#pypy IRC log for Tuesday, 2011-02-01

MrBlueSky_ (~DarkScyth@rrcs-70-63-94-21.midsouth.biz.rr.com) joined #pypy.00:02
jacob22 (~jacob@CPE0015f2633f3a-CM00159a662342.cpe.net.cable.rogers.com) joined #pypy.00:02
nettok (~quassel@ joined #pypy.00:05
gonsor (~quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.00:06
MrBlueSky (~DarkScyth@rrcs-70-63-94-21.midsouth.biz.rr.com) left irc: Ping timeout: 276 seconds00:06
nettok_ (~quassel@ left irc: Ping timeout: 276 seconds00:09
stakkars_ (~tismer@ joined #pypy.00:23
derdon (~derdon@p54A6F11C.dip.t-dialin.net) left irc: Remote host closed the connection00:26
Nick change: Varriount -> Varraway00:26
stakkars__ (~tismer@77-21-85-35-dynip.superkabel.de) joined #pypy.00:29
stakkars_ (~tismer@ left irc: Ping timeout: 240 seconds00:33
fenrrir (~fenrrir@ left irc: Ping timeout: 250 seconds00:34
tmu (~tmu@evil.oulu.nixu.fi) left irc: Ping timeout: 276 seconds00:35
qbprogerAlex_Gaynor: ping00:36
qbprogermissed you earlier :\00:36
JeremyJohnson (~j@S01066cf049e83ae2.vc.shawcable.net) joined #pypy.00:45
Ademan (~dan@adsl-71-141-238-143.dsl.snfc21.pacbell.net) left irc: Quit: leaving00:49
Trundle (~andy@python/site-packages/trundle) left irc: Remote host closed the connection00:54
fmilo (~mist0@ left irc: Quit: fmilo00:56
nhamann (4c10291e@gateway/web/freenode/ip. joined #pypy.01:15
tmu (~tmu@evil.oulu.nixu.fi) joined #pypy.01:16
fmilo (~mist0@ joined #pypy.01:20
jacob22 (~jacob@CPE0015f2633f3a-CM00159a662342.cpe.net.cable.rogers.com) left irc: Remote host closed the connection01:22
jacob22 (~jacob@CPE0015f2633f3a-CM00159a662342.cpe.net.cable.rogers.com) joined #pypy.01:22
getxsick (xsx@unaffiliated/getxsick) left irc: Quit: leaving01:25
stakkars__ (~tismer@77-21-85-35-dynip.superkabel.de) left irc: Quit: schnarch01:41
DasIch_ (~DasIch@p5DC5F2DF.dip.t-dialin.net) left irc: Ping timeout: 240 seconds01:58
nhamann (4c10291e@gateway/web/freenode/ip. left irc: Ping timeout: 265 seconds01:59
fmilo (~mist0@ left irc: Quit: fmilo02:05
silanus_ (~silanus@p5DDEBFA7.dip.t-dialin.net) left irc: Read error: Operation timed out02:09
silanus (~silanus@p5DDE9ACC.dip.t-dialin.net) joined #pypy.02:18
Alex_Gaynorqbproger: pong02:32
qbprogerAlex_Gaynor: is there anything that would be good to benchmark for pypy?02:47
qbprogeri'm working on a benchmark framework02:48
Alex_GaynorI don't have any particular suggestions ATM02:49
qbprogerAlex_Gaynor: alright, I may just go with project euler (primes, totient function, etc)02:51
Alex_Gaynoramaury (for the logs): Based on extreme gdb/pdb sessions I've found out that the issue isn't cookie.pack (not yet at least), the issue is cookie has teh wrong data, which appears to come from self.decoded_chars_used02:53
Nick change: Varraway -> Varriount03:15
sphere_ (~siddharth@ joined #pypy.03:31
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) left irc: Ping timeout: 250 seconds03:35
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.03:37
qbproger_ (~qbproger@ joined #pypy.03:45
qbproger (~qbproger@ left irc: Ping timeout: 264 seconds03:49
Alex_Gaynorok I need a break from this http://paste.pocoo.org/show/330377/03:52
DasIch (~DasIch@p5DC5FDAE.dip.t-dialin.net) joined #pypy.03:53
Arfrever (~Arfrever@gentoo/developer/Arfrever) left irc: Quit: Ex+re (KVIrc 4)03:55
qbproger_ (~qbproger@ left irc: Quit: Leaving04:00
nettok_ (~quassel@ joined #pypy.04:04
nettok (~quassel@ left irc: Ping timeout: 240 seconds04:04
jacob22 (~jacob@CPE0015f2633f3a-CM00159a662342.cpe.net.cable.rogers.com) left irc: Ping timeout: 260 seconds04:19
gonsor (~quassel@kel30.kel.stud.uni-goettingen.de) left irc: Read error: Connection reset by peer04:20
ssbr (~ssbr@gw-wifi4.cdf.utoronto.ca) left irc: Ping timeout: 240 seconds04:23
panni_ (hannes@ip-178-202-69-162.unitymediagroup.de) left irc: Read error: Connection reset by peer04:29
sphere_ (~siddharth@ left irc: Quit: sphere_04:34
Nick change: Varriount -> Varraway04:54
sphere_ (~siddharth@ joined #pypy.04:54
ssbr (~ssbr@gw-wifi4.cdf.toronto.edu) joined #pypy.05:06
gutworth (~benjamin@64-131-13-109.usfamily.net) left irc: Ping timeout: 240 seconds05:16
bobbyz (~bobbyz@c-24-14-151-193.hsd1.il.comcast.net) left irc: Ping timeout: 264 seconds05:23
gutworth (~benjamin@64-131-13-109.usfamily.net) joined #pypy.05:25
gutworth (~benjamin@64-131-13-109.usfamily.net) left irc: Ping timeout: 272 seconds05:31
mrdeep_ (~mrdeep@pool-71-164-66-192.albyny.east.verizon.net) joined #pypy.05:33
mrdeep (~mrdeep@pool-71-164-66-192.albyny.east.verizon.net) left irc: Read error: Connection reset by peer05:33
Nick change: mrdeep_ -> mrdeep05:33
bobbyz (~bobbyz@c-24-14-151-193.hsd1.il.comcast.net) joined #pypy.05:36
htoothrot (~mux@66-169-185-121.dhcp.ftwo.tx.charter.com) left irc: Ping timeout: 272 seconds05:48
htoothrot (~mux@66-169-185-121.dhcp.ftwo.tx.charter.com) joined #pypy.05:50
Nick change: Scriptor -> Scriptor|away05:56
Nick change: Varraway -> Varriount06:00
Nick change: Varriount -> Varraway06:09
Nick change: Varraway -> Varriount06:13
verte (~verte@python/site-packages/verte) joined #pypy.06:21
ssbr (~ssbr@gw-wifi4.cdf.toronto.edu) left irc: Ping timeout: 240 seconds06:26
Gromit (~bear@server00.klix.ch) joined #pypy.06:33
fijal (~fijal@41-134-41-226.dsl.mweb.co.za) joined #pypy.06:39
Nick change: Varriount -> Varraway06:41
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) left irc: Ping timeout: 240 seconds06:51
ssbr_ (ssbr@173-230-191-202.cable.teksavvy.com) joined #pypy.06:54
ssbr__ (ssbr@173-230-191-202.cable.teksavvy.com) joined #pypy.07:02
verte (~verte@python/site-packages/verte) left irc: Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org07:04
whyking (~quassel@p5B39DFC1.dip.t-dialin.net) joined #pypy.07:04
ssbr_ (ssbr@173-230-191-202.cable.teksavvy.com) left irc: Ping timeout: 264 seconds07:05
nettok (~quassel@ joined #pypy.07:11
kenaan03fijal 10r8026114 07M(pypy/extradoc/planning/jit.txt): This is done in jitviewer ...07:13
nettok_ (~quassel@ left irc: Ping timeout: 240 seconds07:14
kenaan03fijal 10r8026214 07M(pypy/extradoc/planning/jit.txt): Kill a redundant task. We have other one, more precise below ...07:15
kenaan03fijal 10r8026314 07M(pypy/extradoc/planning/jit.txt): A comment ...07:15
kenaan12fijal default 11ec46f2017ef8 15/pypy/config/pypyoption.py: Before anyone wakes up, disable geninterp. Will run benchmarks and see07:18
kenaan12fijal default 112f12f4b023ab 15/pypy/module/pypyjit/test/test_pypy_c_new.py: I don't want to, but hg is making me commit this work in progress07:18
kenaan12fijal default 110ec8b9b67a09 15/pypy/config/pypyoption.py: merge heads07:18
zejn (~Gasper@ joined #pypy.07:19
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.07:24
brtzsnr (~alexandru@ left irc: Ping timeout: 240 seconds07:26
ssbr_ (ssbr@173-230-191-202.cable.teksavvy.com) joined #pypy.07:28
ssbr__ (ssbr@173-230-191-202.cable.teksavvy.com) left irc: Ping timeout: 260 seconds07:29
zk (identsucks@whatit.is) joined #pypy.07:36
zk (identsucks@whatit.is) left #pypy.07:37
zk (identsucks@whatit.is) joined #pypy.07:38
Nick change: necktone -> chintone07:41
antocuni (~antocuni@host34-120-dynamic.11-79-r.retail.telecomitalia.it) joined #pypy.07:50
fijalcan't log in to tannit07:51
fijalantocuni: can you?07:52
Action: antocuni tries07:52
antocuniseems not07:52
JeremyJohnson (~j@S01066cf049e83ae2.vc.shawcable.net) left irc: Quit: Leaving07:52
fijalantocuni: I found more nonsense in socket.py07:53
fijalin init of socket07:54
fijalyou store all delegated methods07:54
fijaland in close you also store them07:54
fijalso say send07:54
fijalis not a method, it's a function on an instance07:54
fijalthat makes a difference if you create lots of sockets07:54
fijal(like http)07:54
antocuniare we talking about interp_socket.py?07:55
antocuniah, I see07:56
bivab (~david@ip-88-152-254-229.unitymediagroup.de) joined #pypy.07:56
antocuniwell, I suppose it makes some kind of sense for "normal" python07:56
fijalit's confusion for no reason07:57
fijalto start with07:57
antocuniwhy not? Sometimes you don't want to inherit07:57
antocunidelegation is a well known pattern07:57
fijalyeah, but storing methods on instance is a bit pointless07:57
fijaldelegation is fine07:57
fijalwhat's wrong with07:57
antocunithe only difference here is that you save a bit of typing07:57
fijaldef send(self, data): if self.closed: return foo else: self._sock.send(data)07:58
fijalyou can do metaprogramming if you're *that* concerned07:58
fijalbut you pay for it even on CPython07:58
antocuniah wait07:58
antocuniindeed, I didn't see that close() rebinds them07:58
antocunithen yes, I agree that it's confusing07:58
fijalseriously I'm down to the point where I'm willing to type it all08:00
fijalit's not *that* much typing08:00
antocuniI suppose it's fine if you just do it08:00
antocunihi ronny08:00
fijalsomething's wrong on tannit08:06
bivab (~david@ip-88-152-254-229.unitymediagroup.de) left irc: Remote host closed the connection08:11
fijalalthough buildbot seems to run08:13
antocunifijal: tannit answers to pings08:14
fijalrun ssh -v see what happens08:14
antocuniit "just" hangs on ssh08:14
fijalyou almost get the prompt08:14
fijalseems like /bin/bash is broken or something08:15
antocuniyes, saw it08:15
diracdelta (~christoph@c-67-188-126-146.hsd1.ca.comcast.net) joined #pypy.08:21
ErKa (keryell@m3.wifi.enstb.org) joined #pypy.08:21
nettok (~quassel@ left irc: Ping timeout: 272 seconds08:27
antocuni (~antocuni@host34-120-dynamic.11-79-r.retail.telecomitalia.it) left irc: Ping timeout: 276 seconds08:28
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) left irc: Read error: Connection reset by peer08:30
hpk (~hpk@HSI-KBW-109-193-198-175.hsi7.kabel-badenwuerttemberg.de) joined #pypy.08:32
amaury_ (5bc3480e@gateway/web/freenode/ip. joined #pypy.09:17
bivab (~david@ip-88-152-254-229.unitymediagroup.de) joined #pypy.09:30
Ademan (~dan@nat-sonicnet.noisebridge.net) joined #pypy.09:31
antocuni (~antocuni@host34-120-dynamic.11-79-r.retail.telecomitalia.it) joined #pypy.09:41
arigato (~arigo@fwstups.cs.uni-duesseldorf.de) joined #pypy.10:06
fijalarigato: hi10:07
fijalarigato: pyflate-fast got 25% slower10:10
fijalalso, removing geninterp seems to be a win overall10:11
amaury_fijal: can you check whether tests still pass with geninterp=False?10:15
amaury_IIRC there were issue with os.environ10:15
fijalwhich tests?10:15
amaury_try to change an environment variable, and see if pypy correctly loads the environment10:16
amaury_last time I translated with -O0, the environment was captured at translation time10:16
Shinato (~John@osbk-4db06e66.pool.mediaWays.net) joined #pypy.10:20
Kaskuka (~John@osbk-4db06d7d.pool.mediaWays.net) left irc: Ping timeout: 240 seconds10:20
fijalfijal@hans:~/Downloads/pypy-c-jit-41512-88c9d8b65c47-linux64$ FOO=3 ./bin/pypy -c 'import os; print os.environ["FOO"]'10:21
fijalamaury_: that's what you asked about?10:21
amaury_so it works :-)10:21
fijalthe question is now, if tests pass (or fail not less than usual), can we just remove geninterp?10:22
fijalor it's there, why remove it?10:22
amaury_I'm not sure to understand geninterp=False10:29
amaury_what will happen to compiled applevel code?10:29
fenrrir (~fenrrir@ joined #pypy.10:30
antocuniamaury_: already now not all the compiled applevel code is geninterpreted10:31
antocunie.g., __builtin__/app_io.py is marked as "NOT_RPYTHON", which effectively means "don't use geninterp on this"10:32
amaury_so this code is embedded as is in the program10:33
amaury_and run as normal Python code?10:33
lac people still having problems logging into tannit?10:34
antocunilac: yes, it's still not working for me10:35
lacok, will send mail to get it fixed.  you think that its a problem with bash?10:35
antocunilac: fijal thinks so10:35
fijal'I said it's a possible explanation10:35
antocunibut indeed, running ssh -vvv looks like the authentication is successful, then something hangs10:36
fijalI guess it's something after you run bash because it captures Ctrl-C10:36
arigatoI would say the hard drive died, but it should not really be that easy with the setup of two hard drives in parallel10:38
fijalarigato: buildbot run managed to run and finish10:38
fijalso I don't think it's an  HD10:38
fijalit's hard to disprove based on that10:39
lacwell, there are people in the office, though they might be at lunch now.10:39
lacthere was no company wide mail about a fire in the machine room, or any such.10:40
antocunibtw, if I run "ssh tannit ls", it still hangs but CTRL-C works10:40
arigatoalso, macmini-mvt is down10:43
zejn (~Gasper@ left irc: Quit: Leaving10:44
fijalany opinions on removing geninterp?10:46
lizardo (~lizardo@ joined #pypy.10:51
arigatofijal: we should first check which places still use it, and think if it might cause slowdowns10:51
fijalit did cause 10% slowdown on spitfire10:52
fijalI'll look at that10:52
fijalarigato: will you look at pyflate-fast?10:52
arigatoI mean, look not just by looking at benchmark results10:52
fijalwe did look a while ago10:53
fijalat least once10:53
fijal"which parts are in applevel but they potentially might be in interp level"10:53
arigatoI mean looking at all app_*.py files and at the appexec or similar calls that end up in geninterp10:53
fijalyes, we did10:53
fijalI can do it again though10:53
arigatowell it was not done correctly if there is a 10% slowdown somewhere10:54
fijalthere is 25% slowdown on pyflate fast10:54
fijalcoming few days ago10:54
fijalwhen you merged jit-longlong10:54
fijal(is likely unrelated though)10:55
Action: arigato lunch10:55
fijalwhat I'm saying is we should have maybe some uniform policy on looking at slowdowns10:55
arigatoI see the point10:55
arigatoI'm mostly recovered from the sprint but not quite10:55
fijalI see :)10:56
fijalwon't bother you then10:56
ErKa (keryell@m3.wifi.enstb.org) left irc: Ping timeout: 250 seconds10:58
fenrrir (~fenrrir@ left irc: Ping timeout: 255 seconds11:06
diracdelta (~christoph@c-67-188-126-146.hsd1.ca.comcast.net) left irc: Quit: diracdelta11:18
brtzsnr (~alexandru@ joined #pypy.11:32
fijalantocuni: ping?11:34
diracdelta (~christoph@c-67-188-126-146.hsd1.ca.comcast.net) joined #pypy.11:37
Ademan (~dan@nat-sonicnet.noisebridge.net) left irc: Quit: leaving11:38
fijalantocuni: I have a problem with pdb11:39
fijalyou don't-want-to-know-what11:39
arigatoI guess I need to understand and use pthread_atfork()11:45
fijalstill fighting with cannot find gc roots?11:46
arigatonot really "fighting" per se11:46
diracdelta (~christoph@c-67-188-126-146.hsd1.ca.comcast.net) left irc: Quit: diracdelta11:48
amaury_arigato: we also have fork hooks11:48
amaury_in module/posix/interp_posix.py11:48
kenaan12fijal default 1179a284629147 15/lib-python/modified-2.7.0/socket.py: Cleanup socket. Added 10 lines, but it's so much cleaner by now11:48
lactannit had a kernel panic11:49
laciko has rebooted it now11:49
amaury_fijal: _delegate_methods is used in ssl.py12:02
arigatoamaury_: right, it's exactly the same as pthread_atfork(), but done portably12:04
fijalamaury_: ah, ok12:05
amaury_fijal: fortunately, the same cleanup works there as well12:05
fijalamaury_: well, I suppose good :)12:12
fijalwe don't need an obscure hack any more12:12
kenaan12fijal default 11f7e470d7823c 15/: Fix ssl.py and test_ssl.py (by chance, one more test passing on nightly run as well)12:12
fijalno no no no12:13
fijalcan I back out changes to specific files?12:13
derdon (~derdon@p54A6EB4B.dip.t-dialin.net) joined #pypy.12:13
antocunifijal: yes?12:13
fijalantocuni: I can't set PYTHONPATH to fancycompleter or pdb12:14
antocunifijal: can you be more precise please? I do use them by putting their dirs in PYTHONPATH12:15
fijalthen you can't import 'test'12:15
fijalfrom stdlib12:15
fijalbecause that imports test from those dirs12:16
antocunifijal: ah indeed, you're right12:16
antocuniin fact, I don't have the dirs in pythonpath12:16
antocuniI have symlinks to another dir12:16
antocunifijal: well, I'd not say that it's a "problem"12:17
fijalcan you rename it to testing?12:17
kenaan12fijal default 110872e654e4d7 15/: Backed out changeset f7e470d7823c12:17
kenaan12fijal default 11177524b5d89d 15/: Backed out changeset 0872e654e4d712:17
antocuniwhy can't you just install them?12:17
fijalthat's a noop12:17
fijalantocuni: I can12:17
amaury_fijal: huh?12:17
fijalamaury_: huh what?12:18
amaury_you backed out the backed out change12:18
fijalbecause I can't do what I want12:18
fijalI want to backout only specific files12:18
fijaland not the whole changeset12:18
amaury_"hg help backout" shows that you can add file names12:19
fijalbut I can't make it run12:19
amaury_but a --nocommit would be better12:19
fijalthat's what I'm missing12:19
amaury_me too12:20
amaury_it does not exist12:20
fijalah :)12:20
kenaan12fijal default 11815cadb9214e 15/pypy/: Proper backout12:20
fijalamaury_: there is a hack12:20
kenaan12fijal default 1159399209047b 15/pypy/: merge12:20
fijalyou remove commit message12:20
fijalyou close editor12:20
fijalit crashes "abort, no commit message"12:20
fijaland you have your things modified12:20
arigato"hg pull -u" is rather useless, because it won't update if it doesn't also pull at least one change12:22
amaury_how do you resolve conflicts with hg?12:27
arigatohg resolve --list12:27
amaury_does it open a vi-like windows with 3 buffers?12:27
arigatoit can, if you have some extension installed12:27
antocuniarigato: I think it does by default12:27
amaury_this seems to be the default on tannit12:27
amaury_but I don't want it12:28
antocuniI use "[ui] merge = internal:merge" in my hgrc12:28
antocunithis way, I get svn-style conflict markers12:28
arigatoI get svn-style conflict markers too, but I didn't do anything12:28
arigatomaybe because I compiled hg from sources so I don't have the 3-window editor at all?12:29
amaury_I noticed that on my Windows box, it spawns the P4merge graphical tool12:29
antocuniarigato: the 3-window editor is really vim, AFAIK12:30
amaury_(I happen to have a Perforce client installed)12:30
arigatoantocuni: no clue then12:30
antocunimaybe it's because of some combination of hgrc settings and env variables12:30
antocunianyway, I'm more than happy with internal:merge now12:31
arigatoI'll add "merge = internal:merge" to my .hgrc to be on the safe side :-)12:31
fijal (~fijal@41-134-41-226.dsl.mweb.co.za) left irc: Remote host closed the connection12:33
arigatofrom the man page:12:40
arigato"""       If  attr  is  used to create multiple threads, then the caller must change the stack12:40
arigato       address attribute between calls to pthread_create(3); otherwise,  the  threads  will12:40
arigato       attempt to use the same memory area for their stacks, and chaos will ensue.12:40
zejn (~Gasper@ joined #pypy.12:47
mvt (~mvantelli@ joined #pypy.12:48
zejn (Gasper@ left #pypy.12:48
lacaccording to http://hgbook.red-bean.com/read/a-tour-of-mercurial-merging-work.html12:54
lacif you have a favourite merge tool12:54
lacyou canmake mercurial use it by setting the HGMERGE environment variable12:54
lacah http://mercurial.selenic.com/wiki/MergeProgram#How_Mercurial_decides_which_merge_program_to_use12:57
lacthey go down a list looking for candidates to use.  12:58
gutworth (~benjamin@64-131-13-109.usfamily.net) joined #pypy.13:24
gutworth (~benjamin@64-131-13-109.usfamily.net) left irc: Read error: Connection reset by peer13:30
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.13:35
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.13:36
ErKa (keryell@m3.wifi.enstb.org) joined #pypy.13:46
elmom_ (~elmom@hy-ovpn1-104.vpn.helsinki.fi) joined #pypy.13:53
kenaan12arigo default 11cf0bdb323fec 15/pypy/config/pypyoption.py: Re-enable geninterp.  For me, running "translate.py -O2" crashes without it.13:58
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.14:02
fijal (~fijal@ joined #pypy.14:02
fijalarigato: ping14:06
fijalwhat's the thing with -O2?14:06
arigatoFor me, running "translate.py -O2" crashes without geninterp14:07
fijalok, but how?14:07
fijaland why?14:07
fijaland how -Ojit does not?14:07
fijalit seems very weird that -O0 and -Ojit works14:07
arigatoI don't know, I'm working on some other problem right now and I need a working translation14:07
fijalfair enough :)14:08
arigato[translation:ERROR]  Exception: <Cache 'ApplevelCache' (4 items)> recursive building of <ApplevelClass filename='/hom14:08
arigatoe/arigo/hg/pypy/trunk/pypy/module/__builtin__/app_file_stub.py' can_use_geninterp=False>14:08
fijalok, got it14:08
fijalfor what is worth app_file_stub starts with NOT_RPYTHON14:10
arigatono clue really14:10
fijalI'll investigate14:10
fijalfor one I wonder if __filestub is used from anywhere14:16
amaury_not since r3419214:18
Alex_Gaynoramaury_: did you see the debugging I got done yesterday?14:20
kenaan12fijal default 11a7a3f5f337c4 15/pypy/module/__builtin__/__init__.py: Remove filestub. nobody uses it14:22
kenaan12fijal default 11aa1a6d5a1ef1 15/pypy/config/pypyoption.py: fixed issue, redisable geninterp14:22
fijalarigato: I hope it's fixed by now14:23
arigatothank you14:23
fijalit's a good cleanup14:24
fijalI'm anyway leaving, feel free to reenable it if you run into problems14:24
amaury_fijal: did you remove app_file_stub.py?14:24
fijalwill look it up when I'm back14:24
fijalamaury_: yes14:24
amaury_the pypy-svn message does not show it14:25
fijalapparently no :/14:25
kenaan12fijal default 11a46f6fc27ded 15/pypy/module/__builtin__/app_file_stub.py: Remove app_file_stub.py14:25
amaury_neither does the irc message14:25
amaury_ah ok :-)14:25
Alex_Gaynoris float('nan') RPython?14:28
fijalbecause it's constant folded14:28
Alex_Gaynoris there some better way to do it?14:29
amaury_Alex_Gaynor: rarithmetic.NAN14:29
kursor (~kursor@p57BB1816.dip0.t-ipconnect.de) joined #pypy.14:30
amaury_because float('inf') does not work with Python2.514:30
elmom_ (~elmom@hy-ovpn1-104.vpn.helsinki.fi) left irc: Ping timeout: 260 seconds14:39
sphere_ (~siddharth@ left irc: Quit: sphere_14:40
jacob22 (~jacob@CPE0015f2633f3a-CM00159a662342.cpe.net.cable.rogers.com) joined #pypy.14:49
fijalarigato: -O2 compiles14:49
MrBlueSky_ (~DarkScyth@rrcs-70-63-94-21.midsouth.biz.rr.com) left irc: Read error: Connection reset by peer14:49
gonsor (~quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.14:49
MrBlueSky (~DarkScyth@rrcs-70-63-94-21.midsouth.biz.rr.com) joined #pypy.14:50
kenaan12alex_gaynor default 1157363f9b7479 15/pypy/objspace/std/: Negative values can be raised to the nan power.14:51
kenaan12alex_gaynor default 1113c0dc295d74 15/pypy/module/__builtin__/app_file_stub.py: Merged upstream.14:51
Action: arigato writes a scary test function using threads, fork(), and __del__s14:51
fijalarigato: I run away14:51
Alex_Gaynorarigato: fun, reid kleckner had to do such things for Unladen Swallow14:52
elmom_ (~elmom@hy-ovpn2-82.vpn.helsinki.fi) joined #pypy.14:52
fijalfor fun and amusement, run this under pypy14:52
fijalsee you14:52
fijal (~fijal@ left irc: Quit: Leaving14:52
arigatonow that I think about it14:53
arigatoCPython completely leaks memory when doing fork() with threads14:54
arigatowhen calling fork(), any object with a reference from the stack of another thread becomes dead but won't be freed14:54
amaury_arigato: this is probably not a problem14:57
amaury_IIUC there are two common usages of fork()14:58
amaury_a server that forks to handle requests: the child dies quickly15:00
amaury_or to spawn a new process: exec() will release memory anyway15:00
arigatoand then there is the fork() in multiprocessing, which does not really enter these two categories15:01
sphere_ (~siddharth@ joined #pypy.15:02
amaury_multiprocessing is a monster IMO15:03
amaury_it mixes threads and fork15:03
amaury_which don't play well together15:03
arigatoagreed, but that's why I am trying to fix things so that at least they don't end up in a "cannot find gc roots!"15:03
Alex_Gaynoris the buildbot down?15:17
jacob22tannit had to be rebooted earlier today. That may have affected something.15:18
Alex_Gaynorjacob22: nope, it was just my wifi dropping :(15:18
arigatogood, now I have a test that fails cleanly as expected for asmgcc+threads+fork15:19
derdon (~derdon@p54A6EB4B.dip.t-dialin.net) left irc: Remote host closed the connection15:20
ErKa (keryell@m3.wifi.enstb.org) left irc: Ping timeout: 264 seconds15:29
jbaker__ (~jimbaker@c-24-8-39-78.hsd1.co.comcast.net) left irc: Quit: jbaker__15:34
gonsor (~quassel@kel30.kel.stud.uni-goettingen.de) left irc: Remote host closed the connection15:36
jbaker_ (~jimbaker@c-24-8-39-78.hsd1.co.comcast.net) joined #pypy.15:36
jacob22 (~jacob@CPE0015f2633f3a-CM00159a662342.cpe.net.cable.rogers.com) left irc: Ping timeout: 250 seconds15:37
jbaker_ (~jimbaker@c-24-8-39-78.hsd1.co.comcast.net) left irc: Quit: jbaker_15:42
jbaker_ (~jimbaker@c-24-8-39-78.hsd1.co.comcast.net) joined #pypy.15:43
jacob22 (~jacob@CPE0015f2633f3a-CM00159a662342.cpe.net.cable.rogers.com) joined #pypy.15:46
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) left irc: Ping timeout: 242 seconds15:51
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.15:51
Alex_Gaynoramaury_: I'm not pretty confident that the bug occurs within tell_w() (or something it calls) and isnt't in either __init__, read(), or incremental newline decoder16:01
amaury_it's probably pdb time :-/16:04
gonsor (~quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.16:04
Alex_Gaynoramaury_: I've been doing parallel pdb/gdb sessions for a while16:04
Arfrever (~Arfrever@gentoo/developer/Arfrever) joined #pypy.16:05
amaury_didn't you write an applevel test for module/_io?16:07
kkris (~kris@188-22-127-159.adsl.highway.telekom.at) joined #pypy.16:07
Alex_Gaynoramaury_: well I've got a test script, I haven't committed it because it isn't really a fully reduced test case16:08
idnaroh hey, http://morepypy.blogspot.com/2011/01/jit-backend-for-arm-processors.html is awesome16:09
idnar(although it's a pity the Android NDK is so awful)16:10
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.16:11
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.16:19
Alex_Gaynorok I need abreak from this.  There is a bug in tell_w, I don't know what it is16:22
jbaker_ (~jimbaker@c-24-8-39-78.hsd1.co.comcast.net) left irc: Quit: jbaker_16:23
Nick change: Varraway -> Varriount16:23
arigatoAlex_Gaynor: you're not talking about W_TextIOWrapper.tell_w?16:24
Alex_Gaynorarigato: yes16:24
Alex_Gaynorarigato: and I've already fixed pack() to not overflow ints, so it isn't that16:24
arigatoah ok16:24
jimbaker_ (~jbaker@c-24-8-39-78.hsd1.co.comcast.net) joined #pypy.16:24
jimbaker_ (~jbaker@c-24-8-39-78.hsd1.co.comcast.net) left irc: Client Quit16:25
jimbaker (~jbaker@c-24-8-39-78.hsd1.co.comcast.net) joined #pypy.16:25
amaury_Alex_Gaynor: can you submit your changes to PositionCookie?16:26
Alex_Gaynoramaury_: let me clean up my pwd a tad, then I'll commit it16:26
mvt (~mvantelli@ left irc: Quit: Leaving16:27
kenaan12alex_gaynor default 113f6b33b21cc9 15/pypy/: Try to get test_seek_and_tell from lib-python test_io passing.16:33
Alex_Gaynoramaury_: there it is16:33
derdon (~derdon@p54A6F053.dip.t-dialin.net) joined #pypy.16:35
ousado (~ousado@188-192-241-129-dynip.superkabel.de) left irc: Changing host16:36
ousado (~ousado@unaffiliated/ousado) joined #pypy.16:36
arigatohow I would love a pthread function that, given a pointer to anything in the stack, returns the id of the thread owning the stack16:44
arigatoor even just give the boundaries of the stack of the current thread (which can be the main thread)16:45
Alex_Gaynoramaury_: I... I think it works.  I had an overly aggressive assert in there that verified it was using the exact CPython cookie, which it doesn't since we're using ulonglong for all vals, but I think it works.16:45
Alex_GaynorI'm going to translate16:45
Alex_Gaynoramaury_: Is the precise value of the cookie part of the public API?16:49
amaury_Alex_Gaynor: I don't think so16:49
amaury_except that the lower bits should be the position16:49
Alex_Gaynorthey are, its just that there's probably a bunch of extra 0 bits on 32-bit systems16:50
Alex_Gaynoroh and I think it might be wrong on big endian systems16:51
arigatonote that the position is a long long (i.e. a 64-bit) on all systems16:51
Alex_Gaynorarigato: sizeof(off_t) is 4 on my machine16:52
arigatoare you sure it's 4 even if you first #define a bunch of things?16:52
Alex_Gaynordepends what thing16:58
Alex_Gaynor#define off_t long long makes it 8 :)16:58
amaury_arigato: pthread_attr_getstack() retrieve the stack address + size17:01
arigatoamaury_: no, that's another level17:03
arigatoit's just reading out of the 'attr' structure17:03
amaury_can't you just store the info at thread creation,17:03
amaury_then simply look up this table?17:04
arigatowell you don't have complete info for the main thread, to start with17:04
arigatoAlex_Gaynor: sorry, I'm not finding it any more17:05
amaury_arigato: pthread_getattr_np could be useful17:05
arigatobut not portable :-)17:06
Alex_Gaynorarigato: finding what?17:06
arigatoAlex_Gaynor: the #defines17:06
Alex_Gaynorarigato: ok, no worries "8 bytse should be enough for anyone"17:06
amaury_"the returned thread        attributes object will report the actual stack address that the implementation        selected"17:06
arigatoAlex_Gaynor: it's   #define _FILE_OFFSET_BITS 6417:07
arigato/* This must be set to 64 on some systems to enable large file support. */17:07
arigatoamaury_: yes, but that function is really Linux-only17:08
Nick change: Varriount -> Varraway17:09
dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) joined #pypy.17:14
Action: amaury_ now understands what the "_np" suffix means17:15
Alex_Gaynorgoogle was surprisingly helpful for "_np"17:15
stakkars_ (~tismer@ joined #pypy.17:17
stakkars_ (~tismer@ left irc: Ping timeout: 240 seconds17:25
diracdelta (~christoph@c-67-188-126-146.hsd1.ca.comcast.net) joined #pypy.17:26
amaury_arigato: I just found this post about the LLVM gc infrastructure17:27
amaury_the bottom of the main article contains interesting items17:27
Alex_Gaynoramaury_: I'm hitting an assert (translated) inside of readline_w :/17:29
Alex_Gaynorcomes from _ll_stringslice__rpy_unicodePtr_Signed_Signed17:29
Alex_Gaynorrunning under lldebug now17:29
amaury_Alex_Gaynor: certainly from expression like line[start:endpos]17:30
Alex_Gaynoryup, doesn't say what the cause of the assertion is though17:31
amaury_start < end17:31
amaury_start <= end17:31
amaury_which test/function crashed?17:32
amaury_ah, multi_line_test() uses readline17:33
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Computer has gone to sleep.17:33
stakkars_ (~tismer@ joined #pypy.17:37
arigatoamaury_: still waiting for the link to the LLVM blog post :-)17:38
amaury_ah, sorry, forgot to press Enter17:38
amaury_I don't know what to answer to the mailing list17:42
amaury_about support for win64 17:42
Alex_Gaynor"Come onto IRC if you want to discuss" :)17:43
elmom_ (~elmom@hy-ovpn2-82.vpn.helsinki.fi) left irc: Ping timeout: 240 seconds17:44
arigatoamaury_: "no support at all so far" is an answer, but probably not good enough any more17:44
amaury_Alex_Gaynor: good idea17:44
Alex_Gaynoramaury_: copying multi_line_test passes ok as an AppTest, I don't suppose this will be easy17:44
arigatomaybe we could really #define long long long ?17:44
arigatoor something a bit more manageable17:45
arigatolike having lltype.Signed be really a 64-bit integer type17:45
amaury_or force lltype.Signed to longlong17:45
amaury_during one second, I was worried about performance implications17:46
arigatoit would probably require another round of fixes due to rffi.LONG != lltype.Signed, but it looks like the easiest solution for now17:46
fmilo (~mist0@ joined #pypy.17:46
amaury_but not on a 64bit machine...17:46
arigatoat least for languages like Python, I think it's reasonable to have Signed be a 64-bit int17:46
amaury_let's do it after pypy 1.517:51
arigatore LLVM: did they ever start the job described in that blog post?  as far as I'm concerned, it sounds indeed like "the next big thing" for LLVM17:52
Alex_Gaynorwhat is the plan for 1.5, release after we have full passing tests for 2.7?17:52
arigatoAlex_Gaynor: yes, ideally17:53
Alex_Gaynoramaury_: I've now manually put the assertions that ll_stringslice does into readline_w and it still passes :/17:53
amaury_"make lldebug", then17:55
Alex_GaynorI did, it doesn't seem very useful, do I need to gdb in or something?17:55
Alex_Gaynorwell that makes some sense17:56
amaury_unless you have a clear RPython stack17:56
arigato"make lldebug" is made so that gdb stops at the assertion failure directly, and you can inspect the value of the variables17:57
Alex_Gaynorah ok, good17:57
Alex_Gaynoramaury_: is there a way to get something even more debuggy, I've got some values optimized out, and also gdb seems unable to find the source code18:06
arigatoAlex_Gaynor: sorry about the values optimized out, there is no obvious way to improve, apart from getting a no-jit version which can then be compiled without gcc's -O218:07
Alex_Gaynorarigato: this is no-jit, its the default -O18:07
Alex_Gaynorno I lied, its -O118:07
Alex_Gaynorno, its default -O18:07
Action: Alex_Gaynor isn't good at reading his history18:08
arigatothen it's surely possible to edit the Makefile to remove "-O2" to gcc18:08
stakkars_ (~tismer@ left irc: Quit: schnarch18:08
arigatofinding the source code: it should just work, as long as you don't move things around (did you move the pypy-c somewhere else?)18:08
amaury_or run tests 5 times...18:09
Alex_Gaynoramaury_: Yup :(18:09
amaury_my temp dir is crippled with directories named usession-xxx-SAVE18:10
amaury_for this reason18:10
arigatoah, I usually use the "done" script from18:10
Nick change: Scriptor|away -> Scriptor18:16
Action: arigato -> dinner18:19
arigato (~arigo@fwstups.cs.uni-duesseldorf.de) left irc: Quit: See you18:20
blueadept (~username@unaffiliated/blueadept) joined #pypy.18:21
Trundle (~andy@python/site-packages/trundle) joined #pypy.18:40
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.18:43
ssbr_ (ssbr@173-230-191-202.cable.teksavvy.com) left irc: Ping timeout: 272 seconds18:50
nettok (~quassel@ joined #pypy.18:57
kenaan12hakanardo jit-fromstart 119d324e858411 15/pypy/jit/metainterp/: Allow the entire optimizer and its state to be cloned to allow the optimization to be resumed at any poin...18:59
ErKa (keryell@keryell.pck.nerim.net) joined #pypy.19:05
Nick change: Varraway -> Varriount19:10
Nick change: Varriount -> V19:11
Nick change: V -> Varriount19:11
ssbr (~ssbr@gw-wifi6.cdf.utoronto.ca) joined #pypy.19:17
gonsor (~quassel@kel30.kel.stud.uni-goettingen.de) left irc: Ping timeout: 264 seconds19:18
sphere_ (~siddharth@ left irc: Quit: sphere_19:35
tlynn_ (~tlynn@ joined #pypy.19:42
kkris (~kris@188-22-127-159.adsl.highway.telekom.at) left irc: Quit: Leaving.19:46
kkris (~kris@188-22-127-159.adsl.highway.telekom.at) joined #pypy.19:48
stakkars_ (~tismer@ joined #pypy.19:54
fijal (~fijal@ joined #pypy.19:58
ssbr (~ssbr@gw-wifi6.cdf.utoronto.ca) left irc: Ping timeout: 276 seconds20:00
ssbr (~ssbr@gw-wifi6.cdf.toronto.edu) joined #pypy.20:06
sphere_ (~siddharth@ joined #pypy.20:11
whyking_ (~quassel@p5B39ACF0.dip.t-dialin.net) joined #pypy.20:14
gonsor (~quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.20:14
fenrrir (~fenrrir@ joined #pypy.20:18
whyking (~quassel@p5B39DFC1.dip.t-dialin.net) left irc: Ping timeout: 276 seconds20:18
bivab (~david@ip-88-152-254-229.unitymediagroup.de) left irc: Quit: bivab20:21
fijal_ (~fijal@ joined #pypy.20:25
fijal (~fijal@ left irc: Disconnected by services20:26
Nick change: fijal_ -> fijal20:26
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) left irc: Read error: Connection reset by peer20:32
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) joined #pypy.20:33
kursor (~kursor@p57BB1816.dip0.t-ipconnect.de) left irc: Quit: kursor20:38
stakkars_ (~tismer@ left irc: Read error: Connection reset by peer20:44
fijalAlex_Gaynor: you little bastard :)20:49
fijalyou started the whole discussion why web2py is awful20:49
tlynn_ (~tlynn@ left irc: Ping timeout: 240 seconds20:51
bsod1 (~osa1@ joined #pypy.20:52
diracdelta (~christoph@c-67-188-126-146.hsd1.ca.comcast.net) left irc: Quit: diracdelta20:53
fijal (~fijal@ left irc: Ping timeout: 276 seconds20:56
ssbr (~ssbr@gw-wifi6.cdf.toronto.edu) left irc: Ping timeout: 240 seconds20:57
derdon (~derdon@p54A6F053.dip.t-dialin.net) left irc: Remote host closed the connection20:58
bsod1 (~osa1@ left irc: Ping timeout: 276 seconds20:58
kkris (~kris@188-22-127-159.adsl.highway.telekom.at) left irc: Quit: Leaving.21:00
lizardo (~lizardo@ left irc: Quit: Leaving21:01
Gromit (~bear@server00.klix.ch) left irc: Quit: Verlassend21:01
bsod1 (~osa1@ joined #pypy.21:02
bsod1 (~osa1@ left irc: Ping timeout: 240 seconds21:07
stakkars_ (~tismer@ joined #pypy.21:13
whyking_ (~quassel@p5B39ACF0.dip.t-dialin.net) left irc: Remote host closed the connection21:20
stakkars_ (~tismer@ left irc: Quit: schnarch21:29
Alex_Gaynoramaury_: So apparently the crash is "\n"[0:0}21:31
whyking (~quassel@p5B39ACF0.dip.t-dialin.net) joined #pypy.21:34
kenaan12amauryfa default 1164d00b0b042a 15/pypy/objspace/std/floatobject.py: Fix translation by python 2.5 (and pypy 1.4)21:38
ErKa (keryell@keryell.pck.nerim.net) left irc: Ping timeout: 240 seconds21:40
gonsor (~quassel@kel30.kel.stud.uni-goettingen.de) left irc: Ping timeout: 276 seconds21:41
Alex_Gaynoramaury_: I don't see how that can happen in either scenario, the first has `if endpos > start` 0 !> 0`, and the other has `if start > 0 or endpos < len(line)` and 0 !> 021:42
Arfrever (~Arfrever@gentoo/developer/Arfrever) left irc: Quit: Ex+re (KVIrc 4)21:42
amaury__ (5645c561@gateway/web/freenode/ip. joined #pypy.21:43
Alex_Gaynorerr, I should probably learn how `or` works21:43
jacob22 (~jacob@CPE0015f2633f3a-CM00159a662342.cpe.net.cable.rogers.com) left irc: Ping timeout: 264 seconds21:44
Ademan (~dan@adsl-71-141-238-143.dsl.snfc21.pacbell.net) joined #pypy.21:45
Alex_GaynorSo I'm not a GDB wiz, but looking at the params its slice(unicode, 0, 0)21:47
Alex_Gaynorbut going down and looking at the params its slice(unicode, 1, 0)21:47
peterd (~chatzilla@cpc1-cmbg11-0-0-cust61.5-4.cable.virginmedia.com) joined #pypy.21:49
gonsor (~quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.21:57
amaury__this is maybe an optimization22:01
amaury_ (5bc3480e@gateway/web/freenode/ip. left irc: Disconnected by services22:01
Nick change: amaury__ -> amaury_22:01
amaury_Alex_Gaynor: gdb may decide to allocate two variables at the same address22:01
amaury_if they are not used at the same time22:02
Alex_Gaynorshouldn't no -O and -g be enough to give GCC the hint that I don't want tricks22:02
amaury_-O or -O0 ?22:02
Alex_GaynorI just omitted -O entirely22:02
amaury_-O0 "is the default"22:03
amaury_so you are right22:04
Alex_Gaynorwell the condition that's failing is definitelately start - stop > 022:04
amaury_did you mean >=0?22:06
blueadept (~username@unaffiliated/blueadept) left irc: Quit: Leaving22:08
amaury_can you display the "length" variable?22:09
Trundle (~andy@python/site-packages/trundle) left irc: Remote host closed the connection22:10
amaury_it's named l_length_264 in my translation, you have to look at the source22:10
stakkars_ (~tismer@brln-4db85c3c.pool.mediaWays.net) joined #pypy.22:11
Alex_Gaynoramaury_: I've got an l_line_len_0, but its optimized out22:11
amaury_hey, didn't you remove optimizations?22:12
amaury_so it should not be "optimized out"22:12
amaury_by definition22:12
Alex_Gaynoryou should think :/22:12
Alex_Gaynorsigh, looks like -O1 snuck on, let me remove that and recompile22:13
Alex_Gaynordebug_flags had -O1, of course :(22:13
amaury_yes, asmgcroot needs a minimal level of optimization22:14
Alex_GaynorNo JIT here22:14
whyking (~quassel@p5B39ACF0.dip.t-dialin.net) left irc: Ping timeout: 276 seconds22:14
peterd (~chatzilla@cpc1-cmbg11-0-0-cust61.5-4.cable.virginmedia.com) left irc: Remote host closed the connection22:16
ssbr (~ssbr@gw-wifi14.cdf.toronto.edu) joined #pypy.22:21
sphere_ (~siddharth@ left irc: Quit: sphere_22:21
amaury_I'm getting weird things as well22:29
amaury_./pypy-c -m test.regrtest -v test_io22:29
amaury_goes in infinite loop and eats more and more memory22:29
amaury_it looks to be inside  "0L ** 0L"22:30
Alex_Gaynoramaury_: now that I'm at -O0 I see the correct calling params22:33
Alex_Gaynorl_line_len_0 is one22:33
amaury_does it still fail in the same place?22:33
amaury_start, stop?22:33
Alex_Gaynor(1, 0)22:33
Alex_Gaynornow the question is, do we have a bug, or is this simply a slice that's allowed by CPython and returns ""22:37
amaury_it's allowed in CPython, but not in RPython22:38
amaury_completely unrelated: which RPython code produces this? http://paste.pocoo.org/show/330829/22:39
Alex_Gaynorthat's interesting...22:40
Alex_Gaynorwhat function is that in?22:40
Alex_Gaynorno idea how that comes to be22:42
amaury_probably a wrapper common to all methods of W_TextIOWrappers with the same signature22:42
Alex_Gaynorsomething like that I guess22:42
Alex_Gaynoris the right solution "if start > endpos: line = """22:42
antocuni (~antocuni@host34-120-dynamic.11-79-r.retail.telecomitalia.it) left irc: Ping timeout: 276 seconds22:45
amaury_Alex_Gaynor: yes, but I'd prefer a test22:46
Alex_Gaynoramaury_: I can write one, but it won't fail I suppose, because on CPython that "just works"22:46
amaury_first, line 692 there should be an "assert start <= endpos"22:46
amaury_(in the paragraph: "# Our line ends in the current buffer")22:47
Alex_Gaynorok I'm going to port the full CPython test then22:49
kenaan12wlav reflex-support 110cefe1fd53e7 15/pypy/module/cppyy/: check for negative values on unsigned types22:55
Alex_Gaynoramaury_: heh, the existing test_telling also breaks if I put the assert in there, good22:59
Alex_Gaynorerr no it doesn't, my assert was backwards :(22:59
Alex_Gaynortoday seems like a typo day for me, blaming it on the weather23:00
Alex_Gaynoramaury_: http://paste.pocoo.org/show/330836/ passes fine with no modifications :/23:06
amaury_can you display each value of tell() and see if some values are unreasonably large?23:09
stakkars_ (~tismer@brln-4db85c3c.pool.mediaWays.net) left irc: Quit: schnarch23:11
Alex_Gaynoramaury_: we have a few chunksizes to go, but so far they are a) all the same, and b) all under 400023:12
Alex_Gaynorerr the same list of vals for each chunksize23:13
Alex_Gaynornot all the same within a chunksize23:13
Alex_Gaynoraand, done23:13
Alex_Gaynorno change, all tell values are still fine23:14
amaury_Alex_Gaynor: can you test whether "2**-53" still works with pypy-c?23:26
amaury_careful, here it hangs23:27
Alex_Gaynoramaury_: uhh, at a minimum it is SUPER slow23:27
Alex_Gaynoropen question whether its hanging or not23:27
amaury_and start eating all the memory23:27
Alex_Gaynoris this why test_io has been ssooooo slow to start?23:27
amaury_look at memory usage...23:27
kenneth_reitz (~kenneth_r@c-24-127-96-129.hsd1.va.comcast.net) left irc: Read error: Operation timed out23:28
Alex_Gaynoryay gc23:28
amaury_on tannit, its more like 4473Mb23:29
Alex_GaynorIf I wait I'm sure I'll get there23:30
Alex_Gaynornow what on earth is causing this?23:30
Alex_Gaynoramaury_: works fine on py.py23:31
amaury_even with --withmod-math, yes23:31
Alex_Gaynoramaury_: is the bug in dtoa (our impl I suppose) or somewhere else23:34
jacob22 (~jacob@CPE0015f2633f3a-CM00159a662342.cpe.net.cable.rogers.com) joined #pypy.23:34
amaury_two things changed today in this area:23:34
Alex_GaynorI made a change to -float ** nan23:35
amaury_the other is geninterp=False23:35
amaury_why may have unexpected results23:35
amaury_2L**-53L works23:38
amaury_mmm, the call stack looks like this23:40
amaury_IMO pow__Int_Int_None should have been called23:41
Action: amaury_ fears to debug multimethods again23:41
kenaan12wlav reflex-support 1199ad86908454 15/pypy/module/cppyy/test/test_datatypes.py: added test for builtin type conversions23:42
ahall (~dentist@2001:bd0:100::49) joined #pypy.23:44
--- Wed Feb 2 201100:00

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