#pypy IRC log for Saturday, 2011-11-26

zk (~is@whatit.is) joined #pypy.00:00
davisagli (~davisagli@davisagli.com) left irc: Ping timeout: 240 seconds00:02
davisagli (~davisagli@davisagli.com) joined #pypy.00:02
stakkars_ (~tismer@82.113.119.90) joined #pypy.00:16
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 248 seconds00:19
stakkars_ (~tismer@82.113.119.90) left irc: Quit: schnarch00:22
kkris (~kris@80-123-45-114.adsl.highway.telekom.at) left irc: Quit: Leaving.00:25
`fox` (~fox@79.30.109.192) left irc: Ping timeout: 252 seconds00:40
nedbat (~nedbat@python/psf/nedbat) joined #pypy.00:42
derdon (~derdon@p5DE8ADFD.dip.t-dialin.net) left irc: Remote host closed the connection00:45
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Ping timeout: 258 seconds00:55
cwillu_at_work (~cwillu@cwillu-1-pt.tunnel.tserv13.ash1.ipv6.he.net) left irc: Ping timeout: 244 seconds01:00
verte (~verte@python/site-packages/verte) joined #pypy.01:08
ixokai (~ixokai@pdpc/supporter/student/ixokai) joined #pypy.01:16
mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord01:39
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186401:40
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/59001:40
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/28301:40
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/62301:40
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/112401:40
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/80801:40
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72301:40
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/145001:40
canta (~canta@77-20-123-240-dynip.superkabel.de) left irc: Quit: Leaving.01:41
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/28301:42
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/72501:43
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/72601:43
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186401:44
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/112401:44
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/80801:44
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186501:44
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/59001:45
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/62301:45
ixokai (~ixokai@pdpc/supporter/student/ixokai) left irc: Quit: Bye!01:49
Nick change: Gulaway -> Gulopine01:54
JaredW (~jared@061093200075.ctinets.com) joined #pypy.01:57
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72302:06
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/145002:06
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72402:06
JaredW (~jared@061093200075.ctinets.com) left irc: Remote host closed the connection02:08
bbot213Exception: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/60 [12ctismer, win64-stage1]02:20
stakkarsannotation/test/test_annmm.py02:22
stakkarsfails for me on all windows versions (x86 default, x86 win64-stage1, x64 x64-stage1)02:23
Alex_Gaynorstakkars: armin and carl broke flow graphs, if that's the issue02:24
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186502:27
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186602:27
stakkarsAlex_Gaynor: probably is02:28
stakkarscd02:29
Alex_Gaynorstakkars: they only broke it today though, have you merged default in recently?02:30
stakkarsyes, I always do02:31
stakkarsshould I bachout the last commit?02:33
Alex_GaynorI only looked at it very briefly, since all the buildbots have already failed for tonight  I'm just planning on letting carl and armin look at it when they wake up :)02:34
Alex_Gaynoryou could always back it out on your branch though02:34
stakkarswas that c5d2f01c99ee   kill imports?02:37
Alex_Gaynor0bb2bc25fcfc and maybe b78a605fb45202:40
Moku (~John@osbk-4db06b9d.pool.mediaWays.net) joined #pypy.02:44
Cheshirc (~Kim@unaffiliated/cheshirc) joined #pypy.02:45
Nick change: Moku -> Guest899102:45
Shanita (~John@osbk-4d08724f.pool.mediaWays.net) left irc: Ping timeout: 248 seconds02:45
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72402:47
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72502:47
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186603:08
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186703:08
jaredw (~jared@061093200075.ctinets.com) joined #pypy.03:15
kenaan12ctismer win64-stage1 113b640415e101 15/pypy/jit/backend/llsupport/test/test_ffisupport.py: Added a check for is_emulated_long. inefficient: one change for one working test ;-)03:17
kenaan12ctismer win64-stage1 11fda9745e44a0 15/pypy/: Backed out changeset: b78a605fb45203:17
kenaan12ctismer win64-stage1 11d27c05319266 15/pypy/: Backed out changeset: 0bb2bc25fcfc03:17
bbot2Started: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/61 [12ctismer, win64-stage1]03:19
jaredw (~jared@061093200075.ctinets.com) left irc: Ping timeout: 248 seconds03:24
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72503:28
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72603:28
stakkarsrmmap: who has written that stuff without checking the platform?03:37
kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.03:42
verte (~verte@python/site-packages/verte) left irc: Quit: ~~~ Crash in JIT!03:46
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186703:49
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186803:49
Nick change: Gulopine -> Gulaway04:00
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 260 seconds04:02
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72604:08
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72704:08
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186804:28
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186904:28
Cheshirc (~Kim@unaffiliated/cheshirc) left irc: Quit: QuitMsg#163.4.d17.104:29
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/72604:43
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/72704:43
bbot213Exception: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/72704:49
jonathan__ (~jonathan@bas1-montreal29-1242341904.dsl.bell.ca) left irc: Ping timeout: 240 seconds04:52
mwhudson_ (~mwh@121-73-77-183.cable.telstraclear.net) joined #pypy.04:54
bbot213Exception: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/186905:11
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) joined #pypy.05:22
Taggnostr (~quassel@dyn57-215.yok.fi) left irc: Read error: No route to host05:41
Nick change: Gulaway -> Gulopine05:42
Taggnostr (~quassel@dyn57-215.yok.fi) joined #pypy.05:44
zk (~is@whatit.is) left irc: Quit: leaving06:11
qbproger_ (~qbproger@c-76-104-159-248.hsd1.wa.comcast.net) left irc: Ping timeout: 252 seconds06:23
bbot213Exception: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/61 [12ctismer, win64-stage1]06:35
fijal (~fijal@dsl-241-39-19.telkomadsl.co.za) joined #pypy.06:39
fijalhi06:43
Nick change: Gulopine -> Gulaway06:43
fijalbuildbot run out of mem, nice06:43
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) left irc: Quit: Computer has gone to sleep06:44
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 252 seconds06:52
fijal_ (~fijal@197.169.121.162) joined #pypy.06:55
fijal_bad armin, no cookie06:56
fijal (~fijal@dsl-241-39-19.telkomadsl.co.za) left irc: Disconnected by services06:56
Nick change: fijal_ -> fijal06:56
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.07:06
asmeurer (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) left irc: Remote host closed the connection07:10
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.07:19
asmeurer__ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) left irc: Ping timeout: 240 seconds07:26
fijal (~fijal@197.169.121.162) left irc: Ping timeout: 245 seconds07:33
y3di (~yedi@pool-71-162-74-55.bstnma.east.verizon.net) joined #pypy.07:48
bbot213Exception: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/72707:55
glyph (~glyph@glyph.im) joined #pypy.07:59
glyphhello #pypy08:00
glyphdid you know that your buildbots are in a terrible state?08:01
y3di (~yedi@pool-71-162-74-55.bstnma.east.verizon.net) left irc: Ping timeout: 260 seconds08:03
zk (~is@whatit.is) joined #pypy.08:05
whitelynx (~whitelynx@75.110.189.230) left irc: Ping timeout: 248 seconds08:16
mwhudson_ (~mwh@121-73-77-183.cable.telstraclear.net) left irc: Quit: Ex-Chat08:21
Alex_Gaynorglyph: yes, armin broke them08:22
Alex_Gaynor(and carl)08:22
glyphAlex_Gaynor: I'm not just talking about being broken08:23
glyphalthough boy-o are they broken08:23
Alex_Gaynorglyph: what else is wrong with them?08:24
glyphthey can't report results, the slaves are all down, it's not clear to a casual observer (well, to me, at least) what they're testing08:24
glyphlike, http://buildbot.pypy.org/summary isn't just some broken builds :)08:24
hakanardoalmos all tests in jit/metainterp fails on default for me right now...08:25
Alex_Gaynorhakanardo: armin/carl broke them08:25
hakanardowhich rev?08:25
Alex_Gaynorhakanardo: 0bb2bc25fcfc and maybe b78a605fb45208:26
hakanardothanx08:27
Alex_Gaynorglyph: pretty sure that's because so many tests are failing that trying to read a list is using al the memory08:27
whitelynx (~whitelynx@75.111.197.204) joined #pypy.08:28
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Quit: ZNC - http://znc.in08:45
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.08:51
Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Quit: Ex+re09:00
zk (~is@whatit.is) left irc: Quit: leaving09:14
zk (~is@whatit.is) joined #pypy.09:15
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) joined #pypy.09:29
derdon (~derdon@p5DE8A09D.dip.t-dialin.net) joined #pypy.09:31
arigato (~arigo@89.204.138.64) joined #pypy.09:40
kkris (~kris@93-82-45-16.adsl.highway.telekom.at) joined #pypy.09:51
CIA-6803arigo 07roundup * 10#940/Comparison with native application & cpython: 09:53
CIA-68[chatting] Actually, yes, it seems to be the case that it is not enough time for a complete09:53
CIA-68warm-up. If I run it on a larger collection of files ... * 14https://bugs.pypy.org/issue94009:53
Alex_Gaynorarigato: the is_w commit broke translation (and tests) rather spectacularly09:54
arigatoousp09:54
arigatodo you even know which tests broke?09:55
Alex_Gaynoranything that tries to annotate some code, just about09:55
arigatoah09:55
arigatoI see09:55
Action: arigato fixes09:55
fijal (~fijal@197.172.191.33) joined #pypy.10:03
derdon (~derdon@p5DE8A09D.dip.t-dialin.net) left irc: Remote host closed the connection10:04
kenaan12arigo default 117458773f3be0 15/pypy/objspace/flow/: Fix the flow object space for 0bb2bc25fcfc and b78a605fb452.  Oups.10:07
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) left irc: Quit: buildmaster reconfigured: bot disconnecting10:08
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) joined #pypy.10:13
fijalarigato: bad armin no cookie :)10:13
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) left irc: Client Quit10:13
arigatoyes10:13
fijalyou also broke the buildbot :)10:14
arigatoyes10:14
fijalbut I claim holger's java that has 3G of virtual ram hogged did not help10:14
arigatothat's not on wyvern10:14
melinath (~melinath@li341-41.members.linode.com) left irc: Ping timeout: 258 seconds10:14
Alex_Gaynorfijal: holger's JVM broke translation? ;)10:14
fijalah right10:14
fijalarigato: summary is eating ridiculous amounts of memory btw10:15
fijalmaybe we can do something with it10:15
fijallike run it on pypy :)10:15
Alex_Gaynorfijal: a few buildbot runs with passing tests should fix that10:15
fijalthere is also an option to just remove the offending runs10:15
apanda (~apanda@c-67-170-53-221.hsd1.wa.comcast.net) joined #pypy.10:17
apanda (apanda@c-67-170-53-221.hsd1.wa.comcast.net) left #pypy.10:18
bbot2 (~bbot2@wyvern.cs.uni-duesseldorf.de) joined #pypy.10:19
arigatoI'm removing  them now10:19
arigatobut indeed, it's consuming 1.4GB of RAM just to compute the summary<filter nothing> page10:21
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-win-x86-32/builds/541 [12fijal, release-1.7.x]10:22
glyph (glyph@glyph.im) left #pypy.10:22
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/273 [12fijal, release-1.7.x]10:22
bbot213Exception: 15http://buildbot.pypy.org/builders/pypy-c-app-level-win-x86-32/builds/541 [12fijal, release-1.7.x]10:22
bbot213Exception: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/273 [12fijal, release-1.7.x]10:22
arigato?10:22
fijalthose are my windows runs10:23
fijalschedule 17 years ago10:23
fijalI'm seriously still trying to build an executable on windws10:23
fijalbut we did not install the dr watson removal10:23
canta (~canta@77-20-123-240-dynip.superkabel.de) joined #pypy.10:23
Garenfijal: need > 2GB mem for 32-bit process, and --no-continuation10:23
fijalGaren: sorry, I'm missing a bit of context?10:24
arigatohow about we ask if anyone can contribute a windows build?10:24
fijal"build" or "buildbot"?10:24
Alex_Gaynorcan we steal a CPython buildbot?10:24
Garenfijal: at least, those were the two problems I had to build 1.7.x on windows10:24
fijalhow about we fix the buildbot to not pop up those things?10:24
arigatosorry, not feeling like it today (and I like the naive assumption in your question, implying "we just do one fix and all problems are gone" :-)10:25
fijalarigato: I can give it a go if I get credentials10:25
fijalno, but I claim one problem will be gone10:26
arigatoI fear that I cannot give you access without lac or jacob22_ saying it's ok10:26
Garenpypy 1.6 on windows can't do the translation for 1.7 without crashing though, sometime leading up to 1.7 that was fixed fwiw10:26
fijalcan you mail lac or jacob with that question?10:27
arigatosure10:27
fijalGaren: right, use cPython or a nightly build10:27
arigatoGaren: no, you really need to use CPython10:27
arigatobecause when run on top of pypy it's likely going to hit the 2GB limit10:27
Garenyup, that's what I had to do.  I guess you guys don't need any help then. :)10:27
arigatoGaren: well, getting a translated pypy-1.7 would help :-)10:28
fijalarigato: and I need admin credentials10:28
Garenit's not the 2GB limit, sometime around 1.6 ... all the way to near 1.7, self-translate would always fail10:28
fijalor a contact to whoever has admin credentials10:28
arigatoGaren: yes, it's a known issue, but in addition the 2GB limit is probably too low10:28
Garen.. and once 1.7 is built with cpython, I did have to add /LARGEADDRESSAWARE to do a self-translate, but it worked10:28
arigatoah, right, there is this completely obscure and not-on-by-default-I-ve-no-clue-why hack10:29
arigatofijal: well the issue is still that, last I saw, continuations were not working for me on windows10:30
fijalok10:30
GarenIt's because Microsoft thinks old apps will depend on some dirty C hacks that will suddenly break when they have access to more than 2GB ram10:30
arigatowe need stakkars's final word on that, to either fix them or to give up for 1.710:31
arigatoGaren: and Microsoft may be correct, of course10:31
ziemniak (~noob@host-91-192-88-139.elomza.pl) joined #pypy.10:32
arigatoyay.  the windows buildslave has current two dialog boxes waiting for Ok, and one "Welcome to Java! Click to install.."10:39
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/274 [12arigo, release-1.7.x]10:39
fijalwelcom to java is not stopping the build though right?10:40
fijalI think "most" of our ok boxes come from I-know-where10:40
fijaland can be fixed10:40
fijalbut I never got to the point where I can speak to someone who has admin privilages on that machine :/10:40
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1863 [12arigo]10:40
arigatothat would be Iko10:41
fijalarigato: if the own- works, please run the benchmarks10:41
fijalI'm about to go surfing10:41
arigatobenchmarks?10:42
arigatoah, on linux10:42
arigatoyes10:42
fijaljit-benchmark-3210:42
arigatohave fun10:42
fijalI'm also working on "how do we get info about aborts/forces/bad things"10:42
fijalso we can pass judgements whether using py.test hurts us because it forces frames10:42
Garenarigato: here's the link I was looking for: http://blogs.msdn.com/b/oldnewthing/archive/2004/08/12/213468.aspx11:02
CIA-6803serhat 07roundup * 10#940/Comparison with native application & cpython: 11:06
CIA-68That's good news. I improved the script a little bit, and fixed a bug. Please use11:06
CIA-68the new version in the benchmarks.11:06
CIA-68Thanks.. * 14https://bugs.pypy.org/issue94011:06
`fox` (~fox@95.232.111.141) joined #pypy.11:11
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) joined #pypy.11:12
Garenalso interesting: http://bugs.python.org/issue144949611:17
Garenbut sounds to me that unless you have a few pointers from different allocation blocks, and then later perform a comparison or subtraction with them, shouldn't see a problem.11:18
`fox` (~fox@95.232.111.141) left irc: Ping timeout: 248 seconds11:19
kenaan12hakanardo jit-refactor-tests 112839f3eef58b 15/: hg merge default ; hg revert test_zjit.py -r default11:37
kenaan12hakanardo jit-refactor-tests 114bc39a378477 15/pypy/jit/metainterp/history.py: Resurrect check_loops11:37
kenaan12hakanardo jit-refactor-tests 11c6f2c9f35460 15/pypy/jit/metainterp/history.py: merge11:37
kenaan12hakanardo jit-refactor-tests 115c74eff0f437 15/pypy/objspace/: hg merge default11:37
kenaan12hakanardo jit-refactor-tests 11f47785c0df50 15/pypy/module/micronumpy/test/test_zjit.py: convert test11:37
kenaan12hakanardo jit-refactor-tests 11e47a7b49880d 15/pypy/jit/metainterp/test/test_fficall.py: convert test11:37
kenaan12hakanardo jit-refactor-tests 113583510b011c 15/pypy/jit/metainterp/: kill check_loops11:37
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/723 [12hakanardo, jit-refactor-tests]11:39
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1863 [12arigo]11:40
arigatobah, "id" is a bad idea for a method name in W_Root11:42
`fox` (~fox@95.232.111.141) joined #pypy.11:42
arigatobecause the ast.Name subclass also uses "id"11:42
kenaan12hakanardo jit-targets 119a9928f8d64d 15/pypy/: hg merge jit-refactor-tests11:49
berdario (~Tiibiidii@host118-77-dynamic.10-79-r.retail.telecomitalia.it) left irc: Ping timeout: 260 seconds11:50
kenaan12arigo default 11ac1bed85839a 15/pypy/: Rename the method 'id' to 'unique_id', which is less prone to random conflicts with subclasses (e.g. ast.Name has a...11:56
kenaan12arigo op_malloc_gc 11f49a7ba4e56b 15/pypy/jit/backend/llsupport/: Refactor this logic into its own class, shared by the two GC strategies.11:56
kenaan12arigo op_malloc_gc 11ded0faf183e2 15/pypy/jit/backend/llsupport/: newstr, newunicode.12:29
timotimotimo"op" stands for "overpowered"? :)12:33
fijal (~fijal@197.172.191.33) left irc: Read error: Connection reset by peer12:33
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/723 [12hakanardo, jit-refactor-tests]12:49
stakkarshi12:49
arigatohi12:51
hruske (~Gasper@internet-188-196-76-189.narocnik.mobitel.si) joined #pypy.12:53
stakkarswhat about the buildbot, does it work again?12:53
arigatoI'm unaware of it failing to work12:54
stakkarsI saw a problem with a checkin on default and backed that out, but bbot crashed, again12:54
stakkarsbut you know of the faulty checkins?12:55
nedbat (~nedbat@python/psf/nedbat) joined #pypy.12:55
stakkarsI think it has a problem after all builds were failing12:56
hruske (~Gasper@internet-188-196-76-189.narocnik.mobitel.si) left irc: Remote host closed the connection12:56
arigatoah, in this sense --- yes12:57
stakkarsyes means it is restarted?12:59
arigatoyes (why don't you just check for yourself?)13:00
Squarepy (~Squarepy@ip82-139-118-42.lijbrandt.net) joined #pypy.13:01
stakkarssorry, I looked at my test runs and saw it crashed.13:02
arigatothe win32 are in a sorry state, fwiw13:02
stakkarshow can I check for myself just by running another build?13:02
Squarepy (~Squarepy@ip82-139-118-42.lijbrandt.net) left irc: Changing host13:02
Squarepy (~Squarepy@unaffiliated/squarepy) joined #pypy.13:02
arigatosorry, I guess there is some miscomprehension going on based on me not understanding exactly what part you are talking about13:03
stakkarsis there a status log when it was re-started?13:03
arigatoI had to restart it, and as a consequence I killed the running win32 builds, which were anyway going nowhere (they had been running for 8 days or so)13:05
stakkarsgood, thanks.13:05
stakkarswin32 hangs on something, I'm not using it any more.13:05
stakkarsI would like to install a win32 builder on snakepit13:06
arigatoright now, fijal would be interested in talking to you, I believe13:08
stakkarsthe misunderstandingg: I was talking about my win64 builds which crashed completely after I merged13:08
stakkarstwo checkins (bolz, arigo)13:09
stakkarsI backed them out and started a new build, but bbot gave me13:09
stakkars"exception" on build 6113:10
arigatoah, yes, sorry for doing these two checkins.  we didn't notice that they broke all translations, because they are purely in the interpreter13:10
arigatoI fixed it by now13:10
arigatoall buildbots since yesterday evening crashed with "exception", probably because they are getting too much error data13:11
stakkarsand that was my question: how can I know if the buildbot was restarted?13:12
arigatoah13:13
arigatoI don't know13:13
stakkarsok, miscomprehension resolved ;-)13:14
arigatoI don't know in general, but the miscomprehension came from the fact that today's crashes end up with buildbot.pypy.org showing "MemoryError",13:15
arigatoat least on some pages like the general summary page,13:15
arigatoso it is of course obvious just by checking this page that I restarted buildbot, because it now works again13:16
stakkarsok13:16
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.13:20
voidspace_ (~voidspace@87-194-212-65.bethere.co.uk) joined #pypy.13:31
voidspace_ (~voidspace@87-194-212-65.bethere.co.uk) left irc: Changing host13:31
voidspace_ (~voidspace@python/psf/voidspace) joined #pypy.13:31
voidspace (~voidspace@python/psf/voidspace) left irc: Ping timeout: 240 seconds13:31
Nick change: voidspace_ -> voidspace13:31
ojii (~ojii@134-117.79-83.cust.bluewin.ch) joined #pypy.13:32
voidspace (~voidspace@python/psf/voidspace) left irc: Ping timeout: 240 seconds13:37
voidspace (~voidspace@87-194-212-65.bethere.co.uk) joined #pypy.13:38
voidspace (~voidspace@87-194-212-65.bethere.co.uk) left irc: Changing host13:38
voidspace (~voidspace@python/psf/voidspace) joined #pypy.13:38
arigato (~arigo@89.204.138.64) left irc: Ping timeout: 244 seconds13:39
whitelynx (~whitelynx@75.111.197.204) left irc: Quit: Leaving13:45
witulski (~stupsi@ip-176-198-244-135.unitymediagroup.de) joined #pypy.13:46
JaRoel (~jaroel|4d@2001:980:7133:1:dd05:425b:e88a:57ee) joined #pypy.13:49
voidspace_ (~voidspace@87-194-212-65.bethere.co.uk) joined #pypy.13:51
voidspace_ (~voidspace@87-194-212-65.bethere.co.uk) left irc: Changing host13:51
voidspace_ (~voidspace@python/psf/voidspace) joined #pypy.13:51
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) left irc: Ping timeout: 258 seconds13:51
voidspace (~voidspace@python/psf/voidspace) left irc: Ping timeout: 240 seconds13:52
Nick change: voidspace_ -> voidspace13:52
witulski (stupsi@ip-176-198-244-135.unitymediagroup.de) left #pypy.13:56
mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.13:59
mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host13:59
mfoord (~anonymous@python/psf/voidspace) joined #pypy.13:59
_iori_ (~iori@110-133-45-54.rev.home.ne.jp) joined #pypy.14:03
bivab (~david@ip-62-143-59-5.unitymediagroup.de) joined #pypy.14:15
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 258 seconds14:23
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) left irc: Quit: Leaving...14:35
hruske (~Gasper@2001:15c0:66e9:14:383e:18ff:fe7f:6c95) joined #pypy.14:51
Squarepy (~Squarepy@unaffiliated/squarepy) left irc: Quit: Leaving14:53
jonathan__ (~jonathan@bas1-montreal29-1242341904.dsl.bell.ca) joined #pypy.14:55
kenaan12ctismer win64-stage1 113c028f4484c5 15/pypy/rpython/lltypesystem/ll2ctypes.py: starting to replace sys.maxint15:02
kenaan12ctismer win64-stage1 1144a146430807 15/pypy/rlib/rmmap.py: fixed mmap.resize for win6415:03
kenaan12ctismer win64-stage1 1130a323a07b06 15/pypy/: merge defauli15:03
kenaan12hakanardo jit-targets 11e307e21d62af 15/pypy/jit/metainterp/test/test_fficall.py: fix tests15:04
kenaan12hakanardo jit-targets 113a8d0807cd75 15/pypy/jit/metainterp/test/test_warmstate.py: fix test15:04
kenaan12hakanardo jit-refactor-tests 119c22905d5d42 15/pypy/: hg merge default15:04
kenaan12hakanardo default 11d9ef0a8f3fa2 15/pypy/: merge jit-refactor-tests. This replaces check_loops with check_resops which counts every operation including th...15:04
kenaan12hakanardo jit-refactor-tests 119ab059335d1f 15/: close merged branch15:05
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/724 [12hakanardo, jit-targets]15:06
bbot213Exception: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/724 [12hakanardo, jit-targets]15:06
kenaan12hakanardo jit-targets 1157f992618fe2 15/pypy/: hg merge default15:07
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/725 [12hakanardo, jit-targets]15:07
mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord15:11
mat^2 (mathias@87.55.0.198) joined #pypy.15:17
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 240 seconds15:18
nedbat (~nedbat@python/psf/nedbat) joined #pypy.15:26
fijal (~fijal@197.175.151.177) joined #pypy.15:28
nedbatfijal: thanks for looking into the sandboxing issue on Webfaction.  I should probably start a pypy-dev thread about it, given the timezones between us.15:29
fijalnedbat: hahaha :)15:30
fijalI'm here now15:30
fijalbut in general yeah, pypy-dev is safer (or bugs.pypy.org)15:30
fijalso there are two issues15:30
fijalone is that we depend on ll2ctypes, which we should not15:30
fijalthe other is that execution from /tmp is necessary to run many things in pypy15:30
nedbatfijal: is the /tmp a compile-time constant I can change to something else?15:31
fijalprobably15:31
fijallet me look15:31
fijalit's whatever tempfile.gettempdir() returns15:32
fijalso I'm sure there is an environment variable or something15:32
nedbatfijal: ok, great.  I haven't learned enough about the pypy code to be able to spelnuk for this sort of info myself yet.15:32
zooko (~user@97-118-101-86.hlrn.qwest.net) joined #pypy.15:33
nedbatfijal: what does the ll2ctypes dependency cause?15:33
nedbatfijal: what problem?15:33
fijalexecuting stuff in /tmp15:34
fijalfor one15:34
fijalit's probably import mess or something15:34
nedbatfijal: from tempfile.py:     for envname in 'TMPDIR', 'TEMP', 'TMP':15:35
nedbatso that seems movable without a recompile.15:35
fijalyes15:35
nedbatfijal: Looks like that makes it runnable!  os.environ['TMPDIR'] = mydir  Thanks!15:40
fijalcool15:40
fijalnedbat: you know that there are not that many users of pypy sandbox?15:40
nedbatfijal: I have realized that.15:40
fijalit's still probably way more secure than anything else out there just because of the way it's done15:41
fijalbut it might have rough edges on the usability side15:41
nedbatfijal: I had to implement os.fstat to get my code to run.15:41
fijalcool15:41
fijaldid you submit a patch back upstream?15:41
nedbatfijal: yes, and it spews many lines to the output.15:41
nedbatI'm going to when I have a test for it.15:41
fijalcool15:41
nedbatI'm also talking to geoff thomas at mit who went through some of this.15:41
fijaloh cool15:42
Action: fijal didn't know15:42
nedbateventually i'll also want writable filesystems in the sandbox, so there will be more implementation work in that layer15:43
fijalcool15:44
fijalwe would welcome a sandbox maintainer btw15:44
nedbati'm not sure I'm the guy for that, but I am dipping my toe in.15:45
fijalyou never know :)15:46
fijalwhy not btw?15:46
fijalI mean, someone who wants to edit some code is probably already better than the current situation15:46
nedbatfijal: you are right about all of that.  I have existing side project commitments, that's all.15:46
nedbatfijal: and I don't know yet what skills are needed, or how much of the code base the work would span.15:47
nedbatfijal: but i was motivated to get it this far, so I'm already in the 1%  #occupypypy :)15:47
fijalright15:47
fijalwell, as far as skill goes, writing tests and asking questions15:47
fijalas far as commitments goes, you're the only person who really cares, so I suppose expectations are low :)15:48
nedbatfijal: any idea why sandbox is so lonely?15:49
fijalbecause we don't use it15:49
fijaland while it's a neat idea, it's also unfinished15:50
nedbatfijal: it seems like lots of people are working hard to execute untrusted code on servers, and this is a nearly-working solution.15:50
fijalyes15:50
fijalwell15:50
nedbatcan you give me a thumbnail of what remains to finish it?15:50
fijalthis is one of the areas in pypy, where I for example have no personal interest, but would work for money for example15:50
fijalmostly how do you interface with other pieces of code is what has to be done15:50
fijallike for any given sandbox, you usually want to interact with some other code15:51
mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.15:51
mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host15:51
mfoord (~anonymous@python/psf/voidspace) joined #pypy.15:51
fijalit depends what you do, but some sort of framework would be nice15:51
nedbatyes, I can see that.  It took me a while to wrap my head around the three different examples of how to invoke it.15:54
fijal (~fijal@197.175.151.177) left irc: Read error: Connection reset by peer15:54
fijal (~fijal@41.54.165.196) joined #pypy.15:54
nedbatfijal: first I have to get it to not write 300k of log when executing "a = 17"15:55
fijalthat should be relatively easy15:55
fijalprobably by setting debug = False15:55
fijalor DEBUG = False15:55
fijalor so15:55
fijalnedbat: do you have commit rights btw?15:56
nedbatfijal: I figure it's something simple. 15:56
nedbatfijal: no, do I want those?  :)15:56
fijalthat's not a question for me :)15:56
fijalgenerally we find that giving anyone commit rights works well15:56
fijalif you're unsure, make it a branch15:56
nedbatfijal: true15:56
fijalwe *never* had an issue with that15:57
fijalso, do you want some?15:57
nedbatfijal: sure.  I can decide not to use it.15:57
fijalsure15:57
fijalwhat's your bitbucket username>15:58
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) joined #pypy.15:58
nedbat"ned"15:59
fijalnice shot :)15:59
nedbatit should have been nedbat to match everything else, but if "ned" is available, it's hard to resist.15:59
fijalnedbat: ok, cool, here we go15:59
fijalfeel free to create a branch and commit your stuff there16:00
nedbatthanks!16:00
fijalif you ask to merge, we will complain about tests :)16:00
fijalbranches have no policies16:00
nedbatfijal: ok, that's reasonable.16:00
oal (u4126@gateway/web/irccloud.com/x-sbljvlxafhkylpzb) left irc: Max SendQ exceeded16:08
oal (u4126@gateway/web/irccloud.com/x-kmnmmhsnowwhyinc) joined #pypy.16:10
oal (u4126@gateway/web/irccloud.com/x-kmnmmhsnowwhyinc) left irc: Max SendQ exceeded16:11
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds16:13
mat^2 (mathias@87.55.0.198) left irc: Ping timeout: 240 seconds16:16
qbproger_ (~qbproger@c-76-104-159-248.hsd1.wa.comcast.net) joined #pypy.16:23
oal (u4126@gateway/web/irccloud.com/x-gaxgashaxfmgqkza) joined #pypy.16:25
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) left irc: Quit: Computer has gone to sleep16:27
fijalmwhudson: ping16:27
davisagli (~davisagli@davisagli.com) left irc: Excess Flood16:28
davisagli (~davisagli@davisagli.com) joined #pypy.16:29
bivab (~david@ip-62-143-59-5.unitymediagroup.de) left irc: Quit: bivab16:32
kkris1 (~kris@93-82-47-186.adsl.highway.telekom.at) joined #pypy.16:35
kkris (~kris@93-82-45-16.adsl.highway.telekom.at) left irc: Ping timeout: 276 seconds16:36
amaury__ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.16:38
fijal (~fijal@41.54.165.196) left irc: Read error: Connection reset by peer16:41
amaury__ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 252 seconds16:43
fijal (~fijal@41.49.63.198) joined #pypy.16:46
tav (~tav@host-92-20-44-92.as13285.net) left irc: Ping timeout: 252 seconds16:47
tav (~tav@host-92-20-59-250.as13285.net) joined #pypy.16:52
mfoord (~anonymous@python/psf/voidspace) left irc: Quit: mfoord16:55
`fox` (~fox@95.232.111.141) left irc: Ping timeout: 244 seconds17:00
zooko (~user@97-118-101-86.hlrn.qwest.net) left irc: Ping timeout: 258 seconds17:04
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) joined #pypy.17:05
stakkarsfijal: hi!17:06
stakkarsI'm wondering if "allocate_in_far_regions" makes sense on windows?17:07
stakkarsdoes somebody know if Windows can allocate without reserving memory?17:07
amaury__ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.17:16
stakkarsamaury__: hi! Do you know if windows reserves memory on mmap()?17:21
stakkarsiow., does "allocate_in_far_regions" make sense, or should I disable it?17:21
fijal (~fijal@41.49.63.198) left irc: Read error: Connection reset by peer17:22
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/725 [12hakanardo, jit-targets]17:24
Alex_Gaynorronny: ping17:26
ronnyAlex_Gaynor: pong?17:26
ronnyAlex_Gaynor: sup?17:26
Alex_Gaynorronny: I'm getting some pretty scary errors on trying to hg pull: (grr paste.pocoo is being slow)17:27
Alex_Gaynorhttp://paste.pocoo.org/show/513248/17:27
Alex_GaynorThe only similar bug I see in the hg bug tracker says its an FS issue, but that's assuming the `hg verify` is clean http://mercurial.selenic.com/bts/issue128117:28
Alex_GaynorAny idea what's up?17:28
lmoura (~lmoura@186.215.206.130) left irc: Ping timeout: 240 seconds17:28
Alex_GaynorFWIW hg id errors out.17:28
ronnyAlex_Gaynor: so your repo broke and is in a interesting state17:31
Alex_GaynorI have no idea what I did :/17:31
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.17:33
fijal (~fijal@41.49.203.203) joined #pypy.17:33
ronnyAlex_Gaynor: for some reason some of your manifest entres got corrupted and are unreadable (manifests entries store links to all filerefs of a revision)17:33
ronnyAlex_Gaynor: it looks like you got some garbage data in the manifest17:33
Alex_Gaynoris there a way to fix it?17:34
fijalhey alex17:34
tav (~tav@host-92-20-59-250.as13285.net) left irc: Read error: Connection reset by peer17:34
Alex_Gaynorhey fijal, my hg repo is totally busted17:34
ronnyAlex_Gaynor: its probably best to make a new clone17:34
fijalstakkars: as far as I know allocate_in_far_regions can be entirely disabled17:34
fijalAlex_Gaynor: nice17:34
voidspace (~voidspace@python/psf/voidspace) left irc: Ping timeout: 240 seconds17:35
stakkarsfijal: I meant: _should_ it be disabled for windows, or is it possible without tons of memory?17:35
fijalstakkars: there is no reason these days it exists at all17:36
fijalso if it's any sort of suspect for anything, just disable it17:36
fijalAlex_Gaynor: I have business idea17:36
fijallet's sell pypy that's faster than the last released version!17:37
Alex_GaynorI think we call that "latest nightly".17:37
ronnypay per nightly?17:37
fijalyeah17:38
fijalbut we can package it in a nice box and call it "released"17:38
ronnyAlex_Gaynor: do you have any changesets yo need in your local repo?17:38
Alex_Gaynorronny: nope, already doing a fresh clone17:38
ronnyAlex_Gaynor: what did you do recently with that repo? (stuff like strip/mq)17:39
Alex_GaynorI pulled, I branched, and I committed.17:40
Alex_GaynorI'v never used strip or mq.17:40
fijalAlex_Gaynor: naughty :)17:40
ronnylol17:41
stakkarsfijal: I thought it was a good idea to have memory far away, to find errors, especially on win6417:42
fijalstakkars: I doubt that17:42
fijaland it can potentially be a performance problem17:42
fijalsince we deal with that in a systematic manner these days17:42
fijalthere is no special code and no dependency17:42
tav (~tav@host-92-20-2-83.as13285.net) joined #pypy.17:43
stakkarsfijal: I'm working on correctness, not performance right now. NP to disable, later.17:43
fijaltav: hey tav17:43
stakkarsb17:43
stakkarsut is the feature no longer existing/obsolete?17:44
fijalstakkars: seriously, the way current code is written makes it much harder to have specific bugs related to this17:44
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) left irc: Quit: Computer has gone to sleep17:44
fijaland it's tested separately17:44
stakkarswhy is the function there, at all?17:44
fijalso there is probably little to no point in keeping it17:44
fijalbecause originally our 64bit support was more fragile17:44
stakkarsthen maybe we shotlt trash this completely?17:45
stakkarsbefore I spend even more time in trying it on windows :-)17:45
stakkarss/:-)/:-(/17:46
fijaljust disable it for now17:46
fijalwe can trash it later17:46
cpets (4f75d39d@gateway/web/freenode/ip.79.117.211.157) joined #pypy.17:52
stakkarsfijal: still my question: does windows reserve memory, or can it act like linux?17:53
rguillebert (~rguillebe@fec76-1-78-234-124-105.fbx.proxad.net) left irc: Ping timeout: 244 seconds18:04
kenaan12ctismer win64-stage1 11246401dd1c9c 15/pypy/interpreter/baseobjspace.py: merge glitch18:05
kenaan12ctismer win64-stage1 115104f75cdb84 15/pypy/rpython/lltypesystem/ll2ctypes.py: hacked a bit on ll2ctypes/do_allocation_in_far_regions stopped this, fijal claims it makes no longer sense18:05
kenaan12ctismer win64-stage1 1157e0cb25305f 15/pypy/: hacked a bit on ll2ctypes/do_allocation_in_far_regions stopped this, fijal claims it makes no longer sense18:05
fijalstakkars: no idea18:08
daniloaf (~daniloaf@150.165.63.86) joined #pypy.18:20
Action: fijal sees a lot of pypy-related tweets in japanese18:20
Alex_Gaynorfijal: it means something in japanese, many of them aren't related to us18:22
rguillebert (~rguillebe@fec76-1-78-234-124-105.fbx.proxad.net) joined #pypy.18:22
fijalno18:22
fijala lot of the are18:22
fijalhttp://mycomjournal.jp/news/2011/11/24/050/index.html18:22
fijalat least as long as it says 1.7 ;-)18:22
cpets (4f75d39d@gateway/web/freenode/ip.79.117.211.157) left irc: Ping timeout: 265 seconds18:27
timotimotimowhat pypy optimisation makes cpython allocate huge amonts of data when evaluating type(range(100000000)), but lets pypy evaluate type(range(1000000000000000000)) in an instant? is pypys range a special object that only evaluates its data when iterated over it?18:35
mattip (~chatzilla@109.66.115.123) joined #pypy.18:36
mattiphi18:36
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) joined #pypy.18:36
nedbat (~nedbat@python/psf/nedbat) joined #pypy.18:36
CIA-6803Scriptor 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 18:36
CIA-68[new] The attached script (mostly generated by another one, but is not artificial,18:36
CIA-68such situations really occur in my program) causes pypy 1.5--1 ... * 14https://bugs.pypy.org/issue94118:36
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) left irc: Client Quit18:36
fijaltimotimotimo: yes18:37
fijalmattip: hi18:37
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) joined #pypy.18:37
CIA-6803fijal 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: [chatting] Oops, this is clearly a bug. * 14https://bugs.pypy.org/issue94118:38
kenaan12mattip matrixmath 1156a9a8fc133e 15/pypy/module/micronumpy/: split tests - work in progress18:38
kenaan12mattip matrixmath 11db1df6669568 15/pypy/module/micronumpy/: tests, implementation of flat18:38
kenaan12mattip matrixmath 115f446b54dc5c 15/pypy/module/micronumpy/: Merge work in progress18:38
kenaan12mattip matrixmath 11f345cd981de5 15/pypy/module/micronumpy/: Merge to avoid creating new head, not ready for review18:38
mattipfijal: I just commited a "flat" iterator, tests and implementation. Could you take a look? specifically: changeset db1df666956818:39
CIA-6803fijal 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 18:40
CIA-68For what is worth (not related to the fact that this *is* a bug), probably18:40
CIA-68executing this in a loop would yield better performance on pypy. Just ... * 14https://bugs.pypy.org/issue94118:40
fijalcool18:40
fijalmattip: is the branch ready to be merged btw?18:40
mattipNo, the path leading to the changeset db1df6669568 is, it includes arange, transpose and flat.18:40
arigato (~arigo@82.113.98.65) joined #pypy.18:41
mattipThe head of the branch is still a wip: setting a shape on a view is messy and broken.18:41
fijalok18:41
apollo13 (~apollo13@unaffiliated/apollo13) left irc: Ping timeout: 240 seconds18:41
daniloaf (~daniloaf@150.165.63.86) left irc: Quit: Leaving18:41
fijalI can probably work with that, but for the future reference, we create multiple branches for that18:41
fijalarigato: hi18:41
arigatohi18:41
mattipI thought if I pick some low lying fruit it would clear some creative juices.18:42
Sho_ (~EHS1@kde/hein) joined #pypy.18:42
fijalmattip: sure, that's great18:42
mattipAhh, OK, I will create a different branch in the future. Still learning the process.18:42
fijalyeah, hg isn't entirely easy to use18:42
mitchellh (~mitchellh@c-98-245-152-34.hsd1.co.comcast.net) left irc: Quit: Computer has gone to sleep18:42
fijalbut it's fine for now18:42
fijalwe usually do branch-per-feature though18:43
fijalarigato: are you using tannit?18:43
kenaan12fijal better-jit-hooks 1184c2db217151 15/pypy/: a branch to experiment with some better JIT hooks18:43
kenaan12fijal better-jit-hooks 11b201d7c9f25f 15/pypy/jit/metainterp/: spring cleaning - remove ABORT_BAD_LOOP (unused)18:43
arigatofijal: no, feel free to18:43
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/955 [12fijal]18:43
Action: fijal feels free to18:43
mat^2 (mathias@87.55.0.198) joined #pypy.18:43
mattipfijal: I read the irc logs so if you have some comments even when I am not around, stick my name on them and I will see it later.18:44
fijalmattip: ok18:44
fijalI did not so fart18:44
arigatouh.  reading issue941: it's a huuuuge non-loopy generated piece of code18:44
fijalyes18:44
fijalI'm fine with saying "it'18:44
fijals slow"18:44
fijalbut it should not crash the JIT18:44
arigatosure18:44
zooko (~user@97-118-101-86.hlrn.qwest.net) joined #pypy.18:44
fijalit should probably abort with TRACE_TOO_LONG or so18:45
fijalzooko: hi18:45
arigatoI'm wondering where the JIT even triggers18:45
mattip (chatzilla@109.66.115.123) left #pypy.18:45
fijalarigato: good point18:45
arigatoah, maybe in the list comprehension line 718:45
fijalpfff18:45
fijalthat would be obscure18:45
cpets (4f75d39d@gateway/web/freenode/ip.79.117.211.157) joined #pypy.18:46
fijalmattip18:46
arigatomy guess is that the JIT crashes because the function is too big, even though it does't attempt to trace through it18:46
fijalmattip: arange does not work for floats18:46
fijalbecause they'll get automatically marked as ints18:46
srin (~srin@pool-70-111-141-236.nwrk.east.verizon.net) joined #pypy.18:47
fijalmattip: it also does not test that it works correctly with floats 18:48
fijal(it does not)18:48
fijalmattip: the flatiter should raise OperationError if object is not of BaseArray type18:49
fijalthere are no tests for creation of numpy.flatiter18:50
voidspace (~voidspace@87-194-212-65.bethere.co.uk) joined #pypy.18:50
voidspace (~voidspace@87-194-212-65.bethere.co.uk) left irc: Changing host18:50
voidspace (~voidspace@python/psf/voidspace) joined #pypy.18:50
fijal_attrs_ can be ommited entirely, but this is broken (it has to contain all attributes like __slots__)18:50
fijaltranspose should be tested at the very least on odd number of dimensions, like 1 or 318:51
fijalit should also handle scalar (somehow)18:51
fijal</review>18:52
zookohi!18:52
CIA-6803Scriptor 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 18:52
CIA-682fijal: I'd like to execute this in a loop if I knew the general "formula" for18:52
CIA-68each mz0[i][j]. But I don't know it, the formulae are generated by ... * 14https://bugs.pypy.org/issue94118:52
stakkars (~tismer@p5DDB7E68.dip.t-dialin.net) left irc: Quit: stakkars18:55
CIA-6803fijal 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 18:57
CIA-68Then what I would do is to generate (from config) an array of indexes and18:57
CIA-68formulas and then apply those in a loop. The generated input would loo ... * 14https://bugs.pypy.org/issue94118:57
apanda (~apanda@c-67-170-53-221.hsd1.wa.comcast.net) joined #pypy.18:58
apanda (~apanda@c-67-170-53-221.hsd1.wa.comcast.net) left irc: Remote host closed the connection18:58
stakkars (~tismer@p5DDB7E68.dip.t-dialin.net) joined #pypy.18:59
kenaan12fijal matrixmath 11282d885f914f 15/pypy/module/micronumpy/REVIEW: add a REVIEW file, just because IRC logs might be lost19:01
fijalarigato: I'm adding hooks for aborts() shout if you don't like that idea19:03
arigatofine19:03
asmeurer_ (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.19:04
fijalso for example you can put a pdb in case of abort()19:04
fijalI hope it'll be useful19:04
apollo13 (~apollo13@unaffiliated/apollo13) joined #pypy.19:04
fijalbut I guess you never know until you try :)19:04
arigato:-)19:05
CIA-6803Scriptor 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 19:05
CIA-68Eval()+loop is faster then instruction set??? Surprising info.19:05
CIA-68I suppose this idea also will help me to simplify some another actions and19:05
CIA-68opti ... * 14https://bugs.pypy.org/issue94119:05
rguillebert (~rguillebe@fec76-1-78-234-124-105.fbx.proxad.net) left irc: Ping timeout: 240 seconds19:07
arigatofijal: can I also use one core of tannit?19:07
arigatoah no, you're running benchmarks19:10
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds19:10
Action: arigato pushes to bitbucket then19:10
kenaan12arigo default 11299a1d66e8a2 15/pypy/jit/metainterp/resume.py: When hitting a too large value, cleanly give up jit-compilation instead of crashing with ValueError.19:11
kenaan12arigo default 119f026ead5acb 15/pypy/: merge heads19:11
fijal (~fijal@41.49.203.203) left irc: Read error: Connection reset by peer19:18
kenaan12ctismer win64-stage1 11af23ec1f5f06 15/pypy/: hacked a bit on ll2ctypes/do_allocation_in_far_regions stopped this, fijal claims it makes no longer sense19:18
kenaan12arigo default 11cd9529760135 15/pypy/jit/metainterp/: Clean-up.19:18
arigatostakkars: I would say it makes no sense for now to do anything with ll2ctypes.py19:18
bbot2Started: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/62 [12ctismer, win64-stage1]19:19
fijal (~fijal@41.55.31.11) joined #pypy.19:20
arigatogrumble, /me broke translation19:21
fijaluh19:21
CIA-6803fijal 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 19:21
CIA-68Looking at your case, the eval is looped only over a relatively short list of19:21
CIA-68elements and eval is not really that much worse than normal code f ... * 14https://bugs.pypy.org/issue94119:21
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/62 [12ctismer, win64-stage1]19:23
bbot2Started: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/63 [12ctismer, win64-stage1]19:23
kenaan12arigo default 1117fd3198ef36 15/pypy/jit/metainterp/optimizeopt/optimizer.py: Translation fix.19:24
arigatosorry, fixed now19:24
fijalgood :)19:25
CIA-6803arigo 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 19:26
CIA-68Thanks! Should be at least fixed by 299a1d66e8a2 and following clean-ups. (But19:26
CIA-68indeed, a long non-looping piece of code run just once is never ... * 14https://bugs.pypy.org/issue94119:26
_iori_ (~iori@110-133-45-54.rev.home.ne.jp) left irc: Read error: Connection reset by peer19:26
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/63 [12ctismer, win64-stage1]19:26
bbot2Started: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/64 [12ctismer, win64-stage1]19:26
bbot213Exception: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/64 [12ctismer, win64-stage1]19:28
bbot2Started: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/65 [12ctismer, win64-stage1]19:29
jamesaxl (~jamesaxl@41.248.194.73) joined #pypy.19:30
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 240 seconds19:31
jamesaxlhi i wanna compile pypy for slackware , could you tell me please pypy dependencies19:31
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/65 [12ctismer, win64-stage1]19:32
bbot2Started: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/66 [12ctismer, win64-stage1]19:32
rguillebert (~rguillebe@fec76-1-78-234-124-105.fbx.proxad.net) joined #pypy.19:32
fijaljamesaxl: sure19:32
fijaljamesaxl: for the debian package they're here: https://bitbucket.org/pypy/build/src19:34
fijallet me fish the dependencies19:34
jamesaxlfijal, yeah sure19:34
stakkarshum. I always get on buildbot:19:34
stakkarsremoteFailed: [Failure instance: Traceback (failure with no frames): <class 'twisted.internet.error.ConnectionLost'>: Connection to the other side was lost in a non-clean fashion.19:34
stakkars]19:34
Action: fijal is lost19:34
arigatofijal: getting-started-python.rst19:34
fijalarigato: ah!19:34
fijalhttp://doc.pypy.org/en/latest/getting-started.html19:35
arigatobut this lists a bit too many of them, depending on what you want19:35
fijalnot this one?19:35
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/66 [12ctismer, win64-stage1]19:35
fijalhttp://doc.pypy.org/en/latest/getting-started-python.html19:35
fijalthis one19:35
bbot2Started: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/67 [12ctismer, win64-stage1]19:35
stakkarsit starts, collects 256 tests, also runs the expected number of pythons, but then this message :-(19:35
fijaljamesaxl: ^^^19:35
fijaljamesaxl: those are all for building though19:35
jamesaxlfijal, honto !!!19:35
fijalyou need only libraries (non-dev versions) for the actual binary19:36
CIA-6803Scriptor 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 19:36
CIA-68fijal: should I manually look for equal formulae to prevent them from being19:36
CIA-68computed twice or more?19:36
CIA-68arigo: just started to build mercurial sna ... * 14https://bugs.pypy.org/issue94119:36
fijaland you don't need libgc19:36
CIA-6803fijal 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 19:37
CIA-68fijal: should I manually look for equal formulae to prevent them from being19:37
CIA-68computed twice or more?19:37
CIA-68sounds like a sane optimization for me :) ... * 14https://bugs.pypy.org/issue94119:37
fijaljamesaxl: does that answer your questions?19:37
arigatoI suspect you need to install them, then do a build, and finally look at the binary to know the exact dependencies19:37
fijalmeh19:37
fijalit seems to me that debian/ubuntu will be the last to have pypy in packages :/19:38
fijala bit sad given how many python people work for canonical19:38
bbot24Finished ??: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/67 [12ctismer, win64-stage1]19:38
bbot2Started: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/68 [12ctismer, win64-stage1]19:39
stakkarsheeeelp, what is happening with bbot?19:39
jamesaxlfijal, no, :) but no prob,, arigatou 19:39
stakkarsit restarts my build all the time 19:39
bbot213Exception: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/68 [12ctismer, win64-stage1]19:40
zookohttps://tahoe-lafs.org/trac/tahoe-lafs/browser/trunk/src/allmydata/immutable/upload.py?rev=5232#L141019:40
zookoSorry, wrong channel.19:41
stakkarsfijal: , arigato : can I have caused this? I did not modify the buildbot, but it looses connection19:43
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.19:44
arigatono clue, sorry19:44
exarkunlink?19:44
CIA-6803Scriptor 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 19:47
CIA-68It definitely is, especially taking into account the multiple code usage after19:47
CIA-68one generation. I appreciate your advices! * 14https://bugs.pypy.org/issue94119:47
bbot2Started: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/69 [12ctismer, win64-stage1]19:48
stakkarsarigato: all ok, thanks. Some resources must have run out.19:52
fijal (~fijal@41.55.31.11) left irc: Read error: Connection reset by peer19:52
stakkarsarigato: I restarted the machine, which took very long, now it works19:53
ssbr (~scorchsab@206-188-65-13.cpe.distributel.net) joined #pypy.20:00
ziemniak (noob@host-91-192-88-139.elomza.pl) left #pypy.20:06
jamesaxldoes pypy2.7 compiled by libssl-1.0?20:12
jamesaxlpypy1.720:14
exarkunjamesaxl: That doesn't quite make sense.20:15
exarkunjamesaxl: Are you wondering if one of the binary packages of PyPy 1.7 requires OpenSSL 1.0?20:15
jamesaxlexarkun, cause i'm uing 0.9.820:17
exarkunAre you saying "Yes, that is the question I am asking"?20:18
jamesaxlexarkun, i gave the last answer, cause you know the question is "YES"20:19
jamesaxlthe answer of this question20:21
exarkun`ldd pypy` says the 32 bit PyPy 1.7 binary is linked against 0.9.8.20:22
jamesaxlexarkun, and i got /lib64/libcrypto.so.0.9.8: no version information :(20:23
exarkunSo you're not asking if it requires 1.0, you're asking if it requires 0.9.820:25
exarkunAnd as far as I know, it does require 0.9.8.20:25
jamesaxlexarkun, yeah it is too bad20:27
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.20:31
stakkarsarigo: an issue with test_flatten.py::test_force_cast20:31
stakkarswhen testing in line 810, I get an exception, which is IMHO correct20:32
arigatostakkars: may I suggest you some priorities?  like don't worry about the JIT at all even before getting basic translations working20:32
stakkarsbecause ULONG is actually UINT20:33
rguillebert (~rguillebe@fec76-1-78-234-124-105.fbx.proxad.net) left irc: Ping timeout: 240 seconds20:33
stakkarsyes, it is just annoying ;-)20:33
arigatostakkars: if I may try again to suggest you a way of progressing: I'd recommend a goal list going around like that (in this order):20:34
stakkarsok?20:34
arigato1. make sure pypy/annotation/test and pypy/rpython/test pass, at least up to obscure INT-vs-Signed issues20:35
arigato2. work on pypy/translator/c/test/test_typed.py20:35
arigato3. try to run "translate.py targetrpystonedalone" until it actually works20:35
arigatothat would already be a good step20:36
stakkarsok, understood20:36
fprimex (~fprimex@fl-67-233-144-228.dhcp.embarqhsd.net) joined #pypy.20:40
arigatoalso, I'm not sure if you fixed it, but win32 _continuation was not working for me20:46
stakkarsthat was fixed, but I'll check again20:47
rguillebert (~rguillebe@fec76-1-78-234-124-105.fbx.proxad.net) joined #pypy.20:50
Ademan (~dan@adsl-71-141-224-79.dsl.snfc21.pacbell.net) joined #pypy.20:52
stakkarsarigato: just tried it.20:53
arigato[platform:ERROR] NMAKE : fatal error U1073: don't know how to make 'switch_x86_msvc.a.obj'20:53
arigatois what I get20:53
apollo13 (~apollo13@unaffiliated/apollo13) left irc: Ping timeout: 260 seconds20:53
stakkars_continuation works on win32 default and my branch20:53
arigatoI get the error when translating a complete pypy, or when running the test in rlib/test/test_rstacklet.py20:54
arigatobut module/_continuation/test works indeed20:54
stakkarsaha!20:54
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/955 [12fijal]20:56
nedbat (~nedbat@python/psf/nedbat) joined #pypy.20:59
stakkarsI see. the nmake version is a bit under-developed, still...21:02
CIA-6803Scriptor 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: arigo: mercurial snapshot still fails. * 14https://bugs.pypy.org/issue94121:04
ronnyarigato: wrt #941 - i wonder if it triggers the jit on another level since the code is so long21:07
ronny(thats just a random tought that poped into my mind when reading the trace)21:07
apollo13 (~apollo13@unaffiliated/apollo13) joined #pypy.21:11
amaury__ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 248 seconds21:18
stakkars_ (~tismer@p5DC469D4.dip0.t-ipconnect.de) joined #pypy.21:28
apanda (~apanda@c-67-170-53-221.hsd1.wa.comcast.net) joined #pypy.21:28
stakkars_ (~tismer@p5DC469D4.dip0.t-ipconnect.de) left irc: Ping timeout: 252 seconds21:32
stakkars_ (~tismer@p5DC469D4.dip0.t-ipconnect.de) joined #pypy.21:35
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) joined #pypy.21:37
tilgovi (~randall@c-98-210-155-124.hsd1.ca.comcast.net) left irc: Changing host21:37
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.21:37
amaury__ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.21:37
stakkars_ (~tismer@p5DC469D4.dip0.t-ipconnect.de) left irc: Client Quit21:38
nedbati'm running a sandbox on a server, and it seems to take 2 seconds to run one line of Python in the sandbox and return the result.  The one line is "a = 17".21:42
nedbatI see a warning like "cannot find your CPU L2 cache size in /proc/cpuinfo", so there's startup work happening.  How can I build pypy so there is the least overhead, even if executing Python itself is slower?21:42
CIA-6803arigo 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: Scriptor: works for me. * 14https://bugs.pypy.org/issue94121:46
ronnynedbat: the sandbox s rather expensive to start up, since all direct syscalls are instead marshaled via stdio21:52
nedbatronny: yeah :(   21:53
ronnynedbat: is there really a need to start a new sandbox per command?21:54
ronnyonce its running, and has the code, anything thats not io can be fast21:54
nedbatronny: I'm thinking about that, but I'm looking for overhead that can be skipped in any case.21:54
nedbatronny: I'm executing student code on a server.21:54
stakkars_ (~tismer@p5DC469D4.dip0.t-ipconnect.de) joined #pypy.22:06
CIA-6803Scriptor 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: Do I have to use tip (17fd3198ef36) or something other? * 14https://bugs.pypy.org/issue94122:07
CIA-6803arigo 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 22:09
CIA-68Yes, 17fd3198ef36 is the only revision recent enough to contain the relevant22:09
CIA-68changesets (299a1d66e8a2 to 17fd3198ef36). * 14https://bugs.pypy.org/issue94122:09
ronnynedbat: you can have a pool that starts new ones every time one is taken for work22:10
ronnynedbat: that way the sudents have less startup cost22:10
ronny(just have it load common libs, then wait for a command22:10
CIA-6803Scriptor 07roundup * 10#941/15000+ matrix filling instructions kill pypy-c: 22:10
CIA-68Oh, I'm sorry, I've just made a mistake while symlink creation. It works. Thanks22:10
CIA-68a lot! * 14https://bugs.pypy.org/issue94122:10
arigatoronny: but then we loose one of the purposes of using sandboxing, by allowing malicious code from one student to (at least) do random things in the next student's code22:11
nedbatarigato: yes, I've been pondering what is possible when sandboxing sequentially.22:16
apanda (~apanda@c-67-170-53-221.hsd1.wa.comcast.net) left irc: Read error: Connection reset by peer22:17
nedbatalso, not sure how to have the sandbox wait for commands, it's stdin and stdout are tied to the parent process.  I'm sure it's possible, i just still haven't sorted out the layers well yet.22:17
ronnyarigato: thats not what i meant, i meant he should start up a dozen of sandboxes, and students take them aways, so he starts a new one22:17
ronnyarigato: its just so the startup cost is paid before the students send code22:18
nedbatronny: pre-started, with a queue of ready-to-go processes.22:18
apanda (~apanda@c-67-170-53-221.hsd1.wa.comcast.net) joined #pypy.22:18
ronnynedbat: thats what i mean22:18
nedbatronny: and what would the sandbox proc block on to wait for the code to run?22:20
ronnynedbat: well, you would have to design that22:22
ronnynedbat: i think execnet might be a good tool to manage that22:23
robertj (~Rob@97-81-71-34.dhcp.athn.ga.charter.com) joined #pypy.22:24
nedbatronny: but the sandbox doesn't allow network connections yet...22:25
amaury__ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 248 seconds22:25
robertjooh got here just in time for my question about sandboxing :P22:25
nedbatrobertj: what are you going to do with the sandbox?22:26
robertjare there any facilities to track memory usage back to some kind of execution context?22:26
robertji'd like to let users run untrusted code but need a way to limit memory usage22:27
robertjand i know the typical answer is 'don't do that that!'22:27
nedbatrobertj: you can limit the sandbox to a certain amount of memory.  But it's the entire sandbox that is limited.22:27
ronnynedbat: execnet can start slaves via popen22:27
ronnynedbat: execnet basically starts slave processes via a proces opening facility (like open/ssh), then you can send code fragments over there22:28
arigatoin this case, it looks a bit overkill22:30
nedbatarigato: are you talking about execnet?22:30
arigatoyou can use whatever system calls the sandboxed process think it is doing (and which is handled by the outer process) to introduce delays22:30
arigatonedbat: yes22:30
arigatoe.g. you run the subprocess until it asks to load "x.py", then you wait for the actual code from the student to show up, and you give it immediately as an answer to this "x.py" request22:31
nedbatarigato: you mean via clever implementation in the ll_os functions in sandlib.py?22:32
arigatothen it's just a matter of organizing the subprocess so that it loads all needed modules first, and then does "import x"22:32
arigatoyes22:32
nedbatarigato: right: I'd have ten subprocesses all blcoked on 22:33
nedbatall blocked on "import exercise"22:33
arigatoyes, or "execfile('exercise.py')"22:33
bbot24Failure: 15http://buildbot.pypy.org/builders/own-win-x86-64/builds/69 [12ctismer, win64-stage1]22:33
arigatoit's hackish of course, but it should work22:33
arigatowe're missing a better sandlib.py that would be developed as a real interface handling such hacks transparently22:34
nedbatarigato: I've already got "import exercise" in place, and I've made one small change to sandlib.py, so it's within reach.22:35
arigato:-)22:35
nedbatso it sounds like the only way to speed startup of the sandbox is to reduce the number of faked system calls it has to endure.22:36
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 240 seconds22:37
ronnyarigato: wrt sandbox desgn, would it make sense to have direct imports for parts of sys.path22:37
ronny(so the sandbox would actually use direct python code loading)22:38
arigato(...which is also true btw of the real pypy (and the real CPython): they do tons of system calls, particularly at start-up, which has a cost)22:38
arigatoronny: unsure what you're thinking22:39
nedbatarigato: well, i noticed the CPU L2 cache warning, and that seems to be an optimization effort, which I don't need in this case.22:39
arigatonedbat: yes22:39
ronnyarigato: vasically stuff like the stdlib doesnt have to be loaded via stdin marshaling, it could be loaded with direct calls22:39
ronnyarigato: even more so, propper pure python packages in sys.path could be loaded directly as well if acceptable22:39
arigatoronny: true, but I'll leave it as an exercise to the reader to implement the correct logic in rsandbox.py22:40
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/15122:40
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/95622:40
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/72522:40
ronnyheh22:40
arigatowhich is very far from "sys.path" considerations22:40
ronnynedbat: mind to have your students fix that?22:40
ronny(thay are manpower)22:40
arigatoronny: I was mostly joking, actually22:41
ronnywell, that one will be on the students (maybe?)22:41
arigatowhat I mean is: it's a non-trivial problem, and I don't want to be relied upon to come with a solution22:41
nedbatronny: it weakens the design of the sandbox to have tricky logic that decides some imports are safe.22:42
arigatodefinitely22:42
ronnynedbat: the basc idea is to have a rpython based import hook, which loads precompiled code directly from save locations instead of reading it via the proxy logic22:46
ronnyarigato: btw, would it make sense for pypy to have some extra logic to have those nasty stat cascades for import path finding to be gone22:47
arigatoronny: well you can also translate a version of pypy with some modules pre-imported, if you really care about the load time22:48
arigatoronny: extra logic: feel free22:48
ronnyarigato: the basic iea is to have a sandbo thats able to do direct loading from normal install paths (just forbid wrting)22:49
nedbatthese are fascinating. i'm not quite in a place to take advantage of them, though.22:50
ronnyarigato: and wrt import cascades, im just pissed off that python sometimes does about 80-90 stat calls to get stuff22:50
fprimex (~fprimex@fl-67-233-144-228.dhcp.embarqhsd.net) left irc: Quit: Leaving22:50
ronnyim love with my personal one true golden rule - to be fast - do less22:51
kkris1 (~kris@93-82-47-186.adsl.highway.telekom.at) left irc: Quit: Leaving.22:51
arigatomy own golden rule is: less code is better, unless (1) you really need it to implement what you want, or (2) it really makes it faster22:54
ronnyarigato: kinda true as well, maintenance is a bitch22:55
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) joined #pypy.22:55
arigatoimporting logic is anyway beyond all hopes of being even remotely simple code, but if you feel that you can reduce the number of stat() without increasing complexity too much, feel free22:57
arigatofor my part I'm stopping the consideration at "beyond all hopes of being even remotely simple"22:58
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 248 seconds22:59
ronnyarigato: basically it needs easyly accessible knowdegde about paths23:04
CIA-6803schmichael 07roundup * 10#942/CPython 2.7 faster than PyPy 1.7 in select/socket/ctypes script: 23:08
CIA-68[new] (Only works in Linux... may work in OSX with the socket type changed.)23:08
CIA-68Running "punish_logd.py 1 25000000 log.sock 0" returns the followi ... * 14https://bugs.pypy.org/issue94223:08
Nick change: elliottcable -> mr_fuggles23:09
CIA-6803schmichael 07roundup * 10#942/CPython 2.7 faster than PyPy 1.7 in select/socket/ctypes script: 23:09
CIA-68[chatting] Attach simplelogd.py's config file (pass the filename to simplelogd.py as the23:09
CIA-68only command line argument). * 14https://bugs.pypy.org/issue94223:09
CIA-6803schmichael 07roundup * 10#942/CPython 2.7 faster than PyPy 1.7 in select/socket/ctypes script: 23:12
CIA-68Attach script to test simplelogd throughput. simplelogd appears to be the23:12
CIA-68bottleneck, so I always run this with CPython 2.7 and 1 concurrency (m ... * 14https://bugs.pypy.org/issue94223:12
sunoano (~sa@unaffiliated/sunoano) joined #pypy.23:14
Nick change: mr_fuggles -> ec23:19
kenaan12hager ppc-jit-backend 11fcdbc05a8222 15/pypy/jit/backend/ppc/ppcgen/opassembler.py: Use safe way for relative jumps23:26
kenaan12hager ppc-jit-backend 1162d08e2b17fc 15/pypy/jit/backend/ppc/ppcgen/: Add ugly way to avoid problems with holes in MANAGED_REGS23:26
hruske (~Gasper@2001:15c0:66e9:14:383e:18ff:fe7f:6c95) left irc: Remote host closed the connection23:30
kenaan12boemmels lang-scheme 1129be9de3ed49 15/scheme/object.py: Use __class__.__name__ in __repr__  Simplifies debugging23:35
kenaan12boemmels lang-scheme 116ecdb8380916 15/scheme/test/test_scheme_level.py: New test: Run and assert tests at scheme level  This test creates an Scheme level assert macro, which passes...23:35
kenaan12boemmels lang-scheme 115a01480a6507 15/scheme/: Implemented external representaions of objects and write function23:36
hruske (~Gasper@2001:15c0:66e9:14:f82f:18ff:fe84:c6d2) joined #pypy.23:39
arigato (~arigo@82.113.98.65) left irc: Quit: See you23:42
zooko (~user@97-118-101-86.hlrn.qwest.net) left irc: Ping timeout: 276 seconds23:44
nedbat (~nedbat@python/psf/nedbat) joined #pypy.23:45
--- Sun Nov 27 201100:00

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