#pypy IRC log for Sunday, 2012-01-29

stakkars_ (~tismer@i59F6BDF0.versanet.de) left irc: Quit: schnarch00:04
nedbat (~nedbat@python/psf/nedbat) joined #pypy.00:08
CIA-603dripton 07roundup * 10#950/startup_hook in readline: 00:21
CIA-6I applied hpaulj's changes, translated pypy, and verified that the fix made it00:21
CIA-6work like CPython on the test script from PyMOTW.00:21
CIA-6It's up as00:21
CIA-6h ... * 14https://bugs.pypy.org/issue95000:21
csenger (~user@dslb-084-058-130-062.pools.arcor-ip.net) left irc: Quit: ERC Version 5.3 (IRC client for Emacs)00:22
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 252 seconds00:38
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 276 seconds00:45
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/79001:04
fijalhm01:36
fijalwhy is set equality slow?01:36
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.01:37
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 245 seconds01:39
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/4901:39
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/103801:39
lahwranI believe I've asked this before, but I don't remember getting much of an answer:01:43
lahwrandoes pypy's GIL have the same contention problems cpython's does?01:43
rguillebertI think Armin reimplemented the GIL to match the last (improved) behaviour01:44
htoothroti was hoping they'd give that BF scheduler a fair shake01:44
htoothrotiirc the existing GIL still has some issues01:45
fijallahwran: I think STM is the current answer01:45
lahwranSTM sounds cool01:46
lahwranI'm just trying to make sqlalchemy's ORM play nice with high-performance twisted :x01:46
lahwrandoubtful STM will be the magic bullet that makes that work01:46
htoothrotwhy are you suspecting GIL issues01:47
lahwranI don't dare mess with cpython threading on a multi-core system, though01:47
lahwranhtoothrot: threading is a possible solution (admittedly not the best one ever)01:47
lahwrancpython has serious contention issues on multi-core systems, though01:48
htoothrotin certain cases 01:48
htoothrotwhy do you suspect you are hitting one of those cases?01:48
rguillebertdoes twisted uses threads ?01:48
CIA-603fijal 07roundup * 10#1027/Do something with longs: 01:49
CIA-6[new] A bit no clue what, but performance of longs is fairly disastrous right now, so I01:49
CIA-6suppose we should do something with it (not attaching an ... * 14https://bugs.pypy.org/issue102701:49
htoothrotrguillebert, it has facilities for using them, eg the adbapi module01:49
htoothrotwhich uses a threadpool to handle calls through DB-API modules 01:49
htoothrotbut the idea is not to use threads01:50
rguillebertso "high-performance twisted" means twisted with no thread right ? :)01:50
kenaan12dripton default 11c8bec90e96cf 15/lib_pypy/pyrepl/: (hpaulj) Fix startup_hook in readline to work like CPython.  Fixes issue95001:51
fijallahwran: high performance & sql alchemy don't go in one sentence01:51
CIA-603fijal 07roundup * 10#950/startup_hook in readline: [resolved] I hope fixed * 14https://bugs.pypy.org/issue95001:52
lahwranwell, drat01:52
lahwranguess you have to pay for all those layers of abstraction with a few cpu cycles, eh? :p01:52
fijallahwran: if you don't use threads, GIL has nothing to do01:52
fijallahwran: we don't even *have* GIL when there are no threads01:52
fijalit's a crappy mess if you ask me and that's what you pay for01:53
fijalcrappy complicated mess01:53
fijalJIT is not very good at decoupling mess01:53
lahwranI mean, I was considering doing all sqlalchemy stuff in other threads01:53
fijalok01:53
lahwranin which case, cpython GIL will get in the way on multiple cores01:53
lahwranI'm only trying to evade waiting for IO anyway, so on a single core, cpython threading will do01:53
fijalno, you're messing stuff up01:54
lahwranwell, drat01:54
fijalyou can never use python on multiple cores01:54
lahwranwait, what?01:54
fijalthat's what the GIL does01:54
fijalit prevents you from running on more than one core01:54
lahwranso you're saying, if you have a multi-core system, go home, no python for you01:54
lahwranOH01:54
lahwranthat's not what I mean01:54
fijalno, you can use python01:54
fijalso how did you pull in multicore here?01:54
lahwranI mean, if you use threading at /all/ on a multi-core system, cpython will have contention issues that take loads of cpu01:55
fijalhow does having a multi-core system make any difference?01:55
lahwranlong story01:55
lahwranmoment01:55
fijalwell the answer is it does not :)01:55
fijalit has contention issues on one core as well IMO01:55
fijalin *some scenarios*01:55
fijalnotably when you do a lot of IO01:55
fijalanyway01:56
fijalI would not worry about it01:56
lahwranhttp://blip.tv/carlfk/mindblowing-python-gil-224337901:56
fijalif you use sqlalchemy you will not be doing much IO01:56
fijaland I'm going to bed01:57
fijallahwran: I think you're a bit prematurely worried01:57
fijaland besides, sqlalchemy is not any faster on PyPy than on Cpython I think01:57
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/69302:00
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/80402:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/87502:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/121502:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/69002:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/153402:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/34302:00
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/194702:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/35302:00
tmcm (~no6@shell.dhp.com) joined #pypy.02:04
gehho2 (~gehho2@ZJ040192.ppp.dion.ne.jp) left irc: Quit: Leaving...02:44
derdon_ (~derdon@p5DE8B2C1.dip.t-dialin.net) left irc: Remote host closed the connection02:45
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 245 seconds02:46
nedbat (~nedbat@python/psf/nedbat) joined #pypy.02:49
DasIch_ (~DasIch@p3E991232.dip.t-dialin.net) joined #pypy.02:57
DasIch (~DasIch@p3E991232.dip.t-dialin.net) left irc: Read error: Connection reset by peer02:58
Nick change: DasIch_ -> DasIch02:58
ericholscher (~eric@173.203.223.63) left irc: Quit: Peace.03:01
Eruquen (Eruquen@server3.raumopol.de) left irc: Ping timeout: 260 seconds03:02
ericholscher (~eric@173.203.223.63) joined #pypy.03:03
Shanita (~John@osbk-4db14e91.pool.mediaWays.net) joined #pypy.03:04
Kaskuka (~John@osbk-4db1576f.pool.mediaWays.net) left irc: Ping timeout: 276 seconds03:04
Eruquen (Eruquen@server3.raumopol.de) joined #pypy.03:08
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/22103:26
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/34303:29
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds03:43
Garen_ (garen@69.76.18.3) left irc: Remote host closed the connection04:12
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/69304:29
DasIch_ (~DasIch@p4FFDE9C4.dip.t-dialin.net) joined #pypy.05:10
DasIch (~DasIch@p3E991232.dip.t-dialin.net) left irc: Ping timeout: 245 seconds05:11
Nick change: DasIch_ -> DasIch05:11
buribu (~buribu@unaffiliated/buribu) left irc: Read error: Connection reset by peer05:25
bbot23Success: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/80405:30
nettok (~quassel@190.148.27.106) left irc: Remote host closed the connection06:14
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/153406:28
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/87506:29
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/69006:30
zain (~textual@46.120.28.30) joined #pypy.06:33
Nick change: ixokai_ -> ixokai06:39
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/121506:41
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/35306:43
asmeurer_ (~asmeurer@dhcp-altamirano-166.resnet.nmt.edu) left irc: Quit: asmeurer_07:10
zain (~textual@46.120.28.30) left irc: Quit: Computer has gone to sleep.07:26
metvik (~metvik@unaffiliated/metvik) left irc: Ping timeout: 245 seconds07:28
berdario (~Tiibiidii@host142-83-dynamic.2-87-r.retail.telecomitalia.it) joined #pypy.07:33
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/194707:43
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1948 [py3k]07:43
zain (~textual@46.120.28.30) joined #pypy.08:06
zain (~textual@46.120.28.30) left irc: Quit: Computer has gone to sleep.08:19
dmitrey (~quassel@77.121.195.200) joined #pypy.08:20
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) left irc: Read error: Connection reset by peer08:22
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.08:33
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 244 seconds08:40
thinker (~thinker@123-194-52-90.dynamic.kbronet.com.tw) joined #pypy.08:44
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1948 [py3k]08:47
zain (~textual@bzq-79-182-196-50.red.bezeqint.net) joined #pypy.08:58
AlekSi (~AlekSi@176.57.74.133) joined #pypy.09:05
xcombelle (~xcombelle@AToulouse-551-1-67-164.w92-146.abo.wanadoo.fr) joined #pypy.09:09
asmeurer__ (~asmeurer@dhcp-altamirano-166.resnet.nmt.edu) joined #pypy.09:18
arigato (~arigo@xdsl-188-154-145-222.adslplus.ch) joined #pypy.09:19
xcombelle (~xcombelle@AToulouse-551-1-67-164.w92-146.abo.wanadoo.fr) left irc: Ping timeout: 260 seconds09:19
arigato...ok, we can't push to bitbucket because someone is holding a lock?09:25
arigato(hi)09:25
fijalhi09:29
fijaleh09:29
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.09:29
fijalhow do I debug a phone connecting to wifi that has a button "connect" adn that's pretty much it09:29
arigatoit's great when it just works out of the box09:31
xcombelle (~xcombelle@AToulouse-551-1-156-133.w109-214.abo.wanadoo.fr) joined #pypy.09:34
asmeurer__ (~asmeurer@dhcp-altamirano-166.resnet.nmt.edu) left irc: Quit: asmeurer__09:35
fijalnot so great when it does not09:46
fijalI'm afraid linux is going into the same direction09:47
fijalbtw09:47
fijalarigato: did you finish yday the stackroot thngie?09:47
fijaldoes not seem so?09:48
arigatono09:49
arigatoI did one additional fix09:49
arigatothen translated a pypy-c09:49
arigatothen the pypy-c instantly segfaults09:49
arigatothat's when I decided "tomorrow"09:50
fijalhaha09:50
fijalok09:51
fijalI missed that I only have "screen has terminated"09:51
fijalI suppose that counts as 'not working'09:51
arigato:-/09:53
arigatojust started a new screen, but busy booking for pycon right now09:53
fijalok09:53
fijalI'm busy upgrading my router's firmware09:54
fijalso if I disappear you know what happened :)09:54
fijalyou have 2 screens btw09:54
berdario (~Tiibiidii@host142-83-dynamic.2-87-r.retail.telecomitalia.it) left irc: Quit: Ex-Chat09:55
berdario (~Tiibiidii@host142-83-dynamic.2-87-r.retail.telecomitalia.it) joined #pypy.09:55
arigatotrue09:55
fijalI'll be probably off most of today09:56
fijalbut let me know when you'e done with the travelling salesman problem09:56
fijalhttp://armisael.silix.org/2012/01/speeding-up-json-decoder-in-pypy/09:57
fijalbtw, people putting jitviewer to good use09:57
Nick change: `fox` -> firefox09:59
Nick change: firefox -> volpino10:01
Nick change: volpino -> LadyViolet10:01
fijal_ (~fijal@196.215.151.81) joined #pypy.10:12
xcombelle (~xcombelle@AToulouse-551-1-156-133.w109-214.abo.wanadoo.fr) left irc: Ping timeout: 276 seconds10:13
fijal (~fijal@196.215.151.81) left irc: Ping timeout: 276 seconds10:14
antocuni (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) joined #pypy.10:18
Nick change: fijal_ -> fijal10:22
xcombelle (~xcombelle@AToulouse-551-1-88-3.w92-149.abo.wanadoo.fr) joined #pypy.10:26
dmitrey (~quassel@77.121.195.200) left irc: Ping timeout: 252 seconds10:41
antocunibah11:02
antocuniour copy of pyrepl and the official one at pypy/pyrepl diverged :-(11:02
CIA-603ypekka 07roundup * 10#1028/Putting numpypy arrays to (multiprocessing) queue fails: 11:03
CIA-6[new] Queueing numpypy arrays fails within pickle.py. Attached code works with11:03
CIA-6CPython/multiprocessing/numpy combination but breaks the PyPy/Num ... * 14https://bugs.pypy.org/issue102811:03
fijalantocuni: I guess it's my fault sorry11:05
fijalwhy would you ever put numpy in multiprocessing queues....11:05
antocunifijal: not only yours, there are 4-5 checkins which are not in the official repo11:05
mikefcit fails because numpypy arrays don't pickle11:05
mikefcI was looking at this today for one of my projects.11:06
antocuniI'm thinking about adding a test which checks that our copy of pyrepl won't diverge11:06
fijalmikefc: adding pickle support for numpy arrays seems simple enough IMO11:06
mikefcat the very least i think tostring and fromstring would be needed before I'd attempt a __reduce__ for pickle support.11:06
fijalwe have fromstring I think11:07
antocuniwow, "only" 1146 failing tests in the py3k branch11:08
mikefcyup. need a tostring as well. Not sure there's any chance we can be compatible with numpy pickled arrays though.11:08
antocuni(they were 17346 five days ago)11:08
mikefcit took me quite a while to work a dummy __reduce__ method into applevel numpy. 11:09
mikefcfijal: I don't know how you keep everything in pypy in your head.11:09
fijalmikefc: easy, spend 5 years with pypy and forget about other stuff11:10
mikefcheh. good to hear you have some $ support now.11:10
fijalarmin knows more than I do btw11:10
antocuniah no, they were 1621 of course, 17345 was the number of passing tests :-)11:11
csenger (~user@dslb-084-058-156-080.pools.arcor-ip.net) joined #pypy.11:14
CIA-603Garen 07roundup * 10#934/subprocess hangs in 1.7 (but not in 1.6 or 1.5): 11:15
CIA-6In 1.7 and most recent win32 nightly (that built, from 1/22) pypy throws the11:15
CIA-6following error pretty quickly:11:15
CIA-6Traceback (most recent call last ... * 14https://bugs.pypy.org/issue93411:15
dmitrey (~quassel@77.121.195.200) joined #pypy.11:17
derdon (~derdon@p5DE8B2C1.dip.t-dialin.net) joined #pypy.11:20
fijalcsenger: hey11:20
csengerhi fijal 11:20
xcombelle (~xcombelle@AToulouse-551-1-88-3.w92-149.abo.wanadoo.fr) left irc: Ping timeout: 252 seconds11:20
fijalI looked at your changes to benchmark runner11:21
csengerThere's much going on in pypy :)11:21
fijalthey all make sense IMO and make runner.py less adhoc11:21
fijalas always :)11:21
Nick change: LadyViolet -> `fox`11:21
dmitrey (~quassel@77.121.195.200) left irc: Remote host closed the connection11:21
csengerfijal: should I merge in default and issue pull requests for both packages?11:22
fijalthat was my question - did you update then buildbot to work with new options?11:22
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds11:22
csengerYes, I update both factories, jit and cpython.11:23
fijalok11:23
csengerI just ran the ai test, but it produces the same upload date for jit.11:23
csengers/date/data11:23
fijalI'll look into merging that11:24
fijalprobably not today though11:24
fijalI'm technically "off"11:24
csengerOkay. I'd look into python3 benchmarks in the next days.11:24
fijalI personally don't care about that ;-)11:25
fijalbut makes sense for python-dev11:25
fijalanother issue - I deployed speed on ep.io11:25
csengerThought so :)11:25
fijalshould we use that instead?11:25
fijalI don't know if having jython would not be a better option btw11:26
csengerI've no real oppinion about that beside that the maschine is running with 24 cores idle, so no need to do anything ;)11:26
csengerA build to run benchmarks on jython11:27
csengerWe could reach out to the jython community to get that running.11:27
AlekSiHello. Is https://github.com/pypy/pypy an official mirror?11:27
csengerfijal: I'll ask on the jpython mailing list if they are interested. 11:29
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) left irc: Remote host closed the connection11:29
fijalAlekSi: no11:29
fijalcsenger: well, the 24 cores is seriously not usable for benchmarks :/11:30
fijalyou run into cache contention11:30
fijalI think it's 2-processors so maybe you can run 2 benchmarks independently, but that's it11:30
fijalyou can endless amount of pypy translations11:30
fijalwe can try to have it run 4 times a day though11:30
AlekSifijal: is it managed by someone from PyPy team then, or by complete outsider?11:30
fijalAlekSi: by a complete outsider11:30
fijalI think someone might be using it, but it comes with no guarantee11:31
fijalit should be a verbatim copy11:31
fijal(should)11:31
AlekSifijal: how can I contact him/her?11:31
csengerAh, I thought the benchmark process is bound to 1 core, not 1 processor.11:31
fijalI have no idea :)11:31
fijaltav and ronny did something there11:31
fijalboth can be found on IRC11:31
fijalcsenger: but you have shared cache11:31
fijalso even on 2 independent cores, you still have them interfering11:31
fijalcsenger: do I make any sense?11:34
csengerI don't know much about processor achitecture and how that influences benchmarks, so I think yes.11:35
xcombelle (~xcombelle@AToulouse-551-1-88-3.w92-149.abo.wanadoo.fr) joined #pypy.11:35
AlekSifijal: thanks, mailed him11:37
csengerThen it might make sense to move speed.python.org somewhere else at some point.11:37
fijalAlekSi: what are you trying to achieve?11:37
fijalcsenger: I was thinking about ep.io11:37
fijalsince it's free :)11:37
csengerBut it also means that pypy benchmark and cpython benchmark cannot run in parralel.11:37
fijaljesse said something there will be a dedicated host11:37
fijalno, they totally cannot11:37
fijalthey both should have an exclusive lock11:38
AlekSifijal: I'm interested in tags for releases in git for some automation.11:38
csengerThey do not share the lock. Will that work?11:38
fijalAlekSi: for some automation of what?11:38
fijalAlekSi: and why not use hg?11:38
fijalcsenger: do tehy both hold the lock?11:38
AlekSifijal: hg will require some changes in my machinery 11:40
csengerThere are two locks defined:11:40
csengerTannitCPU = locks.MasterLock('tannit_cpu', maxCount=6)11:40
csengerSpeedPythonCPU = locks.MasterLock('speed_python_cpu', maxCount=24)11:40
csenger 11:40
CIA-603mikefc 07roundup * 10#1028/Putting numpypy arrays to (multiprocessing) queue fails: 11:46
CIA-6[chatting] Not really a bug.11:46
CIA-6Numpypy arrays currently lack a __reduce__ method, so they can't be pickled. Stay11:46
CIA-6tuned. * 14https://bugs.pypy.org/issue102811:46
fijalAlekSi: what are you doing?11:48
fijalcsenger: so both cpython and pypy benchmarks should grab SpeedPythonCPU11:48
fijalI'll make some modifications because there is no point in running nojit for example11:48
csengerI think they should share one SlaveLock that they can exclusively get if both builds will run on the same slave.11:48
fijalyes, SpeedPythonCPU11:49
fijalshould be shared between them11:49
csengerIt can be a locks.SlaveLock.11:49
csengerThe MasterLock will block on the master, so the same build will not run on a second slave like tannit which is not necessary.11:50
fijalok11:50
fijalah11:50
fijalI know why it is the way it is11:50
Nick change: pedronis_ -> pedronis11:50
fijalbecause we have multiple slaves on tannit11:50
fijal(maybe?)11:51
fijalantocuni: do you remember?11:51
Action: fijal off11:51
pedroniscsenger: were bugs in slavelock fixed in buildbot, at some point they tend to make your slave unusable, like if you tried to use the ping functionality from the web ui etc11:52
ronnyfijal: i actually have nothing to do with the git mirror by tav11:53
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.11:53
fijalronny: ok11:53
ronnyfijal: but someone should change the descriptions on that repo11:54
csengerpedronis: dunno (even that that existed :)11:55
fijalronny: from what to what?11:55
fijalronny: you (or I) can jsut ask tav11:55
ronnyfijal: its not clear that its unofficial and wont get pulled from, there are some fork that people actuallt commit to11:56
fijalok11:56
fijalI'll mention that11:56
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.11:56
arigatobah, found it11:57
arigato(undodb-gdb is great sometimes)11:58
pedronisbuildbot tracker doesn't give anything positive or negative about that sort of stuff, I suppose it was in the past too easy to just say no to using locks11:59
fijalarigato: cool :)12:00
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.12:01
arigatowaa12:04
antocunifijal: I think I decided to use only masterlocks because slavelocks were not powerful enough and/or buggy12:09
antocunibut I don't remember the exact details12:09
nigelb (~nigel@ubuntu/member/nigelb) left irc: Remote host closed the connection12:17
Eventh (~Even@219.162.202.84.customer.cdi.no) joined #pypy.12:25
nedbat (~nedbat@python/psf/nedbat) joined #pypy.12:41
arigatofijal: the lock on the pypy repo doesn't go away12:45
arigatodo you know how to report it to the bitbucket team?12:46
antocuniarigato: lock?12:46
arigatocan't push anything12:46
arigatohttp://paste.pocoo.org/show/542663/  (takes 5 or 10 minutes to abort)12:47
antocuniarigato: pff12:47
antocuniarigato: support@bitbucket.org 12:48
arigatook12:48
Trundle (~andy@89.204.139.27) joined #pypy.12:48
Trundle (~andy@89.204.139.27) left irc: Changing host12:48
Trundle (~andy@python/site-packages/trundle) joined #pypy.12:48
arigatoI suspect it's some internal operation, because contegix.com is bitbucket's storage provider12:49
arigatobut it's been 2-3 hours now12:49
mattip (6dba09a2@gateway/web/freenode/ip.109.186.9.162) joined #pypy.12:50
mattiparigato: there was progress on the win32 build, but it still crashed.12:51
mattip(hi btw)12:51
arigatomattip: hi.  yes, I saw12:52
arigatoI have probably a fix12:52
arigatobut I cannot push it to bitbucket and run the buildslave12:52
mattipgood. Just checking if there is something I can do.12:52
antocuniarigato: you can push directly to the buildmaster repo12:52
arigatoah, the buildslaves don't pull from bitbucket?12:53
antocunisorry, to the buildslave repo12:53
antocuniyes, they pull12:53
antocunibut if you push your changes, then they will be there anyway12:53
arigatoah, yes12:53
antocuniah, unless the run is aborted because of the failed pull12:53
arigatono, pull works12:54
antocuniok12:54
arigatothanks for the idea, but I started a run anyway on bigboard in my account :-)12:54
antocuniok :-)12:54
Action: mattip nothing more to contribute. ttfn12:57
mattip (6dba09a2@gateway/web/freenode/ip.109.186.9.162) left irc: Quit: Page closed12:57
arigatofijal: pypy-c seems to work fine in the stackroot-speedup branch now13:02
antocuniany actual speedup?13:02
arigatonot on small examples13:02
xcombelle (~xcombelle@AToulouse-551-1-88-3.w92-149.abo.wanadoo.fr) left irc: Ping timeout: 255 seconds13:19
xcombelle (~xcombelle@AToulouse-551-1-88-3.w92-149.abo.wanadoo.fr) joined #pypy.13:20
berdario (~Tiibiidii@host142-83-dynamic.2-87-r.retail.telecomitalia.it) left irc: Ping timeout: 276 seconds13:24
berdario (~Tiibiidii@host142-83-dynamic.2-87-r.retail.telecomitalia.it) joined #pypy.13:32
amaury_is bitbucket in read-only mode?13:39
arigatoamaury_: on pypy/pypy13:43
arigatobut on the other repos it works13:43
amaury_remote: waiting for lock on repository /data/netapp02/netapp02_sas01_21/data/d-64/r-143064 held by 'bitbucket12.managed.contegix.com:31242'13:43
amaury_this is the error I get13:44
arigatome too13:44
arigatoI already mailed support@bitbucket.org13:44
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy.13:59
dmitrey (~quassel@77.121.195.200) joined #pypy.14:00
metvik (~metvik@unaffiliated/metvik) joined #pypy.14:01
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection14:02
csenger (~user@dslb-084-058-156-080.pools.arcor-ip.net) left irc: Quit: ERC Version 5.3 (IRC client for Emacs)14:06
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.14:10
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer14:10
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 252 seconds14:11
squiddy (~squiddy@g224193165.adsl.alicedsl.de) joined #pypy.14:32
dmitrey (~quassel@77.121.195.200) left irc: Ping timeout: 252 seconds14:40
mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) left irc: Quit: ZNC - http://znc.sourceforge.net14:42
xorAxAx (~alexander@moinmoin/coreteam/alexander) left irc: Quit: Nobody expects the Spanish Inquisition!14:42
nickik2 (~nick@217.150.233.216) joined #pypy.15:05
arigatofijal: stackroot-speedup has a marginal positive benefit on "pypy translate.py"15:07
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.15:13
dmitrey (~quassel@77.121.195.200) joined #pypy.15:17
csenger (~user@dslb-084-058-156-080.pools.arcor-ip.net) joined #pypy.15:43
Nick change: tito_ -> tito15:46
tito (~tito@178.73.216.222) left irc: Changing host15:46
tito (~tito@pdpc/supporter/active/tito) joined #pypy.15:46
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.15:48
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Quit: Lost terminal15:49
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.15:49
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 248 seconds15:50
nettok_ (~quassel@190.148.27.106) joined #pypy.15:56
antocuni (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds16:01
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy.16:07
derdon_ (~derdon@pD9E1D9C1.dip.t-dialin.net) joined #pypy.16:12
nettok_ (~quassel@190.148.27.106) left irc: Remote host closed the connection16:13
derdon (~derdon@p5DE8B2C1.dip.t-dialin.net) left irc: Ping timeout: 248 seconds16:15
__main__ (~main@adsl-99-173-15-158.dsl.pltn13.sbcglobal.net) left irc: Read error: Connection reset by peer16:25
nedbat (~nedbat@python/psf/nedbat) joined #pypy.16:29
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds16:35
voidspace (~voidspace@87-194-212-65.bethere.co.uk) joined #pypy.16:37
voidspace (~voidspace@87-194-212-65.bethere.co.uk) left irc: Changing host16:37
voidspace (~voidspace@python/psf/voidspace) joined #pypy.16:37
ssbr (~ssbr@python/site-packages/ssbr) joined #pypy.16:38
lucian<ssbr> pypy: error while loading shared libraries: libssl.so.0.9.8: cannot open shared object file: No such file or directory16:38
luciani get the same16:38
ssbrbedankt lucian 16:38
luciani'm on ubuntu, ssb is on fedora16:39
rguillebertyou don't have the same version of libssl as the buildbot16:40
arigatoit's only a warning, ignore it.  it's gone from the trunk version16:40
arigatoah no sorry16:41
rguilleberton ubuntu you can install libssl0.9.816:41
arigatoyes16:41
lucianso I can. thanks16:42
arigatoyou may need to install libssl0.9.8 from sources on some systems, but it's actually very easy16:42
arigatogood16:42
ssbrNo love for fedora :(16:42
arigato(I did it in a debian chroot)16:43
luciancool, works16:43
luciandoes the license allow you to ship it?16:43
rguillebertit's the apache licence so probably16:44
rguillebertbut I don't think it should be16:44
lucianrguillebert: why? few systems appear to have it16:44
Action: ssbr is building from source16:45
rguillebertbut it's not really the unix way16:45
ssbrrguillebert: neither is a monolithic JIT compiler framework16:46
lucianneither are a lot of things :)16:46
ssbr(LLVM is a bit more unixy as far as such things go)16:46
amaury_libssl is subject to security updates16:46
lucianamaury_: good point16:47
rguillebertarguably, your distribution is supposed to make pypy work on it not the other way around16:47
ssbrSo wait, where do I get this libssl.so.0.9.8 from?16:48
ssbropenssl 0.9.8 doesn't seem to put it anywhere with the default options, nor is it mentioned in the readme16:49
lahwranis this the optimal place to ask questions about py.test?16:50
lahwranif so, I'm trying to figure out how to best use it with coverage16:50
exarkunssbr: You probably didn't build the .so, then.16:50
ssbrexarkun: Nope, probably not.16:50
lahwrancurrently looking at http://pypi.python.org/pypi/pytest-cov/0.14 which looks pretty good16:50
nedbatlahwran: pypy is not responsible for py.test16:51
lahwranah, okay16:51
ssbrexarkun: I am asking about it and how to get it (via building from source, or not).16:51
rguillebertssbr, you can probably get the package from a previous fedora version16:51
exarkunssbr: Maybe #pypy isn't the best place to ask about OpenSSL build issues?16:51
Action: lahwran dances on over to #python16:51
lucianlahwran: try #pylib16:52
lahwranah, perfect, thanks16:52
ssbrexarkun: I'll concede that.16:52
arigatossbr: here's what I did:16:55
arigatohttp://www.openssl.org/source/16:56
arigatopicked the latest 0.9.816:56
arigato(seems to be  openssl-0.9.8t.tar.gz  right now)16:56
arigatothen build and install from sources16:56
arigatosomething like "configure; make; make install" works16:57
arigato"./config; make; make install"16:57
ssbrarigato: that's what I did, yeah :S16:58
ssbrThere's also a 0.9.8s , which I am trying16:58
ssbrI don't know the difference. I suspected version letters, but...16:58
ssbrYeah, no dice there either.17:00
arigatodoesn't work? how?17:03
ssbrno .so files .17:04
ssbrgoogling says I should ./config shared17:04
ssbrwell, that makes sense, doesn't it. :S17:04
arigatoindeed, I see "./config shared" in my history17:07
chronitis (~chronitis@88-104-120-175.dynamic.dsl.as9105.com) joined #pypy.17:07
ssbrWelp, that one's solved. Now on to "libbz2.so.1.0"17:09
ssbrOK, that one solved by symlinking (hopefully won't blow up on me). It still doesn't like SSL ("no version information"), but I think I'm good to go :)17:12
ssbrthanks for your help, arigato, lucian17:12
arigato:-)17:12
ssbrwooo17:15
amaury_some day pypy will ship in several modules17:15
amaury_so openssl will be required only when doing "import _ssl"17:16
ssbrPyPy's C-compiled, JIT-compiling regexp engine runs the test suite in 0.3 seconds. My JIT-compiled, interpreting regexp engine runs it in 2.4 :D17:16
ssbr(within an order of magnitude!)17:16
kenaan12arigo pypy.org[extradoc] 116cb63133ae6d 15/: Mention openssl 0.9.8.17:16
Alex_Gaynoramaury_: FWIW David Beazley expressed some interest at looking at the seperate compilation issue17:17
amaury_aha17:17
Alex_Gaynoramaury_: I warned him it was a hard problem (tm)17:19
lucianamaury_: that would be nice, but i don't think it's that big a deal17:19
antocuni (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) joined #pypy.17:22
cwillu (~cwillu@cwillu.com) left irc: Ping timeout: 252 seconds17:28
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.17:30
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Remote host closed the connection17:48
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.17:50
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection17:50
cwillu (~cwillu@cwillu.com) joined #pypy.17:56
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 276 seconds17:56
antocuni (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds17:56
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer17:58
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.17:58
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer18:01
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.18:03
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Ping timeout: 276 seconds18:08
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.18:09
kenaan12arigo extradoc 1194ccf2eee106 15/planning/stm.txt: fix18:14
Action: arigato wrote stm.txt but is too scared to start seriously working on that18:14
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 252 seconds18:15
kenaan12alex_gaynor default 11713d8bc8fbc7 15/pypy/module/micronumpy/tool/numready/main.py: Collect data on flatiter as well.18:17
ssbrI see lots of tutorials on making an interpreter using PyPy. This is great! I remember they didn't exist once. But I can't see anything on making an interpreter in PyPy that's accessible from Python (e.g. regexp bytecode interpreter)18:17
ssbrI'm not really sure what's involved here. mixed modules?18:18
xcombelle (~xcombelle@AToulouse-551-1-88-3.w92-149.abo.wanadoo.fr) left irc: Ping timeout: 240 seconds18:19
kenaan12amauryfa string-NUL 1128448014baa2 15/pypy/: Disallow nul bytes in more functions18:20
kenaan12amauryfa string-NUL 11b591a6e4e204 15/pypy/: Fix remaining test failures18:20
kenaan12amauryfa merge-2.7.2 11795bf94e952a 15/lib-python/modified-2.7/test/test_str.py: Add a copy of test_str.py, to be modified18:20
kenaan12amauryfa merge-2.7.2 11dc65c138fd25 15/lib-python/modified-2.7/test/: Skip check on the error message: pypy's multimethods are quite different18:20
kenaan12amauryfa merge-2.7.2 11674ef5a78812 15/lib-python/modified-2.7/test/test_socket.py: tets_socket: Loosen comparison of error messages.18:20
xcombelle (~xcombelle@AToulouse-551-1-88-3.w92-149.abo.wanadoo.fr) joined #pypy.18:21
amaury_ssbr: no, "mixed modules" are only for the pypy interpreter18:21
ssbramaury_: Hm. That's how _sre does it AFAICT. How would someone else do it?18:22
ssbrOf course I could also easily be wrong. I don't know very much about this at all. Just that all these blog posts and so on tell you how to translate your interpreter, but not compile it as a usable module from pypy18:22
Nick change: htoothrot -> toothr18:24
stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) joined #pypy.18:30
Action: fijal back18:34
fijalarigato: still around?18:34
arigatossbr: do you mean, how to write some interpreter (like regexps) integrated with pypy and accessible for the end user by saying "import yourmodule"?18:35
arigatofijal: yes18:35
ssbrarigato: yes18:35
arigatossbr: then yes, you need to read about "mixed modules" for the pypy interpreter18:35
arigatoe.g. regexps are implemented in pypy.rlib.rsre, and integrated in pypy.module._sre18:36
arigatothe "mixed module" part is in pypy.module._sre18:36
fijalarigato: did you try my example?18:36
arigatofijal: no, I was about to ask you18:36
ssbrarigato: Is there enough stuff out there to learn by reading?18:36
fijalarigato: if you have the built pypy-c on tannit I can do that18:36
arigatofijal: yes18:36
arigatothank you18:36
fijalwhere?18:36
arigatoit's ~arigo/br/stackroot-speedup/64compiled/18:37
arigatopypy-c-r51949-jit-shadowstack18:37
fijalok18:37
fijaland some other shadowstack?18:37
arigatoyes18:37
fijalheh18:37
fijalwhere?18:37
arigato~arigo/pypysrc/64compiled/pypy-c-r51917-jit-shadowstack18:37
arigatossbr: let me find again18:38
ssbrSure. And thanks18:39
arigatossbr: http://pypy.readthedocs.org/en/latest/coding-guide.html#implementing-a-mixed-interpreter-application-level-module18:40
arigatonote that you'll have to worry about various aspects of "wrapping"18:41
arigatoe.g. receive wrapped strings from app-level, unwrap them and pass them to your own interpreter18:41
ssbrAlright. Well, that won't be so bad, since I actually am just doing a regexp interpreter, so there's only a handful of things. :)18:42
ssbrI don't really know how this'll go, but I'll be back if I have any questions18:42
arigatosure18:42
ssbrthanks again! :)18:43
arigatolook at example of code, too, pypy/module/*/*.py18:44
arigatoexamples18:44
ssbrWill do! I usually have trouble with reading the pypy codebase, though. I'm not very good at reading foreign code to begin with, and pypy has its own style that I don't know yet. Also, no comments. :<18:44
ssbr(or docstrings)18:45
ssbrIf it isn't too complicated I should be able to get it, though, especially since I'll actually be doing it rather than just reading. Very different thing, that.18:46
fijalarigato: I can consistently get 1.4s vs 1.5s18:47
fijalso it's "measurable"18:47
fijalnothing impressive18:47
ssbranyway, I'm off. It's bedtime for me. (Goodnight!)18:47
fijalI also think it was worse for asmgcc, could it potentially be?18:47
arigatomaybe yes, because asmgcc has a slower stack scanning18:48
fijaleven for the case where everything survives, so there is a lot more to do in nursery collect I can measure ~3% speedup18:49
fijalmerge I guess?18:49
fijalor should we do asmgcc?18:49
fijalor do we wait for your scanning?18:50
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: 18:51
ssbr (~ssbr@python/site-packages/ssbr) left irc: Ping timeout: 252 seconds18:52
fijalarigato: ?18:54
stakkars_ (~tismer@p5DDB5C07.dip.t-dialin.net) left irc: Quit: schnarch18:54
arigatoah!18:55
arigatohg push works again18:55
Alex_Gaynorfijal: is the branch a 6.6% win on just GC, or on a whole script?18:55
fijalAlex_Gaynor: on the whole script (a convoluted example a bit though)18:56
fijalwell not that much18:56
Alex_Gaynornot that much convulated?18:56
fijalisn't the word convoluted?18:56
kenaan12arigo default 11161714d53cc7 15/pypy/jit/codewriter/: Crash earlier and more explicitly if the jit sees an unknown operation (like cast_ptr_to_adr, as win32 builds see n...18:56
Alex_Gaynoryes, I just can't type18:56
fijalwell18:56
kenaan12arigo stackroot-speedup 11d7ae2df81896 15/pypy/rpython/memory/gctransform/shadowstack.py: Fix.18:56
kenaan12arigo default 11828709b0e0e9 15/pypy/rpython/lltypesystem/rffi.py: Don't look inside these functions.  (Looking in str_from_buffer() causes anyway the jit to crash on cast_ptr_to_adr.)18:56
kenaan12arigo stackroot-speedup 11fb3badd51945 15/pypy/translator/c/test/test_newgc.py: a failing test18:57
kenaan12arigo stackroot-speedup 11c4b086bfb967 15/pypy/: Fix.  "livevars" is the list returned from the previous push_roots, so it already contains one extra item...18:57
kenaan12arigo default 11c0a95b1a7b87 15/pypy/module/micronumpy/tool/numready/main.py: merge heads18:57
fijalit's deep call stack + lots of nursery collects18:57
arigato</spam>18:57
fijalarigato: so, thoughts?18:57
arigatofijal: we can try to adapt asmgcc18:57
fijalAlex_Gaynor: probably a bigger win for asmgcc18:58
Alex_Gaynorfijal: cool18:58
fijalarigato: the question is how much do we believe in your hacks?18:58
fijalbecause if miniscan works then maybe winning few % on asmgcc is not that interesting18:58
arigatotrue18:58
arigatobut my "hack" is likely quite more work18:58
fijalok18:59
Alex_Gaynorminiscan is the conservative scan for roots?18:59
fijalI guess adapting asmgcc is not that hard18:59
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.19:01
arigatoAlex_Gaynor: yes19:01
arigatofijal: I suppose19:01
fijalarigato: no clue :)19:02
fijalarigato: I think my stance is that I probably won't hack on asmgcc right now and you're of course free to do whatever you please19:04
arigatook :-)19:04
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.19:05
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 245 seconds19:06
fprimex (~fprimex@fl-67-233-144-239.dhcp.embarqhsd.net) joined #pypy.19:13
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.19:15
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer19:19
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.19:23
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Ping timeout: 248 seconds19:33
chronitisdo duplicate rffi.llexternal declarations of functions matter or is it automatically de-duplicated at some point in the build?19:35
chronitis(eg, "fcntl" is defined at least in rlib/_rsocket_rffi and module/fcntl with identical signature)19:36
arigatoit's just a declaration.  in both cases it ends up as just a call to the C function, in the generated C code19:36
chronitisso, not a problem them?19:37
arigatoexactly19:37
chronitisthanks19:37
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 276 seconds19:42
Husio (husiatyn@oceanic.wsisiz.edu.pl) left irc: Remote host closed the connection19:43
dmitrey (~quassel@77.121.195.200) left irc: Remote host closed the connection19:46
fijalarigato: so do you have any thoughts?19:46
arigato (~arigo@xdsl-188-154-145-222.adslplus.ch) left irc: Ping timeout: 252 seconds19:47
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.19:47
AlekSi (~AlekSi@176.57.74.133) left irc: Remote host closed the connection19:51
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.20:05
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer20:05
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer20:11
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.20:11
jacob22_ (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.20:12
jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Ping timeout: 276 seconds20:12
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer20:15
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.20:16
dmitrey (~quassel@77.121.195.200) joined #pypy.20:18
csenger (~user@dslb-084-058-156-080.pools.arcor-ip.net) left irc: Remote host closed the connection20:18
fijalif I were to write a PEP to write a new GC in pypy I would never do it20:22
Alex_Gaynorfijal: if we had to write a PEP every time we had a new dict type we're run out of 3 digit numbers20:23
timotimowould it be a PPEP?20:25
fijaltimotimo: as soon as it is a PPEP I will go and do something else20:25
fijaland armin and alex20:25
arigato (~arigo@xdsl-188-154-145-222.adslplus.ch) joined #pypy.20:28
fijalmeh20:29
fijalarigato: so do you have any thoughts?20:29
fijalabotu how to proceed with stack scanning?20:29
fprimex (~fprimex@fl-67-233-144-239.dhcp.embarqhsd.net) left irc: Quit: Leaving20:32
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: Ping timeout: 245 seconds20:33
arigatofijal: not really20:33
fijalhow about we merge this branch and wait for the miniscan to be done?20:34
csenger (~user@dslb-084-058-156-080.pools.arcor-ip.net) joined #pypy.20:34
arigatook20:34
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Ping timeout: 260 seconds20:35
`fox` (~fox@host155-107-dynamic.14-87-r.retail.telecomitalia.it) joined #pypy.20:36
Action: arigato runs a few more tests in the branch20:40
fijalI don't know :/20:40
fijalmaybe we can track asmgcc20:40
fijalbut does seem like it's easier to wait :)20:41
arigatoit's annoying :-/20:41
fijalarigato: maybe I can help you with miniscan?20:41
arigatoyesm if you want20:41
fijalseems like the best way to edit a pdf is print it and scan it20:41
fijalor maybe I can just do a .png edit?20:41
arigatoactually if you want you can pick whatever topic I discussed in extradoc/planning/stm.txt and say "let's do that first"20:42
arigatoor even discuss if the topic makes sense or no20:42
fijalok20:42
Action: fijal picks20:42
fijalmultiple nurseries sounds like something generally useful20:43
kenaan12arigo stackroot-speedup 11231f3159915e 15/pypy/rpython/memory/gctransform/shadowstack.py: Fix comment.20:46
fijalas in - sounds like a reasonable thing to implement first that's relatively independent20:49
arigatowell, it's a point pointless with a GIL20:50
arigatomaybe not completely, but then we run into other problems20:50
arigatolike what if the program creates 100 threads --- do we have 100 nurseries of 6MB each?20:51
arigatoin stm.txt the nurseries have a much stronger role of being the thread-local container of *writable* objects20:52
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy.20:54
fijalI did not say it makes sense without stm :)20:58
nickik (~nick@217.150.233.216) left irc: Quit: Verlassend20:59
arigatoyou said "...sounds like something generally useful" so I thought you said exactly that, sorry21:04
arigatoanyway, /me is entering "evening mode" where he doesn't do anything useful21:05
fijalhehe21:06
fijalok21:06
fijalI don't think I'll do anything useful either21:06
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!21:06
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.21:07
oal (u4126@gateway/web/irccloud.com/x-akxqsxxqxqewcimp) left irc: Max SendQ exceeded21:08
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy.21:08
dmitrey (~quassel@77.121.195.200) left irc: Remote host closed the connection21:12
fenrrir (~rodrigo@189.70.242.144) joined #pypy.21:16
oal (u4126@gateway/web/irccloud.com/x-apopkvkvwfnvinym) joined #pypy.21:18
xcombelle (~xcombelle@AToulouse-551-1-88-3.w92-149.abo.wanadoo.fr) left irc: Quit: I am a manual virus, please copy me to your quit message.21:18
`fox` (~fox@host155-107-dynamic.14-87-r.retail.telecomitalia.it) left irc: Ping timeout: 272 seconds21:18
Eventh (Even@219.162.202.84.customer.cdi.no) left #pypy.21:23
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.21:26
xorAxAx (~alexander@ampleforth.srv.alexanderweb.de) joined #pypy.21:31
xorAxAx (~alexander@ampleforth.srv.alexanderweb.de) left irc: Changing host21:31
xorAxAx (~alexander@moinmoin/coreteam/alexander) joined #pypy.21:31
squiddy (~squiddy@g224193165.adsl.alicedsl.de) left irc: Quit: Leaving21:32
mattip (~mattip@109.67.106.238) joined #pypy.21:34
mattipI have started a branch to clean up the win32 build21:39
mattipany objections?21:39
Alex_Gaynornope21:40
Eventh (~Even@219.162.202.84.customer.cdi.no) joined #pypy.21:41
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 260 seconds21:44
Husio (husio@darkserver.it) joined #pypy.21:44
stakkars_ (~tismer@88.130.173.175) joined #pypy.21:46
mattip:)21:48
`fox` (~fox@host75-252-dynamic.50-79-r.retail.telecomitalia.it) joined #pypy.21:52
Taggnostr (~quassel@dyn57-362.yok.fi) left irc: Ping timeout: 255 seconds21:53
kenaan12amauryfa merge-2.7.2 117a82f3fbab26 15/pypy/: Add large file support to mmap: 64bit offsets on 32bit platform.  problem: test_large_offset() fails when in...21:56
amaury_could someone review 7a82f3fbab26 I just pushed?21:56
amaury_I don't understand why the test fails21:57
Eventh (Even@219.162.202.84.customer.cdi.no) left #pypy.21:57
amaury_and passes when run with ./pypy-c and the -A option21:58
`fox` (~fox@host75-252-dynamic.50-79-r.retail.telecomitalia.it) left irc: Ping timeout: 255 seconds22:00
45PAAP7UK (~asmeurer@dhcp-altamirano-166.resnet.nmt.edu) joined #pypy.22:01
Nick change: 45PAAP7UK -> asmeurer__22:03
mattip (~mattip@109.67.106.238) left irc: Read error: Connection timed out22:03
take_cheeze (~Adium@pc06224.anc-tv.ne.jp) left irc: Ping timeout: 240 seconds22:03
mattip (~mattip@109.67.106.238) joined #pypy.22:04
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.22:04
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1216 [12amaury, merge-2.7.2]22:10
arigatoamaury_: I can look22:11
amaury_thanks22:11
amaury_in 32bit mod22:11
amaury_I was working on tannit22:12
amaury_note that I had to relax a limit in ll2ctypes.py22:13
stakkars_ (~tismer@88.130.173.175) left irc: Quit: schnarch22:14
mitsuhiko (~mitsuhiko@ubuntu/member/mitsuhiko) joined #pypy.22:15
arigatoyes, I imagine, but that's not the cause here22:17
mattipit passes for me, and doesn't even take too long: http://paste.pocoo.org/show/54274122:17
arigatoif I write a 'B' character at offset 0x4FFFFFFF22:17
mattipcentos 64 bit running 32 bit python22:17
arigatothen I get it22:17
arigatoso it's an issue like calling the wrong version of the mmap function via ctypes22:18
arigatoI fear there are two C definitions and some #defines to select the correct one22:18
arigatodepending on the size of off_t22:19
arigatoyes22:19
arigatowith this change it passes:22:20
arigatohttp://paste.pocoo.org/show/542744/22:20
arigatomattip: that might be a linux-only issue22:20
mat^2 (~mathias@212.130.113.35) joined #pypy.22:21
arigatoI'm unsure how to fix it22:22
arigatomaybe you can try to add a _callable=mmap_for_tests22:24
arigatowith a function mmap_for_tests that just uses mmap.mmap()22:24
gehho2 (~gehho2@ZJ040192.ppp.dion.ne.jp) joined #pypy.22:24
arigatoor maybe just use the 'mmap64' like I did in my paste, but only on Linux with an #XXX warning22:26
arigato(mmap64 is not really documented)22:26
Alex_Gaynorarigato: it comes up with a man mage for mmap2 if I man mmap6422:26
arigatoyes, same here22:27
arigatobut it's actually not mmap2 either22:27
arigatoit's really mmap with off_t being a 64-bit int22:27
amaury_so mmap is actually a #define to mmap64?22:29
arigatohttp://www.gnu.org/software/libc/manual/html_node/Memory_002dmapped-I_002fO.html22:29
arigatoyes, that's described in that page22:30
arigatoit also says the type is actually off64_t22:30
arigatoif you #define _FILE_OFFSET_BITS == 64, then both mmap64 and off64_t are available under the names mmap and off_t22:31
amaury_this was correctly detected by CConfig22:31
arigatoyes22:31
amaury_is mmap64 defined on all posix platforms?22:32
arigatoso either we fix the name of the llexternal to say 'mmap64' on linux, or we hack even more to automatically figure out in ll2ctypes that 'mmap' is #defined to 'mmap64'...22:32
arigatodunno22:32
arigatoit's in the libc, which is kind of standard nowadays, maybe22:32
amaury_we could use it when   rffi.sizeof(off_t) > rffi.sizeof(lltype.Signed)22:33
amaury_HAVE_LARGEFILE_SUPPORT22:33
antocuni (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) joined #pypy.22:35
arigatoyes22:35
arigatolet's try that, indeed22:35
amaury_or we could use macro=True22:36
amaury_in external()22:36
sunoano (~sa@unaffiliated/sunoano) left irc: Quit: Leaving.22:36
mikefc_ (9665bcc5@gateway/web/freenode/ip.150.101.188.197) joined #pypy.22:39
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!22:41
kenaan12mattip win32-cleanup 11882a62b6adfe 15/pypy/module/_ffi/test/test__ffi.py: fix some tests22:42
arigatouh, yes, I didn't know about 'macro'22:43
nedbat (~nedbat@python/psf/nedbat) joined #pypy.22:43
mattiplow lying fruit :)22:44
arigatomattip: :-)22:45
arigatoactually, shouldn't we more generally use 'macro=True' a bit everywhere just in case?22:45
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 252 seconds22:46
mikefc_ (9665bcc5@gateway/web/freenode/ip.150.101.188.197) left irc: Ping timeout: 245 seconds22:47
arigatoI see that it's already used throughout ropenssl for example22:47
kenaan12amauryfa py3k 11124b3b5cc7ac 15/pypy/module/_codecs/test/test_codecs.py: Fix syntax errors in test_codecs22:51
kenaan12amauryfa py3k 11a5bd7b8f5f80 15/: hg merge default22:51
amaury_yes, I probably added it for this module22:52
amaury_openssl has too many macros22:52
arigatothe test passes with macro=True22:53
amaury_cool22:53
arigatoshould I check in?22:54
amaury_please do22:55
Action: amaury_ -> sleep22:55
arigatogood night22:55
arkx (aku@eagleflow.fi) left #pypy.22:55
amaury_arigato: if you have time, please take a look at the string-NUL branch22:55
kenaan12arigo merge-2.7.2 11ff6d67b226da 15/pypy/rlib/rmmap.py: Fix for the new large file tests in pypy.module.mmap.22:56
arigatonot tonight --- but it's ready to be merged?22:56
amaury_it works, but translation can fail for obscure reason22:56
arigatook22:56
amaury_i.e when you call os.open() with a string not proven to have no NUL22:57
arigatoah22:57
amaury_I suggest that this check should be a translation option22:58
amaury_also, I found obscure bugs around SomeList and the TooLateForChange error22:59
amaury_like: s_x1.contains(s_x2) is True,22:59
amaury_but SomeList(s_x1).contains(SomeList(s_x2)) is False22:59
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/103923:00
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/79123:00
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/22223:00
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/5023:00
Action: amaury_ really off23:01
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.23:02
arigatothat's not really a bug23:02
arigatoif you call a function f(x) that has already been annotated in an earlier phase23:02
arigatoif x has been annotated as, say, a longlong, then you can pass a long23:03
arigatobut if x has been annotated as a list of longlongs, then you cannot pass a list of longs23:03
arigatobecause it's too late to reflow the function f23:03
mattipas long as the conversation is about C #defines: ASN1_ITEM_ptr is defined in openssl/asn1.h, but the definition changes according23:06
mattipto OPENSSL_EXPORT_VAR_AS_FUNCTION, which is true for win32.23:06
mattipthis is important to ssl_external('ASN1_ITEM_ptr', [rffi.VOIDP], ASN1_ITEM, macro=True) in pypy/rlib/ropenssl.py 23:08
mattipsince, for win32, it's not a VOIDP but a pointer to a function23:08
mattip(does such a thing exist in rffi?)23:08
mattipThe upshot is that _hashlib, and maybe many others, fail tests since efi does not compile for win32 in tests23:10
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1216 [12amaury, merge-2.7.2]23:10
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/1217 [12amaury, string-NUL]23:10
arigato (~arigo@xdsl-188-154-145-222.adslplus.ch) left irc: Ping timeout: 252 seconds23:11
amaury_mattip: so ASN1_ITEM_ptr takes a lltype.Ptr(ASN1_ITEM_EXP) as parameter?23:14
nickik2 (~nick@217.150.233.216) left irc: Quit: Verlassend23:15
stargaming (stargaming@server3.raumopol.de) left irc: Changing host23:15
stargaming (stargaming@unaffiliated/stargaming) joined #pypy.23:15
mattip (~mattip@109.67.106.238) left irc: Ping timeout: 252 seconds23:18
mattip (6d436aee@gateway/web/freenode/ip.109.67.106.238) joined #pypy.23:19
mattipargh, everything just crashed23:20
mattipan hour of pdb up-down gone!23:20
antocuni (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) left irc: Remote host closed the connection23:21
mattipamaury_: I'm not sure, I was just getting my head around all this and then it all came tumbling down.23:21
mattipit's something like:23:23
mattip#ifndef OPENSSL_EXPORT_VAR_AS_FUNCTION23:23
mattip#define ASN1_ITEM_ptr(iptr) (iptr())23:23
mattip(no that's backwards)23:24
mattip#ifndef OPENSSL_EXPORT_VAR_AS_FUNCTION23:24
mattip#define ASN1_ITEM_ptr(iptr) (iptr)23:24
mattip#else23:25
mattip#define ASN1_ITEM_ptr(iptr) (iptr())23:25
mattip#endif23:25
mattipthat was from the openssl/asn1.h23:25
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 272 seconds23:28
mattip (6d436aee@gateway/web/freenode/ip.109.67.106.238) left irc: Quit: Page closed23:33
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) joined #pypy.23:38
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy.23:48
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection23:50
Trundle (~andy@python/site-packages/trundle) left irc: Remote host closed the connection23:50
derdon_ (~derdon@pD9E1D9C1.dip.t-dialin.net) left irc: Remote host closed the connection23:57
--- Mon Jan 30 201200:00

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