#pypy IRC log for Saturday, 2011-12-17

azanella (~azanella@189.6.80.207) left irc: Ping timeout: 252 seconds00:11
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) joined #pypy.00:13
squiddy (~squiddy@f053084242.adsl.alicedsl.de) left irc: Remote host closed the connection00:15
nedbat (~nedbat@python/psf/nedbat) joined #pypy.00:21
aurojit (~apanda@59.178.214.32) left irc: Ping timeout: 248 seconds00:26
apanda (~apanda@59.178.223.228) joined #pypy.00:29
aurojit (~apanda@59.178.195.143) joined #pypy.00:35
apanda (~apanda@59.178.223.228) left irc: Ping timeout: 252 seconds00:37
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!00:44
rekamso (~textual@63-228-127-70.dia.static.qwest.net) left irc: Quit: weekend00:49
tilgovi (~randall@75-149-58-169-SFBA.hfc.comcastbusiness.net) joined #pypy.00:58
tilgovi (~randall@75-149-58-169-SFBA.hfc.comcastbusiness.net) left irc: Changing host00:58
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.00:58
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/17301:05
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy.01:13
mat^2 (~mathias@212.130.113.35) joined #pypy.01:13
mtigas (~mtigas@users.spokesman.com) left irc: Quit: mtigas01:14
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 252 seconds01:16
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds01:19
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/74701:19
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.01:31
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Ping timeout: 268 seconds01:33
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/189101:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/63601:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/29301:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/64601:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/30801:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/116301:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/83001:34
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/147301:34
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/75301:34
bbot213Exception: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/29301:34
ericflo (~ericflo@75.103.8.110) left irc: Quit: ericflo01:45
Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Quit: Ex+re01:50
Arfrever (~Arfrever@apache/committer/Arfrever) joined #pypy.01:55
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) joined #pypy.01:59
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) left irc: Changing host01:59
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.01:59
stakkars_ (~tismer@p5DDB72FD.dip.t-dialin.net) joined #pypy.02:00
mwhudson (~mwh@linaro/mwhudson) left irc: Ping timeout: 245 seconds02:14
apanda (~apanda@59.178.202.175) joined #pypy.02:46
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/98002:47
Unhelpful (~quassel@rockbox/developer/Unhelpful) left irc: Ping timeout: 240 seconds02:48
aurojit (~apanda@59.178.195.143) left irc: Ping timeout: 268 seconds02:48
Unhelpful (~quassel@rockbox/developer/Unhelpful) joined #pypy.02:51
aurojit (~apanda@59.178.208.3) joined #pypy.02:53
dracman (~draco@212.255.47.152) left irc: Ping timeout: 252 seconds02:53
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy02:54
apanda (~apanda@59.178.202.175) left irc: Ping timeout: 276 seconds02:54
dracman (~draco@212.255.33.143) joined #pypy.02:59
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds03:03
aurojit (~apanda@59.178.208.3) left irc: Ping timeout: 252 seconds03:09
apanda (~apanda@59.178.192.79) joined #pypy.03:13
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection03:16
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) joined #pypy.03:31
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Read error: Connection reset by peer03:31
apanda (~apanda@59.178.192.79) left irc: Ping timeout: 276 seconds03:36
apanda (~apanda@59.178.217.76) joined #pypy.03:36
apanda (~apanda@59.178.217.76) left irc: Ping timeout: 252 seconds03:47
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/189103:47
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.03:49
apanda (~apanda@59.178.200.117) joined #pypy.03:53
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Ping timeout: 240 seconds04:00
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) left irc: Ping timeout: 276 seconds04:14
jperras (~jperras@li245-112.members.linode.com) left irc: Ping timeout: 240 seconds04:35
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) joined #pypy.04:36
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/83004:59
jperras (~jperras@li245-112.members.linode.com) joined #pypy.05:03
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Read error: Operation timed out05:07
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/64605:08
jperras (~jperras@li245-112.members.linode.com) left irc: Ping timeout: 252 seconds05:09
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/147305:13
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/116305:16
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.05:28
Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Ping timeout: 268 seconds05:32
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/63605:36
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood05:38
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.05:38
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy06:05
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.06:13
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) joined #pypy.06:20
jperras (~jperras@li245-112.members.linode.com) joined #pypy.06:21
tav_ (~tav@host-92-20-7-192.as13285.net) joined #pypy.06:26
jperras (~jperras@li245-112.members.linode.com) left irc: Ping timeout: 240 seconds06:33
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/75306:33
apanda (~apanda@59.178.200.117) left irc: Ping timeout: 240 seconds06:33
apanda (~apanda@59.178.220.6) joined #pypy.06:40
ixokai (~ixokai@pdpc/supporter/student/ixokai) joined #pypy.07:09
ixokai (~ixokai@pdpc/supporter/student/ixokai) left irc: Quit: Bye!07:17
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/30807:39
squiddy (~squiddy@g224197016.adsl.alicedsl.de) joined #pypy.07:39
fijal (~fijal@197.110.12.5) joined #pypy.07:43
fijalAlex_Gaynor: heh, getinteriorfield_raw!07:57
Alex_Gaynorfijal: shouldn't we at least discuss with hakanardo or armin or someone?  we currently do no caching on any _raw operations08:00
fijalI know08:00
fijalwhat's the problem anyway?08:00
fijalwe only need get_08:01
Alex_Gaynorit seems safe, any set*_raw should invalidate, and any call08:01
Alex_Gaynoranyway, it's a bit independent of your branch08:01
fijaltrue08:03
fijalbut you can't measure gains without doing it :)08:03
fijalso it's not entirely independent08:03
Alex_Gaynorwell, you can measure gains of registers and fewer int_add08:03
fijalfewer int_add give you next to nothing08:04
Alex_Gaynorwell, it's a few percent you said?08:04
fijalyes08:04
Alex_Gaynorwell, that's something08:04
fijaldepends if java is raging in the background :)08:04
Alex_Gaynor:)08:05
fijalit's 0.610s vs 0.640s on my benchmark08:06
fijalwhich is ~5%08:06
Alex_Gaynor5%08:06
Alex_Gaynornot terrible08:06
Alex_Gaynoranyway, finish the branch so I can review it08:06
fijalI want get*_raw optimization first!08:06
Alex_Gaynorthey're totally independent08:06
fijalyes, true08:07
Alex_Gaynorget the branch merge ready, I'll make time for interiorfield cachine08:07
fijala bit boring to have an optimization that does not work ;-)08:07
Alex_Gaynorwell it does work08:07
fijalhow was the sopa hearing?08:07
Alex_Gaynordelayed until the 21st08:08
Alex_Gaynorthey're a mess, but that's life08:09
fijalit's funny08:09
Alex_Gaynori'm sure it's funny to doctors when they discuss healthcare08:10
fijalI claim healthcare debates are different08:11
fijalbecause they're primarily about financing not about healthcare methods08:11
Alex_GaynorYes, they never actually talk about healthcare ;)08:11
fijalyes08:11
fijalwhich is a good thing08:11
fijalthey should not talk or regulate internet either08:11
Alex_Gaynorprobably not, but programming is cool, because the design is the implementation08:13
Alex_Gaynorit's not like engineering where you have blueprints and construction workers08:13
Alex_Gaynoreverything is code :)08:13
fijalIBM would not agree08:13
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.08:14
fijal (~fijal@197.110.12.5) left irc: Ping timeout: 252 seconds08:24
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy.08:24
dgl (~dgl@109.86.165.231) left irc: Ping timeout: 244 seconds08:26
pedronis_ (~pedronis@73-53.195-178.cust.bluewin.ch) left irc: Remote host closed the connection08:27
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds08:54
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.08:55
jperras (~jperras@li245-112.members.linode.com) joined #pypy.09:01
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 252 seconds09:06
jperras (~jperras@li245-112.members.linode.com) left irc: Ping timeout: 268 seconds09:07
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) left irc: Quit: asmeurer__09:14
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy09:21
jperras (~jperras@li245-112.members.linode.com) joined #pypy.09:26
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.09:30
`fox` (~fox@95.233.109.194) joined #pypy.09:31
jperras (~jperras@li245-112.members.linode.com) left irc: Ping timeout: 252 seconds09:31
norox (~norox@2a00:801::2d0:b7ff:fea0:c753) left irc: Ping timeout: 268 seconds09:41
hakanardoAlex_Gaynor, fijal: caching getfield_raw is bad as it is used to check if a signal has arrived at the end of every loop09:45
Alex_Gaynorhakanardo: ah, indeed, that's a pretty good reason09:47
Alex_Gaynormaybe we need some idea of voliate vs non-voliate reads09:48
Alex_GaynorI guess right now _gc is non-volatile09:49
Alex_Gaynor_gc also has a writer barrier09:49
hakanardoyou'll need armin for that discution :)09:51
kkris (~kris@93-82-38-56.adsl.highway.telekom.at) joined #pypy.10:06
`fox` (~fox@95.233.109.194) left irc: Ping timeout: 252 seconds10:06
fijal (~fijal@197.168.229.4) joined #pypy.10:10
jperras (~jperras@li245-112.members.linode.com) joined #pypy.10:18
jperras (~jperras@li245-112.members.linode.com) left irc: Ping timeout: 255 seconds10:30
Alex_Gaynorfijal: did you see hakanardo's explanation of why caching on _raw isn't sfe ATM?10:34
fijalno?10:35
fijalshould I see logs?10:35
fijalwe're only planning to cache getinteriorfield_raw10:35
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy.10:35
Alex_Gaynorfijal: signal handling.10:36
Alex_Gaynor_raw are the only volatile reads we have ATM10:36
fijalbut not the interior10:36
Alex_Gaynorit's the general principle though10:36
Alex_Gaynorwe need a proper way to distinguish volatile vs. non-volatile reads10:36
fijalyeah well10:37
fijalit's relatively easy10:37
Alex_Gaynor?10:37
fijalwe can just emit getfield_raw_volatile10:37
fijalor so10:37
Alex_GaynorIMO w ought to try to combine _gc and _raw and put the info on teh descr10:38
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.10:42
fijalcool, I have NaN items in my shooping cart10:42
Alex_Gaynor:D10:43
arigato (~arigo@fwstups.cs.uni-duesseldorf.de) joined #pypy.10:44
jperras (~jperras@li245-112.members.linode.com) joined #pypy.10:44
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.10:45
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) joined #pypy.10:48
rinu (~rinu@117.213.57.242) joined #pypy.10:49
rinu (~rinu@117.213.57.242) left irc: Ping timeout: 248 seconds10:54
fijalarigato: oh hi armin :)10:55
arigatohi!10:55
fijalarigato: we were just discussing get_*_raw caching10:55
arigatofor numpy I guess?10:55
fijaland the problem is that we have to deal with volatile getfield_raw about signals10:55
fijalyes10:55
arigato"yay"10:56
fijalwell10:56
fijalwe can cheat and cache only getinteriorfield_raw for now10:56
fijalbut it does sound a bit excessive10:56
fijala bit unclean maybe is the right word10:56
arigatothe C-like solution is to really mark some reads as volatile10:56
fijalyes10:56
fijalwhat do you think about just that?10:57
arigatomight be needed anyway10:57
arigatosooner or later10:57
fijalok11:00
fijalmaybe indeed after refactor-signature merge11:01
fijalarigato: changing the topic a bit, did you think what to do with functions that are traced over and over11:01
fijaland are always overflowing trace_limit11:01
setmeaway (~setmeaway@119.201.52.190) left irc: Read error: Connection reset by peer11:01
setmeaway (~setmeaway@119.201.52.190) joined #pypy.11:02
fijalthis happens when we run apptests on top of pypy11:03
fijal(not the -A one, the full stack)11:03
fijal (~fijal@197.168.229.4) left irc: Read error: Connection reset by peer11:05
fijal_ (~fijal@197.175.70.192) joined #pypy.11:05
fijal_back11:06
Nick change: fijal_ -> fijal11:06
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Quit: Quitte11:07
fijal (~fijal@197.175.70.192) left irc: Read error: Connection reset by peer11:20
fijal (~fijal@41.53.52.7) joined #pypy.11:25
fijalAlex_Gaynor: man, flat iterator is annoying11:35
nedbat (~nedbat@python/psf/nedbat) joined #pypy.11:45
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Ping timeout: 240 seconds11:46
kenaan12hakanardo jit-multilabel 112d900cd7e4e3 15/: hg merge default11:47
kenaan12hakanardo jit-multilabel 11de71b40cee0f 15/pypy/jit/metainterp/: kill ExportedState.jump_args11:47
rinu (~rinu@117.196.130.167) joined #pypy.11:48
fijalAlex_Gaynor: ping?11:48
rinuhow can one run pypy on embedded systems ?11:49
fijalrinu: dunno, what sort of embedded system?11:49
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) joined #pypy.11:50
rinulike in mobile devices11:51
rinumaybe something with the ARM processors11:51
fijalit's pretty hard to run pypy on maybe something11:55
fijalyou have to be pretty specific11:55
fijalthere is an ARM JIT backend, but it's not yet entirely finished11:55
fijalit vastly depends on the OS as well, but you would likely need to do some work11:56
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Ping timeout: 244 seconds11:56
rinuis there a implementation of lua on progress using RPython ?11:58
fijalnot that I'm aware of11:59
fijalwould be pretty cool, but also relatively hard to beat luajit11:59
rinuis luajit faster tan pypy ?12:00
rinuthan*12:00
fijalthat's a very hard thing to compare, since you don't have a good set of common benchmarks12:00
fijalI would believe luajit is faster on numerical workloads12:01
fijalwhich is the only thing we can reasonably share12:01
fijalI don't know about say garbage collection12:01
kenaan12hakanardo jit-multilabel 119756d0eb5d62 15/pypy/jit/metainterp/optimizeopt/unroll.py: replace ExportedState.optimizer with ExportedState.exported_values, which conatians the info needed12:01
rinuisnt lua's syntac more readable and similar to that of python's ?12:03
rinusyntax*12:03
fijalsyntax is really something on the surface only12:03
fijalpython is a much bigger language that lua12:04
kenaan12amauryfa default 111ed3eaa72e9d 15/pypy/interpreter/astcompiler/ast.py: Run ./interpreter/astcompiler/tools/asdl_py.py12:04
fijaland syntax is really irrelevant12:04
kenaan12amauryfa default 11a930e5061f83 15/pypy/interpreter/astcompiler/: Small improvement in generated ast: lineno and col_offset are always at positions 1 and 2, code is simpler and a...12:04
kenaan12amauryfa default 11a699f3064d8f 15/pypy/: Check type of some ast nodes: return.value must be an expr *subclass*, but not an expr!12:04
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.12:04
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) left irc: Ping timeout: 240 seconds12:05
rinuwhy is it that languages like python does not contain pointers ?12:05
fijalbecause of memory safety12:09
fijalwhat would you want to do with them?12:09
rinu5 (~rinu@117.196.130.167) joined #pypy.12:10
arigatorinu: read e.g. http://www.daniweb.com/software-development/java/threads/12462512:10
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) joined #pypy.12:10
apanda (~apanda@59.178.220.6) left irc: Remote host closed the connection12:10
arigatoor http://mail.python.org/pipermail/edu-sig/2008-May/008528.html12:11
rinu (~rinu@117.196.130.167) left irc: Ping timeout: 240 seconds12:11
ltratt (~ltratt@87.113.171.48) joined #pypy.12:16
rinu5is python the only high level programming language used to write another programming language with use of pypy ? 12:17
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.12:18
arigatoyes12:18
rinu5does python have a virtual machine ?12:20
ltratthi all. I'm trying to get an RPython function to be callable from an external C library. Via the wonders of grep, I eventually stumbled acrosslltype.functionptr, which means I seem to at least somewhat satisfy the RPython type system. But the function I pass as _callable doesn't seem to be translatable into C. "ValueError: don't know how to generate code for <fn start_element>"12:20
ltrattThere is a 'graph' argument to functionptr, but frankly I'm buggered if I can work out what it is, how to generate it, or whether I even need to care about it.12:20
arigatorinu5: CPython and PyPy are usually called virtual machines for Python12:21
fijalltratt: how do you want to pass the function to be called?12:21
fijalltratt: do you want to pass it as an argument to some C call or to have a known name in C?12:22
ltratti want to pass a reference to an RPython function to a C library (libxml2) which will then call back to that RPython function.12:22
arigatoltratt: there is an example: qsort() in pypy.rpython.memory.gctransform.asmgcc12:22
fijalyou can definitely check how we implemented it :)12:23
rinu5then how is the VM of python ( pypy ) different from the popular JVM ?12:23
arigato"the" popular JVM?  there are tons of VMs for Java too12:23
fijalltratt: there are examples in say module/_rawffi/callback.py12:23
ltrattfijal and arigato: thanks, i have looked at that asmgcc. but - to an ignoramus such as i - it's a ton of random gloop.12:23
arigatoltratt: grep for qsort()12:24
fijalor module/pyexpat12:24
rinu5ok but why are languages developed that run on JVM while not for python VM ?12:24
arigatojvm = Java12:24
ltrattaha, llhelper looks like it might be relevant...12:25
fijalrinu5: you mean why is say jython running on jvm but nothing runs on python vms?12:25
rinu5yes12:25
fijalJVM was advertised it's a good way to do that12:26
fijalpython VM not12:26
fijalI think JVM is still a better target, but not a very good one nonetheless12:27
arigatoltratt: actually llhelper() is not really needed in normal cases, I think12:27
fijalit failed to deliver say on performance promises12:27
arigatoltratt: it's needed there because qsort() is a _nowrapper=True function12:27
arigatoltratt: indeed, just look at pypy.rpython.lltypesystem.test.test_rffi12:27
arigatotest_c_callback()12:28
rinu5doesnt pypy convert python code to optimised c code ?12:28
arigatoor test_double_callback()12:28
ltrattarigato: i thoroughly failed to understand what test_c_callback is doing12:28
ltratt:)12:28
arigato"eating_callback" is a C function that takes a callback argument that is RPython12:29
fijalrinu5: no12:29
arigatothat's g()12:29
fijalrinu5: pypy runs python code which is then on the fly compiled using a JIT to the assembler12:29
fijalwhen the code gets hot12:29
fijalpretty much what JVM does for Java12:29
ltrattarigato: hmm.12:29
fijalexcept the architecture is different, but it all looks the same for the user12:29
arigatoyou see there that you can use rffi.CCallback([argtypes..], resulttype) as the type of the callback argument of the C function12:30
ltrattah, i hadn't appreciated ccallback.12:31
`fox` (~fox@95.233.109.194) joined #pypy.12:31
rinu5so pypy converts python code directly to assembly language ? then why cant we use python for anything that we use c/c++ for ?12:31
arigatoas you can see in rffi.py, it's really just calling lltype.FuncType()12:31
ltrattin libxml2, i have to pass a struct with a number of function pointers - currently i was typing them as FuncType. i guess they should be CCallback?12:32
arigatoltratt: but it's defined in rffi.py just because it gives a more uniform api12:32
fijalrinu5: we pretty much can, pypy is pretty fast12:32
arigatoltratt: ah, I see.  no, CCallback is just the same as Ptr(FuncType(..))12:32
ltrattah.12:32
fijalalso, just converting to asm does not mean it's efficient12:32
arigatobut a struct containing function pointers really requires you to populate it with llhelper()12:33
ltrattso for each field in the sruct, i assign to the pointer with an llhelper basically?12:33
arigatoyes12:33
rinu5i mean really hard stuff like writing a kernel12:33
ltrattarigato: thanks. let me give it a go...12:33
arigatoltratt: and you probably need to do it in RPython code, as opposed to having such a struct prebuilt12:33
ltrattyep, that bit i'd guessed :)12:34
ltratti'm mallocing memory for the blank struct, then manually populating it12:34
arigatowe have some issues with prebuilt llhelpers12:34
arigatook12:34
fijalrinu5: for a kernel you need some more low-level stuff12:34
fijallike pointers :)12:34
fijalso python is maybe not very good as a language12:34
fijalarigato: I think they either don't work completely or they do work from some point in the past12:35
arigatosee the long comment at the start of rpython.annlowlevel.llhelper()12:35
fijalat some point I either made the error obvious or implemented them :)12:35
arigatofijal: yes, something like that12:35
arigatogood :-)12:35
fijalbut I don't remember which12:35
ltrattarigato: ok, it now compiles :)12:35
ltrattso that's step #112:36
arigatoI think at some point I stumbled over them not working and hacked a special case (but the general case still doesn't work)12:36
arigatoltratt: llhelper(F, f) basically casts the RPython f to a low-level Ptr(FuncType(..)) type12:36
rinu5thats why asked about pointers guys.what are other things needed other than pointers ?12:37
ltrattarigato: yep, llhelper is definitely what i needed12:37
fijalrinu5: pointers you can get via ctypes12:37
fijalbut it's inconvinient12:37
fijalreal time is the hardest part probably12:38
ltrattand then i appear to need to pass a Ptr(FuncType) for F12:38
rinu5thank you so much fijal and arigato , you guys are so friendly...12:39
arigatoltratt: yes, which is the same as you need in the Struct declaration for the fields12:39
rinu5have a good day!12:39
arigatorinu5: see you!12:39
rinu5 (rinu@117.196.130.167) left #pypy.12:39
ltrattarigato: indeed12:40
arigatoltratt: if you feel like writing complete libxml2 RPython bindings, we would certainly be interested in using them for pypy too :-)12:42
ltrattat the moment, i am just trying to get the damned thing to compile ;)12:43
arigato:-)12:43
ltrattconverge only uses a very small subset of libxml2's functionality. a full XML library for all the horrible corner cases is a large amount of work...12:43
fijalltratt: it's maybe worth noting *with tests*12:43
arigatoltratt: yes, I feared so12:44
ltrattarigato: and, worse, most of it is for stuff that no-one in their right mind would ever want to use...12:45
arigatono point in complaining about xml to me :-)12:46
ltratthehe, yes, i guess I am amongst friends in this regard ;)12:46
fijal:-)12:46
arigatollibxml2 is one of the most-wanted third-party Python library for pypy12:46
arigatobut none of us cares :-/12:47
ltratti can hardly say i blame you ;)12:47
ltrattat the very least, the small part implemented in the Converge VM will give someone a head-start if they're brave enough to implement the whole thing as an RPython library.12:48
ltratt[assuming i get it to work ;)]12:48
arigatofijal: ah, or was it lxml?  I think it's a different set of bindings over the same C librar12:48
arigatoy12:48
fijallxml12:49
fijallxml is a mess though12:49
fijalI looked at the code one day12:49
ltrattlxml is a layer over libxml2, so i guess it would still want native libxml2 bindings underneath12:49
fijalltratt: it vastly depends how much you want to reimplement this12:50
fijalbut it occurs to me there are not that many people who would love such job12:50
ltratti agree wholeheartedly on both points!12:50
ltrattthis is the sort of thing you have to pay people to do, because it's too damned boring to do voluntarily...12:50
fijalyes12:51
arigatofijal: it's Cython, right?12:51
fijaland I don't quite see anyone caching out on that12:51
fijalhell, I probably wouldn't even like to do it for money too much12:51
fijalarigato: yes, but a reasonably ugly cython12:51
fijalit's C12:51
fijalwith usage of CPython C API12:51
stakkars__ (~tismer@p5DDB72FD.dip.t-dialin.net) joined #pypy.12:52
stakkars_ (~tismer@p5DDB72FD.dip.t-dialin.net) left irc: Read error: Connection reset by peer12:52
fijalarigato: I envy CPython devs, they can measure a 2% increase in speed13:01
arigato:-)13:01
arigatoLinux kernel devs measure 0.15% increases13:01
kenaan12fijal refactor-signature 11e3cef40afa0d 15/pypy/module/micronumpy/interp_numarray.py: clean up descr_len13:01
fijalwow13:03
fijalI guess they don't have other-stuff-that-runs-at-the-same-time to worry about13:03
arigatoyes13:03
ltrattMy multitime program is a dirt simple way of getting some idea whether "stuff that ran at the same time" corrupted results. http://tratt.net/laurie/src/multitime/13:05
ltratti warn you though: it is dirt simple13:05
fijalltratt: better than no solution at all13:05
ltrattit's proved useful to me13:05
ltrattif you run something 5 times and the standard deviations are small, it gives a reasonable level of confidence about the numbers13:06
fijalI usually do this anyway, but looking by hand13:06
ltrattcool, well multitime might be useful to you then. it's certainly easier to have a script do the maths than try and do it in my head ;)13:08
fijalyes :)13:08
arigato:-)13:08
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 252 seconds13:11
Action: fijal realizes "now we don't need all this code any more"13:11
Action: arigato struggles to find the simplest model to clean up the NEWxxx operation mess in the backend13:13
tav_ (~tav@host-92-20-7-192.as13285.net) left irc: Ping timeout: 240 seconds13:14
ojii (~ojii@147-35.62-81.cust.bluewin.ch) joined #pypy.13:14
ojii (~ojii@147-35.62-81.cust.bluewin.ch) left irc: Client Quit13:15
tav_ (~tav@host-92-20-7-192.as13285.net) joined #pypy.13:21
fijalarigato: op_malloc branch?13:21
kenaan12hakanardo jit-multilabel 114054e3b40851 15/pypy/jit/metainterp/optimizeopt/unroll.py: cleanup13:22
kenaan12hakanardo jit-multilabel 112653c963abd8 15/pypy/jit/metainterp/: rename start_resumedescr to resume_at_jump_descr13:22
arigatofijal: yes13:22
fijal:/13:22
fijalof course part of the problem is the boehm support13:22
arigatonot really13:22
fijalit was a source of some mess there last time I checked13:23
fijalso what's the issue?13:23
fijalmaybe you can rubber duck it with me, I don't seem to be any usable otherwise13:23
arigatoI just want to remove as many special cases as possible from the backend, and move it to llsupport13:24
arigatoI'm thinking now about adding these two backend-only operations:13:24
arigato+    'CALL_MALLOC_GC/*d',      # like CALL, but NULL => propagate MemoryError13:24
arigato+    'CALL_MALLOC_NURSERY/1',  # nursery malloc, const number of bytes, zeroed13:24
fijaland they would replace pretty much everything else?13:28
arigatoyes13:28
fijalsounds good to me13:29
arigatoall the NEWxxx operations13:29
arigatook :-)13:29
fijalit seemed back in the day that a coarse-grained resop is a good think for the backend :)13:29
fijalwhat if CALL_MALLOC_NURSERY fails?13:29
fijalis it like NEW?13:29
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) joined #pypy.13:30
arigatono, CALL_MALLOC_NURSERY really means "try to allocate inline, if that fails do the CALL, and that can in turn fail like CALL_MALLOC_GC"13:30
fijalok13:30
fijalyou know what would be really cool, although hard?13:30
arigato? .-)13:30
fijalto be able to express the conditional stuff in resops13:30
arigatoyes13:30
JaRoel (~jaroel|4d@sink.jaroel.nl) joined #pypy.13:30
fijalso we don't have to handcode it13:31
`fox` (~fox@95.233.109.194) left irc: Ping timeout: 240 seconds13:31
arigatoyes, although I'm not sure it would end up with the same efficient register allocation13:31
arigatoe.g. COND_CALL_stuff13:31
arigatowe can express the check of the byte with several INT_xxx operations13:31
arigatobut that means register allocation13:32
arigatoinstead in COND_CALL_xx it's special-cased13:32
arigatousing particular instructions like TEST [memory], 0xvalue13:32
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) left irc: Read error: Operation timed out13:33
arigato"obviously" the right thing is to expand it as several instructions and have a very good register allocator able to recombine it, but well...13:33
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds13:35
fijalthe thing we completely don't have right now is merging13:38
fijalif you don't have merging, it's a bit pointless13:38
arigatoyes13:38
fijalbut I guess it's possible to play with a better register allocator13:38
fijalmore importantly you would be able to express some things like list_resize or dict_lookup as a small conditional barrier13:38
fijalso you would win somewhere else13:38
arigatoyes13:39
fijalit's probably well worth trying13:41
Action: fijal is moving some stuff on numpy from non-obviously unsupported -> officially unsupported -> removed -> supported13:41
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds13:44
fijalit occurs to me "there is support, but it does not work" is worse than "it's unsupported"13:45
Nick change: DasIch_ -> DasIch13:47
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) joined #pypy.13:58
ltratthmm, head seems broken to me: http://pastebin.com/zRJQsg4M13:59
fijalltratt: uh yes, I broke it13:59
fijalltratt: let me fix it14:00
ltrattok :)14:00
kenaan12fijal refactor-signature 112e85e96a3a5b 15/pypy/module/micronumpy/: Refactoring in progress - code removal.14:01
kenaan12fijal default 11e42742946b48 15/pypy/tool/gcc_cache.py: Temporarily revert 976baeae86ed, seems to break some stuff for obscure reasons14:01
fijalkenaan: hello?14:01
fijalok good14:01
fijalltratt: "for now"14:01
pybot joined #pypy.14:45
-ChanServ (ChanServ@services.)- [#pypy] Welcome!14:45
-ChanServ (ChanServ@services.)- [#stackless] Welcome to #stackless. This channel is for the discussion of matters relating to Stackless Python. Remember that others in this channel may not be in the same timezone as you, so stick around for an answer (even if only in your channel history) if you ask a question. The mailing list is also recommended as being a faster place to get help (http://www.stackless.com/mailman/listinfo/stackless).14:46
izibiexarkun: one time i got this traceback: http://paste.pocoo.org/show/521920/14:48
izibibut only one time14:48
exarkunThat seems useful.14:48
izibiexarkun: and i got another one: http://paste.pocoo.org/show/522085/14:49
izibibut most times the process just exits14:50
`fox` (~fox@95.233.109.194) joined #pypy.14:51
exarkunSomeone else will have to help interpret those, though, I don't know what they mean.14:51
kenaan12hakanardo jit-multilabel 11c69acd44808e 15/pypy/jit/metainterp/optimizeopt/unroll.py: rename to indicate that it belongs to the short_preamble and not the current part14:51
kenaan12hakanardo jit-multilabel 111dfa4f95ba02 15/pypy/jit/metainterp/optimizeopt/unroll.py: not needed since the hack adding ops at the begining of newoperations was removed14:51
kenaan12hakanardo jit-multilabel 1169c7bd4e7bac 15/pypy/jit/metainterp/optimizeopt/unroll.py: always assign self.inputargs14:51
kenaan12hakanardo jit-multilabel 119e63288cc721 15/pypy/jit/metainterp/optimizeopt/unroll.py: labels always have descr14:51
`fox` (~fox@95.233.109.194) left irc: Ping timeout: 252 seconds14:59
rguillebertizibi, it might be in a library15:01
izibirguillebert: one thing i could imagine: arch links it to different library versions than the official binary15:01
izibithat's why i can't test it with the official one. the libs it's linked again are missing15:02
rguillebertyou can monkey patch sys.exit to see where it's called (if it's called by python code)15:03
nickik (~nick@217.150.239.67) joined #pypy.15:04
izibihttp://paste.pocoo.org/show/522093/15:05
izibionly differences are libssl and libcrypto15:06
rguillebertwhich one doesn't work ?15:08
izibi/usr/bin/pypy is the one from arch and doesn't work15:08
izibican't test the official build as lib{ssl,crypto}.so.0.9.8 is missing here15:09
izibii'll compile an older version of openssl and we'll see15:09
rguillebertyour tracebacks don't seem to have anything to do with linking15:12
exarkunif I understand, the tracebacks are from the arch-supplied PyPy, which is linked against the arch-supplied openssl properly15:12
exarkunand izibi hasn't tested moin in the official PyPy binary release, because that one is linked against libraries he doesn't have15:13
squiddy (~squiddy@g224197016.adsl.alicedsl.de) left irc: Read error: Connection reset by peer15:14
izibiexarkun: right15:14
squiddy (~squiddy@g224197016.adsl.alicedsl.de) joined #pypy.15:14
izibimaybe it's more an arch issue, i don't know15:15
arigatoizibi: you can use the more recent nightly builds15:15
arigatothey don't have the libssl.so.0.9.8 issue15:15
arigato(which is just a warning anyway)15:15
izibiarigato: are they linked agains openssl 1.0.0?15:16
arigatono, but as far as I know it works against openssl 1.0.0 too15:16
arigatoI may be wrong15:16
rguillebertbut you still need to recompile it :)15:16
arigatono15:16
arigatoor maybe yes and then I'm sorry for giving misleading comments :-)15:17
rguillebertthe last patch I saw about openssl was because debian used openssl1.0.0 (which dropped SSLv2 afaik) and compilation failed because it wanted the SSLv2 functions15:19
rguillebertmaybe there was a patch after that15:19
izibino, nightlies are linked against 0.9.815:21
rinu5 (~rinu@117.213.60.61) joined #pypy.15:22
rguillebertthey probably aren't ABI-compatible anyway15:24
exarkunThey're like 97% ABI compatible. :)15:26
exarkunAnd PyPy really probably does stick to the compatible part15:26
exarkunBut everything has to suck, because only the ABI as a whole is versioned, and there's no tools that can look at the application and the library and tell you the subset in use is actually safe.15:27
rguillebertizibi, can you try LD_PRELOAD="/path/to/openssl" pypy-1.7/bin/pypy ?15:30
rguillebert(I know...)15:30
rguillebertLD_PRELOAD="/usr/lib/libssl.so.1.0.0" pypy-1.7/bin/pypy15:33
izibi./pypy-1.7/bin/pypy: error while loading shared libraries: libssl.so.0.9.8: cannot open shared object file: No such file or directory15:34
jimbaker` (~jbaker@c-75-71-80-146.hsd1.co.comcast.net) left irc: Ping timeout: 240 seconds15:37
izibii can start the official pypy build with a self-compiled lib{ssl,crypto}, let's see what happens15:37
Shanita (~John@osbk-4d087f7b.pool.mediaWays.net) joined #pypy.15:37
rguilleberthttp://aur.archlinux.org/packages.php?ID=3630815:38
Kaskuka (~John@osbk-4db060c4.pool.mediaWays.net) left irc: Ping timeout: 240 seconds15:38
izibiyeah, just compiled that myself :D15:39
rguillebertoh yes, it's not binaries :p15:40
izibiand it still crashes15:41
izibiwith the official pypy build15:41
rguillebertwith the same error ?15:42
izibiwith no error as it does most times15:42
izibihttp://paste.pocoo.org/show/522108/15:43
izibisome more output before this but just logging information as you can see it there15:43
Nick change: rinu5 -> rinu15:48
rinu (rinu@117.213.60.61) left #pypy.15:49
rguillebertI guess you need to use gdb to know where it crashes in C code15:56
rguillebertor pdb in python code, but it's probably C code15:56
izibii've never used gdb. i can try if someone tells me step by step what to do ;)16:00
exarkungdb --args /path/to/pypy [args to run moin]16:01
exarkunYou'll get a gdb prompt16:01
exarkunwhich lets you do lots of things16:01
exarkunif you want a breakpoint on `exit´, then `break exit´16:02
exarkunif you want to run the program, `run´16:02
exarkunwhen the program is running, C-c will interrupt it16:02
exarkunand drop you into a gdb prompt again16:03
exarkunfrom which point you can inspect things, like the C call stack, `bt´16:03
exarkunor resume the program, `continue´16:03
gehho2 (~gehho2@ZO060028.ppp.dion.ne.jp) left irc: Quit: Leaving...16:03
exarkunif it actually crashes, you should get the gdb prompt again and you can see where it crashed16:03
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.16:03
rinu5 (~rinu@117.196.144.250) joined #pypy.16:03
exarkunif it hits a breakpoint, you'll get the gdb prompt as well16:03
izibiexarkun: "Function "exit" not defined." on break exit. is that normal?16:05
izibiMake breakpoint pending on future shared library load? (y or [n])16:05
exarkunThat doesn't sound normal, no.16:05
exarkunMaybe it'll be okay16:05
exarkunIt's worth a try at least.16:05
izibianyway, i'll try to crash it and we'll see16:05
izibii'll try pypy from the arch package first16:06
arigatoTooMuchClevernessAlwaysComeBackToHitYouError16:08
rinu (~rinu@117.216.74.81) joined #pypy.16:08
rinu5 (~rinu@117.196.144.250) left irc: Ping timeout: 276 seconds16:08
arigatollsupport/descr.py cleverly ensures that some fields like _is_float_field are actually class attributes, not stored on every instance16:09
rinu (rinu@117.216.74.81) left #pypy.16:09
arigatobut of course this was made bogus by the introduction of DynamicFieldDescr16:09
arigatoso now descr.py contains useless obscure cleverness16:09
izibiexarkun: Breakpoint 1, 0x00007ffff5ee74c0 in exit () from /lib/libc.so.616:10
exarkunizibi: cool16:10
rguillebertnow try bt16:10
rguillebertthe command "bt"16:10
exarkunand cross your fingers16:10
izibihttp://paste.pocoo.org/show/522120/16:11
exarkunSo sad16:11
rguilleberterf16:11
peter_zfs (~a@ppp-94-65-219-255.home.otenet.gr) left irc: 16:11
exarkunIt sorta looks like a return-from-main exit?16:11
exarkunBut I don't know16:11
arigatosys.exit() would cause that16:12
exarkunah.16:12
arigatoit raises SystemExit that is caught in the main()16:12
exarkunizibi looked around at the Python code, he didn't find anything there that explains this behavior16:12
arigatopatch sys.exit at the level of Python, I suppose16:13
izibimonkeypatching? or is there a better way?16:14
arigatomonkeypatching is good .-)16:14
rinu5 (~rinu@117.196.135.63) joined #pypy.16:15
arigatothough that wouldn't catch "raise SystemExit(245)"16:15
arigatobetter actually to write (in Python) around the main function:16:15
arigatotry:16:15
arigatoexcept SystemExit, e:16:15
rinu5 (rinu@117.196.135.63) left #pypy.16:15
arigato   import traceback; traceback.print_exc()16:16
arigatosys.exit() just raises SystemExit, so you don't need to patch it if you add the "except SystemExit" trick.16:16
squiddy (~squiddy@g224197016.adsl.alicedsl.de) left irc: Read error: Connection reset by peer16:19
izibii have an idea what could be going on, let's see16:20
izibiwerkzeugs reloader starts a new process. and it just passes through the exit code16:29
izibiyes, it just passes through the exit code16:30
izibihttp://paste.pocoo.org/show/522128/16:32
exarkunSounds like the werkzeugs reloader is buggy, then, if it's turning a segfault into exit status 245.16:36
nedbat (~nedbat@python/psf/nedbat) joined #pypy.16:38
exarkunHm.  PyCrypto generates incorrect RSA signatures on PyPy. :(16:38
izibihttps://github.com/mitsuhiko/werkzeug/blob/master/werkzeug/serving.py#L56216:39
exarkunYea, that's bogus.16:40
exarkunsubprocess.call does not return a value that can be passed meaningfully to sys.exit.16:41
exarkunMaybe you should switch to hosting moin in Twisted. >:P16:41
exarkunFor starters, can you try to reproduce the moin/pypy crash without using the reloader?16:42
izibilet's see16:45
izibii guess with the reloader it would have crashed by now16:45
izibii get errors because of too many open files, but that's a different thing ;)16:46
izibiso who's doing something wrong? werkzeug or pypy? :D16:47
exarkunIf you can't get moin to crash without using werkzeug's reloader, then I'd try getting a much simpler application to crash using werkzeug's reloader16:48
exarkunAnd then perhaps you can either blame werkzeug, or at least have a much simpler example of a bug in PyPy.16:49
exarkunIt's probably also a good idea to file a bug against werkzeug for the subprocess.call/sys.exit confusion.16:50
exarkunThat's just a bug in werkzeug that should be fixed.16:50
exarkun(but it's only making things confusing in this case, I doubt it's actually the cause of the crashes)16:51
izibi"Run the command described by args. Wait for command to complete, then return the returncode attribute."16:51
exarkunizibi: The RPython tracebacks you reported earlier _are_ PyPy bugs though.  You should try to get arigato to look at them. :)16:51
izibiwhat's the difference between returncode and the value you pass to sys.exit? i guess i'm also confused by that :D16:52
exarkunizibi: For example, if a process ends because it receives SIGSEGV, then subprocess.call will return -1116:52
exarkunizibi: But sys.exit(-11) has a completely different meaning16:52
arigatoand of course 245 = 256-1116:53
izibiah! that explains the return code (or however you want to call it) :D16:53
jimbaker (~jbaker@c-75-71-80-146.hsd1.co.comcast.net) joined #pypy.16:58
jimbaker (~jbaker@c-75-71-80-146.hsd1.co.comcast.net) left irc: Changing host16:58
jimbaker (~jbaker@canonical/jimbaker) joined #pypy.16:58
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.17:00
ltrattI think I am doing something dumb with rffi as libxml2 is segfaulting on me. Here's a really minimal libxml2 using C program which doesn't segfault http://pastebin.com/r9zQAXc2 and here's my RPython code http://pastebin.com/XhRpWd9y17:04
ltratti wonder if i am doing something with the strings that i am passing in...17:05
squiddy (~squiddy@g224197016.adsl.alicedsl.de) joined #pypy.17:05
ltrattessentially, i get a segfault in a method called from xmlSAXUserParseMemory in the RPython version17:05
fijal (~fijal@197.175.9.70) joined #pypy.17:06
arigatoare you sure you don't need to initialize 'w'?17:09
arigatoit's not zero-filled17:09
ltrattw? h?17:10
ltratti thought everything was zero filled by default? certainly printing out a couple of parts seemed to suggest it was.17:10
izibiany idea how to find out what's the last call in the python code before it crashes?17:10
ltrattarigato: is there an easy way to zero-fill? i can't see a memset definition lurking anywhere obvious.17:11
arigatoltratt: no, neither malloc nor scoped_alloc zero-fill by default17:11
arigatoyou can call scoped_alloc(xmlType, zero=True)17:12
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 276 seconds17:12
arigatoor in C, call calloc() instead of malloc()17:12
gehho2 (~gehho2@ZJ040192.ppp.dion.ne.jp) joined #pypy.17:13
arigatoizibi: hard.  I would suggest to run it in gdb in a version of pypy compiled with debugging symbols17:13
ltrattok, i could swear i'd seen a comment in PyPy suggesting it was zero filled. My bad! I'm trying "zero=True" now.17:13
arigatoizibi: if that's beyond you, you can put a bug report instead17:13
fijaluh, do we have a segfault?17:14
arigatoizibi: please describe everything that we need to reproduce the crash17:14
arigatofijal: yes17:14
fijal:/17:14
arigatono clue, maybe it's a ctypes misusage or something17:14
fijalis cpyext involved?17:14
fijalizibi: ah, by the way. are you using jinja2?17:14
fijalif so, is this the very last release of jinja2?17:14
ltrattarigato: you're a star - zero=True does what i hoped. i can't believe i fell for that particular mistake. c'est la vie ;)17:15
izibiyes, moin uses ninja217:15
arigato:-)17:15
overminder (~overminde@wf105-220.ust.hk) left irc: Ping timeout: 276 seconds17:16
fijalizibi: upgrade your jinja217:16
fijalthe previous versions of jinja do very dubious things with ctypes17:16
fijallike manipulating tracebacks17:17
maxyz (~maxy@186.18.94.35) left irc: Ping timeout: 240 seconds17:17
izibifijal: i'm currently on 2.617:18
maxyz (~maxy@186.18.94.35) joined #pypy.17:19
fijalshould be good17:20
Action: fijal off17:20
kenaan12arigo op_malloc_gc 11fc4c51ca1ec4 15/pypy/jit/backend/llsupport/: In-progress.17:22
kenaan12arigo op_malloc_gc 112011f2788757 15/pypy/jit/backend/llsupport/test/test_rewrite.py: Fix some more tests.17:22
kenaan12arigo op_malloc_gc 11ac65509a7f46 15/pypy/jit/backend/llsupport/: In-progress.17:22
kenaan12arigo op_malloc_gc 119d09995aae35 15/pypy/jit/backend/llsupport/: Finish test_rewrite.TestFramework.17:22
kenaan12arigo op_malloc_gc 11bab5597339cc 15/pypy/jit/: Start to reorganize dsecr.py to kill the obscure (and now half pointless) inheritance.17:22
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Ping timeout: 268 seconds17:23
arigato (~arigo@fwstups.cs.uni-duesseldorf.de) left irc: Quit: See you17:23
fijal (~fijal@197.175.9.70) left irc: Ping timeout: 255 seconds17:24
ltratt (~ltratt@87.113.171.48) left irc: Quit: using sirc version 2.211+KSIRC/1.3.1217:30
izibithis is how to reproduce the segfault: http://paste.pocoo.org/show/522161/17:33
izibi(at least on archlinux, didn't test it elsewhere)17:33
izibimaybe someone can test this on another distribution17:34
izibioh and you have to run a large number of http requests to crash it17:35
izibifor example with apache bench: ab -n 1000 -c http://127.0.0.1:8080/Home17:35
Da_Blitzizibi: have you tried lodging a bug on the bug tracker?17:36
izibiDa_Blitz: no, not yet, but it would be nice to know if this is specific to arch linux17:37
kkris (~kris@93-82-38-56.adsl.highway.telekom.at) left irc: Ping timeout: 252 seconds17:40
stakkars_ (~tismer@p5DDB72FD.dip.t-dialin.net) joined #pypy.17:43
stakkars (~tismer@p5DDB72FD.dip.t-dialin.net) left irc: Read error: Connection reset by peer17:43
Nick change: stakkars_ -> stakkars17:43
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) left irc: Read error: Connection reset by peer17:48
stakkars_ (~tismer@p5DCE28BD.dip.t-dialin.net) joined #pypy.17:51
kkris (~kris@93-82-35-32.adsl.highway.telekom.at) joined #pypy.17:57
Da_Blitzizibi: i have it exit after a couple of reqeusts as well on debian with pypy 1.7 from the website17:59
izibiDa_Blitz: echo $?18:01
izibidoes it output 245?18:01
Da_Blitzone sec18:02
Da_Blitzthats the one18:02
Da_Blitzin case you need it, debian 6.0.318:03
izibithx18:03
Da_Blitznp18:03
rguillebertI'll investigate after dinner18:12
`fox` (~fox@95.233.109.194) joined #pypy.18:15
stakkars_ (~tismer@p5DCE28BD.dip.t-dialin.net) left irc: Quit: schnarch18:28
`fox` (~fox@95.233.109.194) left irc: Quit: STACCA STACCA STACCA18:29
DasIch_ (~dasich@p3E9903C3.dip.t-dialin.net) joined #pypy.18:31
DasIch (~dasich@p3E991D99.dip.t-dialin.net) left irc: Ping timeout: 252 seconds18:35
arigato (~arigo@89.204.154.123) joined #pypy.18:37
gehho2 (~gehho2@ZJ040192.ppp.dion.ne.jp) left irc: Ping timeout: 252 seconds18:40
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) left irc: Ping timeout: 244 seconds18:40
exarkunrguillebert: How's cython/pypy these days?  Still sleeping?18:49
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy18:50
stakkars_ (~tismer@p5DCE28BD.dip.t-dialin.net) joined #pypy.18:50
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy.18:56
kenaan12hakanardo jit-multilabel 114014ad4062c5 15/pypy/jit/metainterp/optimizeopt/: refactor a bit and make sure import_state import_state sets things up properly even if there is nothing ...19:02
kenaan12hakanardo jit-multilabel 117f2f31e8e4bd 15/pypy/jit/metainterp/: allow the metainterpreter to request an label by placing it in the unoptimized trace19:02
kenaan12hakanardo jit-multilabel 11bd3c0ac15e47 15/pypy/jit/metainterp/: kill old retrace logic no longer used19:02
kenaan12hakanardo jit-multilabel 11e65693cb1de5 15/pypy/jit/metainterp/: kill old interface for calling optimizeopt, its no longer used19:02
kenaan12hakanardo jit-multilabel 119bc137b88407 15/pypy/jit/metainterp/optimizeopt/unroll.py: moved retrace count check up one level19:02
kenaan12hakanardo jit-multilabel 11f98a2c88a795 15/pypy/jit/metainterp/optimizeopt/unroll.py: kill did_import19:02
kenaan12hakanardo jit-multilabel 114f6581643f66 15/pypy/jit/metainterp/optimizeopt/unroll.py: translation fixes19:02
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/637 [12hakanardo, jit-multilabel]19:03
nedbat (~nedbat@python/psf/nedbat) joined #pypy.19:03
rguillebertexarkun, yes, I'm writing an email to get in merged in the main repo19:12
rguillebertbut I'll work on it at the next sprint19:13
exarkuncool19:17
stakkars_ (~tismer@p5DCE28BD.dip.t-dialin.net) left irc: Quit: schnarch19:33
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Read error: Connection reset by peer19:35
kennethr_ (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.19:35
stakkars_ (~tismer@88.130.184.37) joined #pypy.19:49
squiddy_ (~squiddy@g224197016.adsl.alicedsl.de) joined #pypy.19:54
squiddy (~squiddy@g224197016.adsl.alicedsl.de) left irc: Read error: Connection reset by peer19:54
izibiDa_Blitz: did you test on i386 or amd64 btw?20:01
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/637 [12hakanardo, jit-multilabel]20:03
rguillebertizibi, ". ./env/bin/activate"20:04
rguillebertisn't "source ./env/bin/activate" ?20:04
rguilleberts/isn't/isn't is supposed to be/20:05
izibirguillebert: both work20:05
rguillebertI'm not sure20:05
Nick change: squiddy_ -> squiddy20:05
rguillebertoh yes nevermind20:06
izibirguillebert: "This builtin ['.'] is equivalent to source." http://www.gnu.org/s/bash/manual/html_node/Bourne-Shell-Builtins.html20:09
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds20:10
rguillebertyes I thought it was just an equivalent of "bash"20:11
rguillebertAKA run the script in a subprocess20:11
nirbheek"./" != ". "20:13
nirbheekin the former it's a path specifier, in the latter it's a command20:13
rguillebertI have the same bug20:17
rguilleberton 32bits20:17
izibiThomasWaldmann tested on ubuntu 11.10 32bit and there it worked. very strange20:25
rguillebertI have the same distribution20:27
rguillebertI did 100 requests in parallel20:28
izibithere's only one worker thread so this doesn't matter20:28
izibii can crash it by holding down F5 in a browser20:28
rguillebertit works properly with an old pypy I had on my hdd20:33
izibisounds like fun with bisect? :D20:36
mtigas (~mtigas@c-67-185-214-42.hsd1.wa.comcast.net) joined #pypy.20:38
rguillebertI'm translating with debugging symbols20:45
rguillebertso we'll see20:46
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.20:46
maxyz (~maxy@186.18.94.35) left irc: Ping timeout: 268 seconds20:54
maxyz (~maxy@186.18.94.35) joined #pypy.20:55
arigatoizibi: fwiw I'm still getting OperationalError: (OperationalError) near " ": syntax error u'INSERT INTO etc20:57
arigatoI remember being told that it's ok if you start moin once with CPython, but well20:57
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.21:10
hruske (~Gasper@188-230-156-183.dynamic.t-2.net) joined #pypy.21:11
kennethr_ (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Textual IRC Client: http://www.textualapp.com/21:28
kenaan12amauryfa py3k 112258d6a11e06 15/: hg merge defaults21:31
kenaan12amauryfa py3k 116a82b1beaf24 15/pypy/module/_ssl/: Remove "issuer" and "server" attributes from ssl socket. + Fix shutdown test21:31
kenaan12amauryfa py3k 118d91b7ba9f7c 15/pypy/: Fix a crash when the 'str' unwrap_spec is given a non-ascii unicode string. Now it automatically converts to utf_8 ...21:31
kenaan12amauryfa py3k 114c7922f5a847 15/pypy/interpreter/pyparser/data/Grammar3.2: Remove "except Exception, e" syntax. Expected to break a lot of code...21:31
kenaan12amauryfa py3k 113f837e62c8c0 15/pypy/: Fix compiler: the exception handler target is now a NAME (and not any assignment target as before). Still missing: ...21:31
kenaan12amauryfa py3k 11c13feede4946 15/pypy/module/_multibytecodec/: Unicode fixes in _multibytecodec module21:31
kenaan12amauryfa py3k 11e20a4198d35c 15/pypy/module/bz2/: Unicode fixes for the bz2 module21:31
kenaan12amauryfa py3k 118d890adadd94 15/pypy/module/bz2/: Fix BZ2File by importing the applevel version from CPython 3.3.21:31
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1474 [12amaury, py3k]21:31
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1474 [12amaury, py3k]21:33
mtigas (~mtigas@c-67-185-214-42.hsd1.wa.comcast.net) left irc: Quit: mtigas21:41
hruske (~Gasper@188-230-156-183.dynamic.t-2.net) left irc: Read error: Operation timed out21:44
kenaan12amauryfa py3k 11c0d4b0fdc8f1 15/pypy/module/bz2/app_bz2file.py: Try to fix translation21:45
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/1475 [12amaury, py3k]21:45
derdon (~derdon@p5DE8912A.dip.t-dialin.net) joined #pypy.21:46
lahwranooh, py3k22:00
lahwranhow is that coming, by the way? on a scale of 0-2 perhaps, where 0 = starting, 1 = not sure, and 2 = almost done22:01
lahwranor maybe a bit less detailed; first half or second half?22:04
MostAwesomeDudelahwran: How do you quantify such a thing?22:04
MostAwesomeDudeI guess if we had a page somewhere with the list of tasks required to support it...22:04
lahwranI was hoping someone working on it might be able to go "well, it doesn't feel very done, so probably a 0"22:04
lahwranor "it feels pretty close, maybe a 2"22:05
exarkunI expect it's a lot closer to 0 than 2.22:05
lahwranoic.22:05
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 248 seconds22:06
mwhudson (~mwh@linaro/mwhudson) joined #pypy.22:13
Arfrever (~Arfrever@apache/committer/Arfrever) joined #pypy.22:18
nedbat (~nedbat@python/psf/nedbat) joined #pypy.22:29
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/74822:30
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/17422:30
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/98122:30
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 245 seconds22:34
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy.22:37
tav_ (~tav@host-92-20-7-192.as13285.net) left irc: Ping timeout: 252 seconds22:39
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) joined #pypy.22:39
arigato (~arigo@89.204.154.123) left irc: Ping timeout: 252 seconds22:42
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds23:10
kkris (~kris@93-82-35-32.adsl.highway.telekom.at) left irc: Quit: Leaving.23:20
verte (~verte@python/site-packages/verte) joined #pypy.23:24
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) left irc: Quit: quit23:24
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.23:25
maxyz (~maxy@186.18.94.35) left irc: Ping timeout: 268 seconds23:25
stakkars_ (~tismer@88.130.184.37) left irc: Quit: schnarch23:26
maxyz (~maxy@186.18.94.35) joined #pypy.23:27
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds23:52
--- Sun Dec 18 201100:00

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