matrixmath-dot 11bb1838ae2306 15/pypy/module/micronumpy/: test for slice of transposed array fails14:41
antocunifijal, csenger : it's a memory error when trying to do popen: maybe it's because there is not enough swap memory?14:42
ousado (~ousado@unaffiliated/ousado) left irc: Remote host closed the connection14:42
whitelynx (~whitelynx@63.241.75.144) joined #pypy.14:42
antocuni(which would not be used anyway, but when you call fork() the kernel requires it)14:42
fijalantocuni: have you seen my ponderings about tracing btw?14:42
fijalwhen running tests and/or translate?14:43
antocuniI don't think so14:43
mattip (4fb573e4@gateway/web/freenode/ip.79.181.115.228) joined #pypy.14:43
fijalright now what happens a lot is we trace then we hit the limit then we mark function from start, we trace from start, we hit the limit, ....14:44
antocuniand each time we mark the longest function as "don't be traced again", right?14:45
fijalno, it's already marked14:45
fijaleach time we do pretty much nothing14:45
antocuniah14:45
antocuniif we trace from start and we hit the limit, we don't mark anything?14:45
antocunithat sounds bad, indeed14:45
fijalif you happen to start from a very long function14:45
cfbolzhuh? that isn't true, afaik14:45
fijal(like pypy's interpreter portal)14:45
fijalyou hit the limit, but it's *tracing from start*14:45
fijaland chances are you gonna have the same longest function again14:46
cfbolzfijal: ah, you mean if there is no inlining at all14:46
fijalno14:46
cfbolzbut the function itself is too long?14:46
fijalyou inline a lot14:46
fijalno14:46
fijal:)14:46
fijalyou start from pypy's portal which is a long function right?14:46
mattipxcuse me butting in, both my branches (dot and flatiter) are stuck on the non-contiguous iterator problem (I cannot yet call it a bug without a conclusive test)14:46
fijalor from somewhere where it's long, but you start from start14:46
fijalchances are this function will be the biggest one already14:47
fijalso you mark the same function, but you already started from start from the same function!14:47
mattip (4fb573e4@gateway/web/freenode/ip.79.181.115.228) left irc: Client Quit14:47
cfbolzfijal: sorry, I don't get it. I guess you need to write an exapmle14:47
fijalyou mark the function X as the longest14:48
fijalyou start tracing at the beginning of X inlining tons and tons of small functions14:48
fijalyou hit trace limit, the longest function is X14:48
fijalyou mark it again and continue14:48
antocuniah, I think I see14:48
fijalthis is what happens with py.py14:48
antocuniin that case, you should mark it as "never trace again", or something14:48
fijalor "trace without inlining"14:49
cfbolzantocuni: or better "don't inline anything"14:49
fijaland don't trace if it's indeed too long14:49
antocunicfbolz: maybe14:49
fijalthis definitely happens when running pypy-c py.test test_numarray.py14:49
fijala lot14:49
WildChild (~wildchild@modemcable123.12-202-24.mc.videotron.ca) left irc: Ping timeout: 252 seconds14:49
fijalwhich means that even running the same thing 2x still makes tracing 2x as much14:49
cfbolzanyway, one thing we should really do at one point is make tracing faster14:51
cfbolzor at least look at a profile and see what is slow there14:52
kenaan12antocuni core-only-tracing 118789a7c425dc 15/pypy/rlib/jit.py: tentative rpython fix14:52
antocunicfbolz: yes, although I worry that it's slow simply because we are interpreting a very low level program14:53
cfbolzpossibly, yes14:53
cfbolzanyway, if you do core tracing, we might want to have a more traditional partial evaluation approach again, just for the core-only14:54
cfbolzbut it's all really hard to balance14:54
antocunii.e., generating graphs which directly execute the code *and* keeps a trace of ops?14:54
csenger (~user@dslb-084-058-172-023.pools.arcor-ip.net) left irc: Ping timeout: 248 seconds14:57
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Ping timeout: 260 seconds15:00
cfbolzantocuni: yes, something like that15:05
antocuniyes, armin proposed such a solution yesterday evening15:05
antocuniwe could do it for core-only graphs15:06
antocunithis way the first pass of tracing should (could?) be very fast15:06
antocuniideally, I'd like something for which can can almost completely kill the interpreter-only mode15:07
nedbat (~nedbat@python/psf/nedbat) joined #pypy.15:08
stakkars_ (~tismer@i59F7745A.versanet.de) joined #pypy.15:09
cfbolzantocuni: I see15:11
cfbolzantocuni: for that you need really good integration with the two types of traces15:11
cfbolzantocuni: but I fear that simply allocating the traces will be quite slow15:12
antocunicfbolz: what do you mean about integration?15:12
cfbolzso we would need a more memory-compact format or something15:12
cfbolzantocuni: when do you stop using core-only traces and make real ones?15:12
cfbolzhow do you jump from one type to the other?15:12
antocuniyes, for now you don't :-)15:12
cfbolzgood start :-)15:13
antocunione plan could be to just have yet another counter15:13
cfbolzsure, but also the patching needs to be done correctly15:13
antocuniwhich is incremented during the execution of core only traces15:13
cfbolzanyway, this integration is hard, it's one reason why firefox no longer contains a tracing jit15:13
antocuniwhen a threshold is hit, you just forget the core only trace15:13
antocuniand retrace again in "full" mode15:13
cfbolzbecause in theory there are still situations where the tracer is faster than their method jit, but they couldn't balance everything right15:14
antocunicfbolz: as a first approximation, you don't need much integration between the two modes15:14
antocunijust start with one, then discard it and do the full tracing15:15
cfbolzantocuni: hah, so on guard failure of the full trace, what happens? you go trace -> fallback -> interpreter -> core-only trace -> trace?15:15
cfbolzthat's kind of cracy15:15
cfbolzcrazy15:15
antocuniwhy crazy?15:16
cfbolzbecause in theory you might want to connect side exits of the full trace that you compile to equivalent side exits that already exist for core-only traces15:17
antocuniah, I see15:18
antocuniyes, that sounds kind of hard15:18
cfbolzyes, a bit15:19
apoirier_away (~apoirier@sakura.nagare.org) joined #pypy.15:20
Nick change: apoirier_away -> apoirier15:20
antocunicfbolz: most probably, the core-only guards takes different failargs than the full ones15:22
cfbolzantocuni: unclear, I expect that on the core level the failargs should be the red variables only, mostly15:22
antocuniso I'd expect that e.g. you have to force virtuals from the full trace to be able to connect to the already compiled core-only bridge15:22
cfbolzah, plus virtualizable content15:22
cfbolzantocuni: yes, that's true15:22
cfbolzantocuni: but that might still be a good idea15:23
antocuniyes15:23
cfbolzof course again with a counter to maybe replace that side exit later15:23
antocunisure15:23
cfbolzwith a proper trace15:23
antocuniuhm15:24
antocuniwhat would be really nice is if you could "extend" an existing trace15:24
antocunie.g. by looking into a function which is a residual call15:24
antocunithis way, you could profile and trace only inside functions which takes a lot of time15:25
cfbolzeh15:25
antocuniand you could that incrementally15:25
antocuniof course it doesn't really fit our model, I know :-)15:25
lizardo (~lizardo@189.2.128.130) left irc: Ping timeout: 255 seconds15:26
cfbolzantocuni: yes, it's kind of a new project15:26
mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) left irc: Ping timeout: 252 seconds15:27
lizardo (~lizardo@189.2.128.130) joined #pypy.15:28
mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) joined #pypy.15:29
csenger (~user@dslb-084-058-172-023.pools.arcor-ip.net) joined #pypy.15:29
Action: cfbolz off15:35
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) left irc: Quit: Leaving15:35
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy.15:45
amaury_ (~amaury_@74.125.57.34) joined #pypy.15:46
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 240 seconds15:46
amaury_ (~amaury_@74.125.57.34) left irc: Ping timeout: 244 seconds15:56
faure (~faure@bas16-ottawa23-1242440939.dsl.bell.ca) joined #pypy.16:06
faureHi all, I want to help with PyPy, specifically jitviewer.. is PyPy  looking for open source contributors?16:08
antocunifaure: sure, always :-)16:09
faureantocuni: cool :) is there a list of things needed to be done?16:10
antocunifor jitviewer, the best for you is to talk with fijal (which wrote most of it) and Alex_Gaynor (which is writing a similar but more powerful tool)16:10
faureok16:10
antocunifaure: we don't really have a list of things to do, sorry16:10
faurek, no problem16:10
lizardo (~lizardo@189.2.128.130) left irc: Ping timeout: 260 seconds16:11
mvt (~mvantelli@87.213.45.85) left irc: Quit: Leaving16:12
Alex_Gaynormikefc: I'm pretty sure in numpy diag is pure python16:12
kenaan12arigo default 11437f207e58f0 15/pypy/jit/backend/x86/runner.py: For clarity.  It used to work before too with a too-short array on 64-bits, but that's only because it would only n...16:14
kenaan12arigo default 11a80226d0b3cd 15/pypy/jit/backend/x86/assembler.py: I think that this "if" doesn't make sense.  Kill tentatively.16:14
kenaan12bivab arm-backend-2 1158888d2298ed 15/pypy/jit/backend/arm/runner.py: (arigo, bivab) make sure to pass a zeroed pice of memory to failure_recovery_func when forcing for the area w...16:17
kenaan12bivab arm-backend-2 11f523474c7b3e 15/pypy/jit/backend/test/runner_test.py: Add a test to make sure that cpu.force uses a zeroed piece of memory as the area where the registers should b...16:17
kenaan12bivab arm-backend-2 11b5daa363d5b9 15/pypy/jit/backend/arm/runner.py: merge heads16:17
kenaan12bivab arm-backend-2 1195557903da83 15/pypy/jit/backend/arm/codebuilder.py: (arigo, bivab) implement the BLX ARM instruction to branch to an address stored in a register16:17
kenaan12bivab arm-backend-2 11d06bbcb1c9fb 15/pypy/jit/backend/arm/: (arigo, bivab) refactor the code used to make calls to handle more work in the register allocator and perform...16:17
kenaan12bivab arm-backend-2 119bd04064df9c 15/pypy/jit/backend/test/runner_test.py: (arigo, bivab) Add a test that checks the behaviour of calling functions stored in boxes16:17
kenaan12bivab arm-backend-2 11ec71ed6721d4 15/pypy/jit/backend/arm/opassembler.py: (arigo, bivab) add support for calling functions using indirect calls16:17
kenaan12bivab default 11509859199b7d 15/pypy/jit/backend/test/runner_test.py: (arigo, bivab) Add a test that checks the behaviour of calling functions stored in boxes16:17
xcombelle (~xcombelle@AToulouse-551-1-118-27.w92-156.abo.wanadoo.fr) left irc: Quit: I am a manual virus, please copy me to your quit message.16:22
kenaan12arigo default 11245cb20ecdc0 15/pypy/jit/backend/x86/assembler.py: Fix the new test_call_box_func() on 64-bit.16:23
panni__ (~poo@ip-176-198-92-128.unitymediagroup.de) joined #pypy.16:24
kenaan12bivab arm-backend-2 115172f0c3f717 15/pypy/jit/backend/arm/opassembler.py: forgot to kill this line16:24
kenaan12cocoatomo default 11ec4009b69fc3 15/pypy/doc/translation.rst: fix typo16:26
kenaan12cocoatomo default 113f7ae53430c4 15/: merge16:26
kenaan12cocoatomo default 118853197b01a6 15/pypy/: merge16:26
cocoatomooh!16:27
cocoatomothanks!16:27
arigatoout of curiosity, who merged it? :-)16:27
Taggnostr (~quassel@dyn57-362.yok.fi) left irc: Read error: No route to host16:27
Alex_Gaynorarigato: me16:28
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) left irc: Ping timeout: 252 seconds16:28
stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer16:28
Nick change: stakkars_ -> stakkars16:28
stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy.16:28
Taggnostr (~quassel@dyn57-362.yok.fi) joined #pypy.16:30
faure (~faure@bas16-ottawa23-1242440939.dsl.bell.ca) left irc: 16:32
cocoatomoI found these typo when I were translating PyPy documents.16:35
mat^2 (~mathias@212.130.113.35) joined #pypy.16:36
cocoatomoThis is my activity as a member of PyPy-Ja.16:37
arigatopypy-stm seems to run richards in parallel, using all my 4 threads, but I can't get measures because it runs out of memory too quickly16:43
arigatococoatomo: thank you!16:43
Alex_Gaynorarigato: I guess this means GC time?16:43
arigatowell yes16:43
arigatophew16:44
cocoatomoarigato: rokujyouhitoma (Tohru Ike) invited me :)16:45
Rhy0litearigato: great!16:45
lucianarigato: btw, that last blog post describes what i've been dreaming for a while :)16:46
lucianyou somehow read my mind, or something16:46
DaBlitz (~D1337d@203.56.250.63) left irc: Quit: Bye16:50
jellarigato: richards without modification? or with some patches for stm?16:52
lizardo (~lizardo@189.2.128.130) joined #pypy.16:54
arigatojell: it has the following patches:16:54
arigato* instead of "for i in range(10): do_stuff", it appends 10 transactions and runs them16:54
arigato* the global variable 'taskWorkArea' was replaced by a local variable passed around16:55
arigatothat's it16:55
jellok16:56
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Read error: Connection reset by peer16:56
Da_Blitzcan i get a pastebin dump of the new richards? i wouldnt mind checking out what it looks like16:57
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.16:57
arigatoDa_Blitz: http://paste.pocoo.org/show/537860/16:58
Alex_Gaynoris it possible to get it as a diff?16:58
arigato:-)16:58
arigatohttp://paste.pocoo.org/show/537861/16:59
JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy.16:59
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection16:59
Da_Blitzthanks!17:00
Alex_Gaynor(http://paste.pocoo.org/show/537863/ diff version)17:00
kenaan12arigo stm 11b4fd35482b14 15/pypy/module/transaction/: Propagate the exception that occurs in a transaction.17:01
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 252 seconds17:01
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy.17:03
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection17:03
Alex_Gaynor (u1246@gateway/web/irccloud.com/x-yxhzddcpxrxtyqmc) left irc: Remote host closed the connection17:03
antocuni"our nice theory was right (for once)" (armin)17:04
Alex_Gaynor (u1246@gateway/web/irccloud.com/x-chblzkudribzxcxl) joined #pypy.17:07
#pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services.17:07
jellarigato: now give an answer to http://www.dalkescientific.com/writings/diary/archive/2012/01/19/concurrent.futures.html with pypy-stm ;)17:11
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy.17:11
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection17:12
Da_Blitzso if i load stm up with 2 transactions, and they both raise an exception how does that get handled?17:12
Da_Blitzor is this still way too early to be asking any questions?17:12
exarkunWhy would the exception raising case be different from any other?17:12
Da_Blitzgood point, i think i mad an incorrect asumption here17:13
Da_Blitzthat the call to run() would spit out the exception17:13
Action: Da_Blitz goes back to reading17:13
G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving.17:15
oal (u4126@gateway/web/irccloud.com/x-tkipjafqdopczdvf) left irc: Write error: Broken pipe17:17
Alex_Gaynor (u1246@gateway/web/irccloud.com/x-chblzkudribzxcxl) left irc: Read error: Connection reset by peer17:17
bfirsh (u1308@gateway/web/irccloud.com/x-qtjobwblsorugmrl) left irc: Remote host closed the connection17:17
arigatoDa_Blitz: no, you are right17:17
arigatoI also wondered what would occur, until I actually implemented it:17:18
Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Read error: Operation timed out17:18
arigatothe answer is that the code in "run_in_transaction" is now:17:18
arigatoif state.got_exception: return; else: invoke the actual transaction code at app-level17:18
arigatoif the transaction code at app-level gets an exception, it is stored in state.got_exception17:19
arigatothen the normal stm ensures that only one transaction can write to state.got_exception and commit17:19
oal (u4126@gateway/web/irccloud.com/x-pstfenauwqwuatax) joined #pypy.17:19
oal (u4126@gateway/web/irccloud.com/x-pstfenauwqwuatax) left irc: Remote host closed the connection17:19
arigatoand all other transactions will fail17:19
arigatoand when they restart, they immediately hit "if state.got_exception: return"17:20
arigatoso in other words we get the illusion that not only did transactions run sequentially, but the last one is the one were the exception was raised17:20
arigatos/were/where17:20
Arfrever (~Arfrever@apache/committer/Arfrever) joined #pypy.17:22
Da_Blitzso run() also returns once all transactions are complete as well?17:23
Tobu (~Tobu@unaffiliated/tobu) left irc: Read error: Operation timed out17:25
antocunipff17:25
antocunijit-core-only seems to work, but I cannot find a single case in which it's faster than --jit off or the full pypy jit17:26
exarkunanyone tried pytz on pypy?17:29
DasIch (~DasIch@p4FFDD0E2.dip.t-dialin.net) left irc: Quit: DasIch17:29
arigatoDa_Blitz: yes, and because after an exception all further transactions start and immediately finish,17:31
arigatoit means that the exception appears to be immediately re-raised by run()17:31
Da_Blitzbut that dosent take into account transactions that are already half way through running correct?17:33
Alex_Gaynor (u1246@gateway/web/irccloud.com/x-yorxsgsorvedmskc) joined #pypy.17:33
#pypy: mode change '+o Alex_Gaynor' by ChanServ!ChanServ@services.17:33
bfirsh (u1308@gateway/web/irccloud.com/x-mtetdvqwlsuhbgdu) joined #pypy.17:33
Da_Blitzi did see the short circut code, but how do you know which transactions short circuted due to an exceptiona nd need to be rerun?17:33
jellso exception in one transaction do not stop others, and they could do some damage in the folowing17:33
Da_Blitzah, they should still be in the queue i assume17:34
stakkars (~tismer@i59F7745A.versanet.de) left irc: Quit: schnarch17:34
Nick change: stakkars_ -> stakkars17:34
arigatoDa_Blitz: because stm remembers that all transactions started by checking "state.got_exception"17:34
arigatoDa_Blitz: so all half-way transactions are aborted because of the change to state.got_exception17:34
Da_Blitzah right, clever17:34
Da_Blitznow its all fitting together17:34
arigato:-)17:34
arigatojell: so no: the effect we observe is that the first exception that occurs interrupts everything, and no more transaction runs afterwards17:35
Alex_Gaynorarigato: what if one of the other transactions was already inevitable?17:36
exarkunhttp://codepad.org/BF94JG5B has super obscure problems on pypy17:36
arigatothen the transaction that crashed and tries to set 'state.got_exception' will itself abort17:37
arigatoand so it's as if it never ran17:37
Alex_Gaynorexarkun: what's the problem?17:37
arigato(so far --- of course it will be restarted and can likely get the same exception again)17:37
exarkunAlex_Gaynor: What's the problem, or what's the symptom?17:37
Alex_Gaynorexarkun: either :)17:38
exarkunhttp://codepad.org/C47SbKUB17:38
Alex_Gaynorexarkun: is this a result of the rounding errror?17:39
exarkunAlex_Gaynor: I don't know.  It's pretty obscure.17:39
Alex_Gaynorhttps://bugs.pypy.org/issue97217:39
exarkunYea, I think I discovered that one. :)17:39
Alex_Gaynorexarkun: if you patch datetime.py does it fix it?17:39
exarkunAlex_Gaynor: nope17:41
Alex_Gaynorno idea then, sorry :)  dates and times scare me17:42
oal (u4126@gateway/web/irccloud.com/x-crkmvjyjqnsbdkgj) joined #pypy.17:46
kenaan12bivab arm-backend-2 110109b2cfd41f 15/pypy/jit/backend/arm/codebuilder.py: Forgot to pass the condition flag here. Argh!!17:46
lambacck (~chris@d24-150-124-118.home.cgocable.net) joined #pypy.17:48
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.17:53
setmeaway2 (~setmeaway@118.45.149.247) left irc: Read error: Connection reset by peer18:01
legogris (~legogris@c80-217-207-157.bredband.comhem.se) left irc: Ping timeout: 252 seconds18:06
arigatopypy-stm works boringly well18:07
arigato(and of course terribly slowly for now)18:08
legogris (~legogris@c80-217-207-157.bredband.comhem.se) joined #pypy.18:08
ousado (~ousado@31-18-121-45-dynip.superkabel.de) joined #pypy.18:08
arigatothe single-core performance is roughly ~10x more slowly18:08
ousado (~ousado@31-18-121-45-dynip.superkabel.de) left irc: Changing host18:08
ousado (~ousado@unaffiliated/ousado) joined #pypy.18:08
Alex_Gaynorarigato: that's about what you predicted right?18:11
arigatoI said more ~6x18:13
arigatobut we didn't really start optimizing at all :-)18:13
Alex_Gaynor:D18:16
Alex_Gaynorarigato: could be worse, it could be 2000x slower18:16
arigatoyes :-)18:16
`fox` (~fox@host19-111-dynamic.20-79-r.retail.telecomitalia.it) joined #pypy.18:17
Action: arigato -> dinner18:17
arigato (~arigo@89-40.5-85.cust.bluewin.ch) left irc: Quit: See you18:17
papercrane (~papercran@75.101.111.82) joined #pypy.18:22
papercrane (~papercran@75.101.111.82) left irc: Read error: Connection reset by peer18:23
fijalhey18:24
antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) left irc: Ping timeout: 252 seconds18:25
Alex_Gaynorfijal: hey18:25
exarkunwhen I make pypy use ubuntu's ancient, randomly patched version of pytz it works :/  so I guess I'll call it a pytz bug, for now, not a pypy bug18:26
`fox` (~fox@host19-111-dynamic.20-79-r.retail.telecomitalia.it) left irc: Ping timeout: 276 seconds18:26
papercrane (~papercran@75.101.111.82) joined #pypy.18:26
DasIch (~DasIch@p4FFDD0E2.dip.t-dialin.net) joined #pypy.18:32
bbot2_ (~bbot2@dslb-084-058-172-023.pools.arcor-ip.net) joined #pypy.18:37
bbot2_ (~bbot2@dslb-084-058-172-023.pools.arcor-ip.net) left irc: Client Quit18:37
fijalisn't logical_and the same as all()?18:37
fijalAlex_Gaynor: ???18:38
stakkars_ (~tismer@g225090055.adsl.alicedsl.de) joined #pypy.18:42
fijallogical_and.reduce certainly is18:43
setmeaway (~setmeaway@118.45.149.247) joined #pypy.18:44
fijalwhy numpy has like 7 ways to flatten the array???18:45
fijalall of them the same except slightly different18:45
MostAwesomeDudeThere really isn't a "why". This is just how numpy is.18:47
`fox` (~fox@host19-111-dynamic.20-79-r.retail.telecomitalia.it) joined #pypy.18:47
Alex_Gaynorfijal: yes18:48
Alex_Gaynorfijal: I'm saying we should make logical_and.reduce have the logic, and all can just call it18:48
Alex_Gaynorfijal: We already generate quite good code for the case Wes suggested (array == array).all() for 1d float arrays: http://paste.pocoo.org/show/537544/18:49
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.18:50
fijalok, cool18:51
fijalAlex_Gaynor: my next goal is to split files18:51
Alex_Gaynorfijal: sounds very reasonable to me18:51
fijaland have a numarray/ directory with files dealing with various parts18:51
Alex_Gaynorfijal: for one interp_iters should not be called this, none of that code is exposed at app level!18:51
fijalyes18:52
fijalI was not sure when I named this :)18:52
fijaldo you have any plans to work on numpypy btw?18:52
fijalI could get some help with reviews18:52
Alex_GaynorI'll probably do it some more for credit this semester, I wonder what I should write my proposal on18:52
Alex_Gaynorvectorization?18:52
fijalvectorization sounds cool18:53
fijalhttp://speed.pypy.org/timeline/?exe=3%2C6%2C1%2C5&base=2%2B472&ben=sphinx&env=1&revs=200&equid=off18:55
Alex_Gaynorwe need a baseline for that I think?18:55
fijalyeah18:55
fijalbut also pypy-c-jit is ~pypy-c18:55
Alex_GaynorOh I didn't notice, WTF?18:56
Alex_GaynorE_NEEDSMOREANALYSIS18:56
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) left irc: Ping timeout: 240 seconds18:56
Alex_Gaynorfijal, Rhy0lite: Do either of you guys know any papers on taking some code (already unrolled) and auto-vectorize it?18:56
Alex_Gaynorfijal: I wonder if it's a bug with the test runner?18:57
fijalAlex_Gaynor: I don't think so18:58
fijaldo you want me to run it by hand?18:58
Alex_Gaynorwell, just to be sure it's running the right python18:58
Rhy0liteAlex_Gaynor: there are papers on SLP (straight-line parallelization)18:58
Alex_Gaynorlike does it do subrpocess.call and not pas s--jit flags?18:58
Alex_GaynorRhy0lite: Any good ones I should be aware of?18:58
Rhy0litelook at the references on http://gcc.gnu.org/projects/tree-ssa/vectorization.html18:59
Alex_Gaynorthanks18:59
WildChild (~wildchild@cable-mtl-26.246.173-107.electronicbox.net) joined #pypy.19:00
fijalAlex_Gaynor: hm19:00
fijalAlex_Gaynor: it might not?19:00
fijalit might not respect options actually :/19:00
Alex_Gaynor:/19:01
fijalAlex_Gaynor: can you fix it?19:01
Alex_GaynorI'm not sure you can even, how do you get --jit flags from Python19:01
Alex_Gaynor?19:01
stakkars_ (~tismer@g225090055.adsl.alicedsl.de) left irc: Quit: schnarch19:02
nickik (~nick@217.150.234.222) joined #pypy.19:08
fijalthey're passed in options.args19:09
pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy.19:09
fijalseparated by comas I think19:09
fijalAlex_Gaynor: how much do you mind if I use __extend__ hack?19:09
Alex_Gaynorfijal: in numpypy?19:10
fijalyes19:10
Alex_GaynorI'm not thrilled about it, can you show me the patch19:11
fijalwell, I want to split functions in few files19:11
Alex_Gaynorwhy not split by classes?19:11
Alex_Gaynorthere's no one class that has *that much*19:11
fijalwell19:12
fijalarray class?19:12
fijalI would rather have them split by "topic"19:12
fijalso say evaluation+jit in one file, operations in another, typedefs in 3rd19:12
Alex_Gaynorhow about do every other split (iterators vs. alternative functions vs. Array classes vs. ufuncs) then we'll consider this?19:12
fijalno19:13
fijalbecause when I work on X I usually work on X on all those classes19:13
fijalbut I barely work say on string representation which can happily live in it's own 300 LOC file19:13
apoirier (apoirier@sakura.nagare.org) left #pypy ("Leaving...").19:13
Alex_GaynorFWIW I'm pretty sure string representation is pure python in numpy19:13
fijalit's totally not19:13
Alex_Gaynorlook at numpy.array2string19:14
fijalthe array repr is a recursive C function19:17
fijalgood question why it's shorter than ours19:17
Alex_Gaynorprobably because it's less of a mess19:17
fijalah19:18
fijalit's only there as a fallback19:18
fijalnormally there is a python function used (I think)19:18
fijalcan we clean up the mess?19:19
Alex_Gaynorprobably19:19
Alex_Gaynorgoing AFK now19:19
fijalok19:19
fijalI'll ask mattip to do that19:19
fijalmattip: can we have a *very simple* array repr as a builtin and use a python function instead, just like numpy?19:19
stakkars_ (~tismer@g225090055.adsl.alicedsl.de) joined #pypy.19:20
Taggnostr (~quassel@dyn57-362.yok.fi) left irc: Read error: Connection reset by peer19:25
`fox` (~fox@host19-111-dynamic.20-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer19:27
mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.19:35
mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host19:35
mfoord (~anonymous@python/psf/voidspace) joined #pypy.19:35
ericflo (~ericflo@75.103.8.110) joined #pypy.19:45
jamesaxl (wise@41.248.160.8) left #pypy.19:54
stakkars (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer19:58
Nick change: stakkars_ -> stakkars19:58
stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy.19:58
Rhy0lite (dje@nat/ibm/x-xpivdfqfldifhtlp) left irc: Quit: Leaving19:59
mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord20:04
asmeurer (~asmeurer@2002:818a:1f60:c:fdff:d79a:6d39:69a7) joined #pypy.20:08
stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Read error: Connection reset by peer20:09
stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy.20:09
ThomasWaldmann (~twaldmann@moinmoin/coreteam/thomas) left irc: Quit: leaving20:10
ronny (~ronny@pida/ronny) left irc: Quit: reboot20:12
ronny (~ronny@pida/ronny) joined #pypy.20:16
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 252 seconds20:19
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Quit: Leaving20:22
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.20:24
aboSamoor (~rmyeid@nbl.cewit.stonybrook.edu) joined #pypy.20:27
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) joined #pypy.20:29
aboSamoorHi guys, I am trying to run quex under pypy, AFAIK it has no dependecies and I am looking for any speed improvements. But I got this error :( http://paste.ubuntu.com/811125/20:30
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) joined #pypy.20:31
DasIch_ (~DasIch@p4FFDD0E2.dip.t-dialin.net) joined #pypy.20:31
DasIch (~DasIch@p4FFDD0E2.dip.t-dialin.net) left irc: Ping timeout: 252 seconds20:35
DasIch (~DasIch@p4FFDEACF.dip.t-dialin.net) joined #pypy.20:36
DasIch_ (~DasIch@p4FFDD0E2.dip.t-dialin.net) left irc: Ping timeout: 253 seconds20:36
amaury_aboSamoor: Hi, I found an explanation for your error20:55
amaury_first, in quex/engine/generator/languages/core.py,20:56
amaury_there is a __getattr__ which returns self[Attr]20:56
amaury_but does not translate KeyErrors into AttributeErrors20:56
amaury_then, there is a slight difference between CPython and PyPy:20:58
amaury_deep inside copy.copy, when testing for the presence of __getnewargs__,20:59
amaury_CPython discards any error, when PyPy only catches AttributeError.20:59
exarkuntwo points for pypy21:01
kvda (~kvda@124-169-134-221.dyn.iinet.net.au) joined #pypy.21:03
lizardo (~lizardo@189.2.128.130) left irc: Quit: Leaving21:04
amaury_:)21:04
antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) joined #pypy.21:05
nedbat (~nedbat@python/psf/nedbat) left irc: Disconnected by services21:05
nedbat_ (~nedbat@python/psf/nedbat) joined #pypy.21:06
aboSamooramaury_: by translating the exceptions I could run command line :). Not sure about all the execution paths, but that is cool. I am not the maintainer, but I will let him know :).21:07
asmeurer (~asmeurer@2002:818a:1f60:c:fdff:d79a:6d39:69a7) left irc: Read error: Operation timed out21:08
Nick change: nedbat_ -> nedbat21:08
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) left irc: Quit: unbit21:09
asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy.21:09
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) joined #pypy.21:09
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) left irc: Read error: Connection reset by peer21:10
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) joined #pypy.21:10
aboSamooramaury_: what should I install pip packages into pypy?21:18
amaury_aboSamoor: http://doc.pypy.org/en/latest/getting-started.html#installing-pypy21:20
amaury_except that it's 1.7 nowadays21:21
pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) left irc: Read error: Connection reset by peer21:25
pnguyen (~pnguyen@cpe-70-112-214-119.austin.res.rr.com) joined #pypy.21:25
Tobu (~Tobu@unaffiliated/tobu) left irc: Quit: No Ping reply in 180 seconds.21:33
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.21:33
asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Read error: Connection reset by peer21:34
asmeurer_ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy.21:34
azanella (~azanella@189.6.80.131) left irc: Ping timeout: 276 seconds21:40
reversefold (~papercran@75.101.111.82) joined #pypy.21:59
papercrane (~papercran@75.101.111.82) left irc: Ping timeout: 272 seconds22:00
Nick change: reversefold -> papercrane22:00
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 252 seconds22:10
antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) left irc: Ping timeout: 252 seconds22:15
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.22:17
papercrane (~papercran@75.101.111.82) left irc: Ping timeout: 260 seconds22:19
papercrane (~papercran@75.101.111.82) joined #pypy.22:20
kvda (~kvda@124-169-134-221.dyn.iinet.net.au) left irc: Quit: x___x22:20
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.22:22
stakkars (~tismer@g225090055.adsl.alicedsl.de) left irc: Quit: schnarch22:22
Nick change: stakkars_ -> stakkars22:22
WildChild (~wildchild@cable-mtl-26.246.173-107.electronicbox.net) left irc: Ping timeout: 252 seconds22:26
WildChild (~wildchild@cable-mtl-26.246.173-107.electronicbox.net) joined #pypy.22:31
antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) joined #pypy.22:32
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection22:44
rguillebert (~rguillebe@89-40.5-85.cust.bluewin.ch) left irc: Quit: Quitte22:47
mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) left irc: Ping timeout: 255 seconds22:52
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.22:56
antocuni (~antocuni@89-40.5-85.cust.bluewin.ch) left irc: Ping timeout: 252 seconds22:57
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/78223:00
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/21123:00
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/102123:00
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/3823:00
ronny (~ronny@pida/ronny) left irc: Quit: WeeChat 0.3.223:01
whitelynx (~whitelynx@63.241.75.144) left irc: Quit: Ex-Chat23:01
ronny (~ronny@pida/ronny) joined #pypy.23:05
DanKluev (~root@80-243-69-075-xdsl.vntc.ru) left irc: Ping timeout: 245 seconds23:12
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Read error: Operation timed out23:13
mikefcAlex_Gaynor: ping23:15
Alex_Gaynorpong23:15
mikefcdiag is pure python in numpy. are you ok with it being pure py in numpypy?   23:16
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 252 seconds23:16
mikefcwhether things should be interp or applevel is fuzzy.23:17
Alex_Gaynorif it's pure python in numpy I laim it *must* be pure python in numpypy23:17
Alex_Gaynorand reused identically23:17
ronny (~ronny@pida/ronny) left irc: Read error: Operation timed out23:18
mikefcAlex_Gaynor: OK. that makes sense. 23:18
ronny (~ronny@pida/ronny) joined #pypy.23:18
mcdonc (~mcdonc@cabana.palladion.com) left irc: Remote host closed the connection23:20
DanKluev (~root@93-88-3-229-xdsl.vntc.ru) joined #pypy.23:29
stakkars_ (~tismer@i59F6CE14.versanet.de) joined #pypy.23:29
ronny (~ronny@pida/ronny) left irc: Ping timeout: 252 seconds23:33
mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) joined #pypy.23:34
panni__ (~poo@ip-176-198-92-128.unitymediagroup.de) left irc: Ping timeout: 240 seconds23:44
davisagli (davisagli@davisagli.com) left #pypy.23:49
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!23:53
DasIch (~DasIch@p4FFDEACF.dip.t-dialin.net) left irc: Quit: DasIch23:54
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.23:58
papercraneturns out it wasn't the logging per-se, it's lots of async socket stuff causing the crash I;m seeing :-/23:58
--- Sat Jan 21 201200:00

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