#pypy IRC log for Thursday, 2011-12-29

andrewfr (~andrew@modemcable192.47-178-173.mc.videotron.ca) joined #pypy.00:32
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/75900:32
andrewfr (andrew@modemcable192.47-178-173.mc.videotron.ca) left #pypy.00:39
espes (espes@2001:470:1f05:ca1::) left irc: Ping timeout: 268 seconds00:49
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/1500:54
espes (~espes@123-243-221-248.tpgi.com.au) joined #pypy.00:58
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/18700:59
DasIch (~dasich@p4FFDCF99.dip.t-dialin.net) left irc: Remote host closed the connection01:09
DasIch (~dasich@p4FFDCF99.dip.t-dialin.net) joined #pypy.01:22
gutworth (~benjamin@64-131-9-52.usfamily.net) joined #pypy.01:26
DasIch (~dasich@p4FFDCF99.dip.t-dialin.net) left irc: Ping timeout: 252 seconds01:27
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/190401:30
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/65601:30
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/32001:30
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/31001:30
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/65901:30
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/117501:30
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/84201:30
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/149401:30
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/76901:30
nettok_ (~quassel@200.119.174.210) joined #pypy.01:32
nettok (~quassel@200.119.177.100) left irc: Ping timeout: 255 seconds01:34
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace01:53
espes (~espes@123-243-221-248.tpgi.com.au) left irc: Ping timeout: 252 seconds02:18
espes (~espes@123-243-221-248.tpgi.com.au) joined #pypy.02:27
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/99602:39
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds03:02
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/190403:41
thinker` (~thinker@218-167-101-86.dynamic.hinet.net) joined #pypy.03:46
hakanardo (~hakan@h59ec009e.seluldx.dyn.perspektivbredband.net) left irc: Read error: Operation timed out03:49
espes (~espes@123-243-221-248.tpgi.com.au) left irc: Ping timeout: 252 seconds04:03
tav_ (~tav@host-92-20-44-126.as13285.net) joined #pypy.04:07
espes (~espes@123-243-221-248.tpgi.com.au) joined #pypy.04:09
setmeaway (setmeaway@119.201.52.190) left irc: Read error: Connection reset by peer04:15
setmeaway (~setmeaway@119.201.52.190) joined #pypy.04:15
tilgovi (~randall@ool-4357ca60.dyn.optonline.net) joined #pypy.04:20
tilgovi (~randall@ool-4357ca60.dyn.optonline.net) left irc: Changing host04:20
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.04:20
Sweet-P (~Sweet-P@203.205.116.93) joined #pypy.04:24
gutworth (~benjamin@64-131-9-52.usfamily.net) left irc: Ping timeout: 244 seconds04:32
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/149404:44
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/84204:44
mimi_ (~mimi@97-88-28-179.dhcp.roch.mn.charter.com) joined #pypy.04:50
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/65904:50
gutworth (~benjamin@64-131-9-52.usfamily.net) joined #pypy.04:54
hakanardo (~hakan@h59ec009e.seluldx.dyn.perspektivbredband.net) joined #pypy.04:55
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/117505:00
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection05:03
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/65605:13
davisagli (davisagli@davisagli.com) left #pypy.05:14
anish_ (anish@nat/redhat/x-mytqdnhrphwenbqy) joined #pypy.05:34
gutworth (~benjamin@64-131-9-52.usfamily.net) left irc: Ping timeout: 248 seconds05:36
setmeaway (~setmeaway@119.201.52.190) left irc: Read error: Connection reset by peer05:45
setmeaway (setmeaway3@119.201.52.190) joined #pypy.05:45
setmeaway (setmeaway3@119.201.52.190) left irc: Read error: Connection reset by peer05:46
setmeaway (stemearay@119.201.52.190) joined #pypy.05:46
asmeurer (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) left irc: Quit: asmeurer05:52
aleksi (~aleksi@176.57.74.133) joined #pypy.05:57
aleksi (~aleksi@176.57.74.133) left irc: Remote host closed the connection06:05
kenaan (~kenaan@cube.twistedmatrix.com) left irc: Ping timeout: 252 seconds06:10
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/76906:13
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) left irc: Ping timeout: 252 seconds06:18
GDorn (gdorn@li223-147.members.linode.com) left #pypy.06:19
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/32006:23
aleksi (~aleksi@85.235.191.82) joined #pypy.06:38
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) joined #pypy.07:06
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/31007:10
fijal (~fijal@196-210-199-8.dynamic.isadsl.co.za) joined #pypy.07:43
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.07:44
hruske (~Gasper@188-230-156-183.dynamic.t-2.net) joined #pypy.07:50
hpk (~hpk@hq2.merlinux.eu) left irc: Read error: Operation timed out07:55
hpk (~hpk@hq2.merlinux.eu) joined #pypy.07:57
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) joined #pypy.08:04
hpk (~hpk@hq2.merlinux.eu) got netsplit.08:15
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) got netsplit.08:15
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) got netsplit.08:15
Circlefusion (~cf@74-138-199-109.dhcp.insightbb.com) got netsplit.08:15
raydeo (~raydeo@merickel.org) got netsplit.08:15
zk (~is@whatit.is) got netsplit.08:15
hpk (~hpk@hq2.merlinux.eu) returned to #pypy.08:19
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) returned to #pypy.08:19
stakkars (~tismer@p5DDB7A14.dip.t-dialin.net) returned to #pypy.08:19
Circlefusion (~cf@74-138-199-109.dhcp.insightbb.com) returned to #pypy.08:19
raydeo (~raydeo@merickel.org) returned to #pypy.08:19
zk (~is@whatit.is) returned to #pypy.08:19
bivab (~david@ip-62-143-59-5.unitymediagroup.de) joined #pypy.08:20
bivab (david@ip-62-143-59-5.unitymediagroup.de) left #pypy.08:20
bivab (~david@ip-62-143-59-5.unitymediagroup.de) joined #pypy.08:21
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) joined #pypy.08:23
zain (~textual@cpe-68-173-70-124.nyc.res.rr.com) left irc: Quit: Computer has gone to sleep.08:26
kenaan (~kenaan@cube.twistedmatrix.com) joined #pypy.08:27
kenaan12bivab arm-backend-2 11977efd55f3c6 15/pypy/jit/backend/arm/opassembler.py: (arigo, biab) make sure excetions are saved for call_assembler and call_may_force08:27
kenaan12bivab arm-backend-2 1118234110368e 15/pypy/jit/backend/arm/: make the save_exc parameter non-optional08:27
kenaan12bivab arm-backend-2 11aa7017a83625 15/pypy/jit/backend/arm/: fixes to emit_call and call_reaquire_gil08:27
kenaan12bivab arm-backend-2 11bd95dd546f05 15/pypy/jit/backend/arm/: Cleanup08:27
kenaan12bivab arm-backend-2 11c615cc3558ce 15/pypy/: merge default up to e4a0b9e4d23b08:27
kenaan12bivab arm-backend-2 11c61c064e84b4 15/: merge default up to 3ff9c88dca3908:27
kenaan12bivab arm-backend-2 114a920a79a182 15/pypy/jit/backend/: update backend to new interface provided by compute_vars_longevity08:27
kenaan12bivab arm-backend-2 11561c9d8112aa 15/: merge default up to 6fb87770b5d208:27
kenaan12bivab arm-backend-2 1157c6036823fe 15/pypy/jit/backend/arm/: implement chages to the frame manager08:27
kenaan12bivab arm-backend-2 11802e7c9df889 15/pypy/: merge 1a9a44331010 (jit-targets)08:27
kenaan12bivab arm-backend-2 11d8831765a657 15/pypy/jit/backend/: move is_comparison_or_ovf_op to llsuport08:27
kenaan12bivab arm-backend-2 113a1fca872e84 15/pypy/jit/backend/: implement changes needed for target/label functionality08:27
zain (~textual@cpe-68-173-70-124.nyc.res.rr.com) joined #pypy.08:29
zain (~textual@cpe-68-173-70-124.nyc.res.rr.com) left irc: Client Quit08:32
zain (~textual@cpe-68-173-70-124.nyc.res.rr.com) joined #pypy.08:36
zain (~textual@cpe-68-173-70-124.nyc.res.rr.com) left irc: Client Quit08:36
thinker` (~thinker@218-167-101-86.dynamic.hinet.net) left irc: Remote host closed the connection08:46
fijal (~fijal@196-210-199-8.dynamic.isadsl.co.za) left irc: Ping timeout: 240 seconds08:54
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) joined #pypy.09:03
espes (~espes@123-243-221-248.tpgi.com.au) left irc: Ping timeout: 276 seconds09:09
WildChild (~wildchild@modemcable123.12-202-24.mc.videotron.ca) left irc: Read error: Connection reset by peer09:13
WildChild (~wildchild@modemcable123.12-202-24.mc.videotron.ca) joined #pypy.09:13
espes (espes@2001:470:1f05:ca1::) joined #pypy.09:14
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.09:20
nettok_ (~quassel@200.119.174.210) left irc: Remote host closed the connection09:41
verte-{nyctea} (~verte@python/site-packages/verte) joined #pypy.09:56
anish_ (anish@nat/redhat/x-mytqdnhrphwenbqy) left irc: Quit: Leaving10:24
thinker` (~Sinker@114-24-2-146.dynamic.hinet.net) joined #pypy.10:25
anish_ (anish@nat/redhat/x-iutkfagkrxygcmlk) joined #pypy.10:25
thinker` (~Sinker@114-24-2-146.dynamic.hinet.net) left irc: Remote host closed the connection10:26
Sinker (~thinker@114-24-2-146.dynamic.hinet.net) joined #pypy.10:28
anish_ (anish@nat/redhat/x-iutkfagkrxygcmlk) left irc: Ping timeout: 252 seconds10:30
voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.10:32
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host10:32
voidspace (~anonymous@python/psf/voidspace) joined #pypy.10:32
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) left irc: Ping timeout: 244 seconds10:52
derdon (~derdon@pD9E1C99E.dip.t-dialin.net) joined #pypy.10:54
anish_ (anish@nat/redhat/x-erldvqbqewirxboc) joined #pypy.10:58
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace11:01
mat^2 (mathias@87.55.0.198) joined #pypy.11:04
voidspace (~anonymous@python/psf/voidspace) joined #pypy.11:04
lizardo (~lizardo@189.2.128.130) joined #pypy.11:05
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 268 seconds11:06
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.11:10
dracman (~draco@31-17-70-40-dynip.superkabel.de) left irc: Read error: Connection reset by peer11:13
kkris (~kris@93-82-44-248.adsl.highway.telekom.at) joined #pypy.11:20
tav_ (~tav@host-92-20-44-126.as13285.net) left irc: Ping timeout: 240 seconds11:23
azanella (~azanella@189.6.80.207) joined #pypy.11:25
Sinker (~thinker@114-24-2-146.dynamic.hinet.net) left irc: Ping timeout: 268 seconds11:27
tav_ (~tav@host-92-20-28-45.as13285.net) joined #pypy.11:30
nedbat (~nedbat@python/psf/nedbat) joined #pypy.11:37
Sinker (~thinker@114-45-177-184.dynamic.hinet.net) joined #pypy.11:39
`fox` (~fox@79.45.107.142) joined #pypy.11:47
anish_ (anish@nat/redhat/x-erldvqbqewirxboc) left irc: Quit: Leaving12:05
Sinker (~thinker@114-45-177-184.dynamic.hinet.net) left irc: Ping timeout: 252 seconds12:18
Nick change: verte-{nyctea} -> wleslie12:19
wleslie (~verte@python/site-packages/verte) left irc: Quit: ~~~ Crash in JIT!12:29
aleksi (~aleksi@85.235.191.82) left irc: Remote host closed the connection12:32
JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy.12:35
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Ping timeout: 244 seconds12:39
`fox` (~fox@79.45.107.142) left irc: Ping timeout: 240 seconds12:39
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) joined #pypy.12:43
steakknife (~steakknif@pfsense.hackerdojo.com) left irc: Quit: Leaving.12:52
DasIch (~dasich@p4FFDCF99.dip.t-dialin.net) joined #pypy.13:14
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds13:16
DasIch (~dasich@p4FFDCF99.dip.t-dialin.net) left irc: Ping timeout: 240 seconds13:18
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) joined #pypy.13:20
mimi_ (~mimi@97-88-28-179.dhcp.roch.mn.charter.com) left irc: Ping timeout: 255 seconds13:33
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Ping timeout: 240 seconds13:35
anish_ (anish@nat/redhat/x-iyiaezjivoygbxdz) joined #pypy.13:36
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds13:38
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.13:43
Rhy0lite (dje@nat/ibm/x-vonrcckenjkanofl) joined #pypy.13:45
anish_ (anish@nat/redhat/x-iyiaezjivoygbxdz) left irc: Ping timeout: 252 seconds13:48
Rhy0litebivab: hi13:48
bivabRhy0lite: hi13:48
Rhy0liteafter our fix yesterday, I was able to fix a lot of the remaining PPC64 failures13:49
Rhy0litethere are a few strange ones, but mostly the skipped float tests13:49
bivabcool13:50
bivabso both backends are more or less at the same point again13:50
bivabor both versions of the backend13:50
Rhy0liteprobably13:50
Rhy0liteI'm not sure about the exact state of the PPC32 backend13:51
Rhy0liteOne strange failure is the memoryerror test13:51
bivabsven and armin were working on that test, but I do not recall how far they got with it13:52
Rhy0liteokay13:52
Rhy0liteI did not realize that failed on PPC32 as well13:53
Rhy0liteI don't know how the allocation can end up inconsistent13:53
Rhy0liteit's mainly that one and call_assembler13:54
bivabthe idea for the memory error is to allocate something larger than the allocator can handle to trigger an error, is that what you mean?13:56
Rhy0litemaybe I'm confused about the test I mean13:59
Rhy0litemaybe it's test_free_loop_and_bridges13:59
Rhy0liteyes, that one14:00
Rhy0litewhere total mallocs don't match14:00
bivabthe function get_asmmemmgr_blocks is defined twice in ppc_assembler.py, maybe it is related to that issue14:06
bivablet me check14:06
jnoller (~jnoller@vpn.nasuni.com) joined #pypy.14:08
Rhy0liteah!14:09
Rhy0litehmm, and one seems wrong14:11
Rhy0liteno change from removing the one that sets clt.asmmemmrg = []14:12
bivabhm, ok14:12
iliya (~iliya@87-126-20-213.btc-net.bg) joined #pypy.14:26
lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) joined #pypy.14:32
lahwranthis code really needs to be reviewed by a security pro14:44
lahwranwould the pypy jit cause pypy to perform more slowly than cpython while the jit is doing it's thing?14:57
lahwranie, for very short-lived python programs, would cpython be a better choice?14:57
lahwranor perhaps pypy without jit14:58
lesshastelahwran: I don't know the answer but logically there must be some startup costs from pypy14:58
lesshastelahwran: how short is short?14:58
lahwran0.56 seconds on cpython, 0.74 seconds on pypy14:59
lesshasteout of interest.. if you run it only once you presumably don't care about the time difference14:59
lahwranno, it will be run very often15:00
lesshasteand if you run it many times, wouldn't it make sense to do that from within some other program?15:00
lahwranno15:00
lahwranwell15:00
lahwran:/ I suppose it might15:00
lahwranthe program is pypy_interact.py, the master component of the sandbox15:00
lesshastelahwran: you can time the startup cost fairly easily15:01
lahwranoh?15:02
lesshastelahwran: by writing a program that just prints "hello" to the screen for example15:02
lesshasteand timing it15:02
lesshastewell.. that will tell you something :)15:02
lesshastemaybe not everything15:02
lahwranthe startup time of the jit will be longer for a larger program15:02
lahwranit shouldn't be significant, though; it's not a particularly large program, maybe a couple thousand lines on the longest path15:03
lesshastelahwran: for detailed questions you may find the mailing list more helpful 15:04
lesshasteI think the real gurus are all asleep now15:04
lesshasteor working :)15:04
lahwranhnnng I don't like mailing lists15:04
lesshastethen ask your question here with some pastbin'ed example and wait for 12 hours :)15:05
fijal (~fijal@197.170.75.66) joined #pypy.15:05
lahwranspeak of the devil15:05
lesshastelahwran: :)15:05
lahwranlesshaste: nah, I really need to sleep15:05
lesshastelahwran: maybe they are waking up now :)15:05
lahwranand it's not significant, anyway; it turns out there's nothing wrong with running it on cpython15:06
lahwranlooks like it was already using the pypy internal marshal.py15:06
lesshastehi fijal 15:06
fijalhi15:06
fijalwhat's up?15:06
lesshastetrying to work out how to make my code fast without rewriting it in C :)15:06
lesshastehow about you?15:06
lesshasteany ideas for things to help pypy make this faster? http://paste.pocoo.org/show/527346/15:07
lesshasteI see there is to be a talk on this topic soon 15:08
lesshastefrom the blog15:08
lesshastethe bottleneck is  hds = [hamdist(unknown, pattern[i:l+i]) for i in xrange(l)] which is called a lot.. I believe15:08
fijalwhat's hamdist?15:09
fijaland what's pattern?15:09
fijallesshaste: you can look at traces15:09
lesshastehamdist(str1, str2) just outputs the Hamming distance between two equal length strings15:09
fijalbut I need to fix jitviewer to show code15:09
lesshastethat is the number of symbols that differ15:10
lesshastea pattern is just a string of digits15:10
fijalone way to speed it up would be to pass the original string and indexes to hamdist15:10
fijalinstead of a slice15:10
fijalI bet this is what you do in  C15:10
fijalI can't say upfront if this is the case here15:10
lesshasteinteresting.. thanks15:10
fijalJIT in general should be able to figure such stuff out15:15
fijalbut not in this case15:15
fijallesshaste: note that this is a potential optimization we *could* do, but we did not so far15:15
lesshastefijal: cool15:15
lesshasteI heard some rumour of sse work too :)15:16
lesshastewhich sounds like a good SoC project to me15:16
fijalfor numpy only so far15:16
lesshastemade the change... 34 seconds to... 36 :)15:18
lesshasteoh15:18
lesshasteerr15:18
lesshastelet me do that again15:19
lesshaste34 seconds to ...2915:19
lesshastewell that's something15:20
lesshasteyou would have thought hamdist could be done amazingly quickly15:21
kenaan12edelsohn ppc-jit-backend 118e95fc00f638 15/pypy/jit/backend/ppc/ppcgen/ppc_assembler.py: Remove duplicate, broken definition of get_asmmemmgr_blocks. Store TOC at correct offset for PPC64.15:21
lesshastefijal: it's called 73 million times. I would have thought each call would be about < 100 instructions15:22
fijallesshaste: you have to think how python operations are implemented15:22
fijallesshaste: I'll look into that soon, but not right now15:22
fijalget back to me ok?15:22
lesshastethanks!15:22
lesshastelet me paste the modified version15:22
lesshastehttp://paste.pocoo.org/show/527356/15:23
lesshasteI added comment line saying that if it's too fast on your computer, increase l to 615:24
fijal:]15:24
lesshastemy computer is slow so l=5 takes about 30 seconds :)15:24
lesshasteand 30 seconds is enough for me :)15:24
fijal (~fijal@197.170.75.66) left irc: Read error: Connection reset by peer15:26
fijal (~fijal@197.172.62.161) joined #pypy.15:27
nedbat (~nedbat@python/psf/nedbat) joined #pypy.15:28
gutworth (~benjamin@64-131-9-52.usfamily.net) joined #pypy.15:34
bivabRhy0lite: found the reason for the failing test_free_loop_and_bridges test15:38
Rhy0litegreat15:38
bivabthe generated function descriptors are not allocated correctly15:38
Rhy0litethat makes sense as a point of leak15:39
Kaskuka (~John@osbk-4db176e0.pool.mediaWays.net) left irc: Ping timeout: 240 seconds15:40
Kaskuka (~John@osbk-4d087292.pool.mediaWays.net) joined #pypy.15:40
bivabyes, I will refactor the code a bit and add the fix15:41
Rhy0litegreat.  that means all support other than float is close to complete15:46
lahwrannedbat: I made a few changes to how --timeout works, uses select() and capping time.sleep now instead of a thread15:46
lahwrannot sure how to contribute them, though15:47
fijalRhy0lite: what's missing from complete completion?15:49
Rhy0litefijal: test_assembler_call and test_cond_call_gc_wb_array_card_marking_fast_path15:51
Rhy0liteand test_memoryerror15:51
fijalyou don't need gc_wb_fastpath15:53
fijalfor stuff to work15:53
fijalas in you can try already translating if you fix test_call_assembler15:54
fijalat least small interpreters15:54
fijalis test_zll_random passing?15:54
Rhy0liteI haven't tried test_zll_random again15:54
fijalRhy0lite: ok, this should pass before anything else15:56
fijaland will catch errors15:56
Rhy0liteyes15:56
Rhy0litejust celebrating the progress15:56
kenaan12hager ppc-jit-backend 110c305ddabb69 15/pypy/jit/backend/ppc/ppcgen/ppc_assembler.py: (bivab) Fix allocation of function descriptors for PPC6415:56
fijalok :)15:57
Rhy0litebivab: yea!15:57
bivabRhy0lite: :)15:57
mfoord (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.16:03
mfoord (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host16:03
mfoord (~anonymous@python/psf/voidspace) joined #pypy.16:03
voidspace (~anonymous@python/psf/voidspace) left irc: Ping timeout: 252 seconds16:05
Nick change: mfoord -> voidspace16:05
mfoord (~anonymous@python/psf/voidspace) joined #pypy.16:06
lahwranwhat does __pypy__.builtinify do?16:09
fijallahwran: you know how C functions and python functions differ in CPython?16:09
lahwranno16:09
voidspace (~anonymous@python/psf/voidspace) left irc: Ping timeout: 240 seconds16:09
fijalif you attach a C-function to a class, it does not become a method16:10
fijalpypy functions have the same object whether they're in python or rpython16:10
fijalso we have a wrapper that makes them more builtin-like16:10
fijalobscure, isn't it?16:10
mfoord (~anonymous@python/psf/voidspace) left irc: Ping timeout: 240 seconds16:10
lahwranit just deletes __get__?16:10
lahwranis that really it? ._.16:11
fijalit's "enough"16:11
lahwranwhat is the purpose of doing that?16:11
lahwran_marshal seems to do it16:12
fijalbecause people complain16:12
fijal"but on CPython it behaves differently!"16:12
fijalthe example is:16:13
fijalclass A(object):16:13
lahwranexcuse me, I'm just going to shove my hand through my face16:13
fijal close = os.close16:13
fijalnow, can you run A.close()?16:13
lahwranI see16:13
fijalor A().close()?16:13
fijaldepending how close is implemented (python vs C) you might or might not16:13
voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.16:14
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host16:14
voidspace (~anonymous@python/psf/voidspace) joined #pypy.16:14
Alex_Gaynorfijal: A should clearly subclass int :)16:14
fijallahwran: generally "to fix it's easier than explain to tons of people"16:14
fijalAlex_Gaynor: hey16:14
lahwranseems like a bad thing to do anyway, but I can understand trying to be as compatible as possible with cpython16:14
lahwranI'm just trying to figure out _marshal16:15
Action: Rhy0lite -> lunch16:15
lahwranis _marshal normally run through rpython?16:16
mat^2 (mathias@87.55.0.198) left irc: Ping timeout: 240 seconds16:16
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds16:18
davisagli (~davisagli@davisagli.com) joined #pypy.16:20
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds16:20
kenaan12fijal jitviewer 119e5303808452 15/: try slightly harder to display source, even in extraordinary conditions16:22
fijalyes16:22
fijallahwran: so no wonder importing a version from lib_pypy takes longer16:22
lahwraneh?16:23
fijalAlex_Gaynor: jitviewer should work16:23
fijallahwran: import marshal in pypy just imports a builtin moduel16:23
fijalAlex_Gaynor: at least pieces of it16:23
fijallesshaste: ok, enough of jitviewer works for me to give a look16:23
fijallesshaste: point 1 - we can't (yet) vectorize whatever you're doing16:23
fijalC can probably16:23
fijallesshaste: 1. put stuff in a function, don't run it as globals16:28
fijallesshaste: meh, there are definitely a few things that can be improved :/16:31
Alex_Gaynorfijal: cool16:32
fijallesshaste: I'm forced to go to a party, will look more later16:32
Alex_Gaynorfijal: CPython compiles very quickly you know16:32
fijalAlex_Gaynor: assembler and jumps don't quite work16:32
fijalthe rest is fine16:32
fijalAlex_Gaynor: yeah, figured out16:32
fijalAlex_Gaynor: feel like adding it to speed.python buildbot?16:32
Alex_Gaynornot really16:33
fijalI think lesshaste found a pretty bad case for the jit16:34
fijal (~fijal@197.172.62.161) left irc: Read error: Connection reset by peer16:35
iliya (~iliya@87-126-20-213.btc-net.bg) left irc: Quit: Ex-Chat16:37
lesshastethanks fijal16:39
lesshastefor when you get back :)16:39
G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving.16:44
witulski (~stupsi@ip-176-198-244-135.unitymediagroup.de) joined #pypy.16:46
JaRoel (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection16:51
witulski (~stupsi@ip-176-198-244-135.unitymediagroup.de) left irc: Ping timeout: 252 seconds16:51
mat^2 (mathias@87.55.0.198) joined #pypy.16:58
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.17:02
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection17:06
kkris1 (~kris@93-82-45-133.adsl.highway.telekom.at) joined #pypy.17:11
kkris (~kris@93-82-44-248.adsl.highway.telekom.at) left irc: Ping timeout: 252 seconds17:12
Nick change: exarkun_ -> exarkun17:15
#pypy: mode change '+o exarkun' by ChanServ!ChanServ@services.17:15
nettok (~quassel@200.119.156.207) joined #pypy.17:18
nettok_ (~quassel@200.119.162.14) joined #pypy.17:31
nettok (~quassel@200.119.156.207) left irc: Ping timeout: 252 seconds17:33
asmeurer (~asmeurer@c-174-56-21-245.hsd1.nm.comcast.net) joined #pypy.17:35
derdon (~derdon@pD9E1C99E.dip.t-dialin.net) left irc: Remote host closed the connection17:38
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace17:41
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood17:45
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.17:47
antocuni (~antocuni@host117-71-dynamic.58-82-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds17:56
nettok (~quassel@200.119.151.199) joined #pypy.17:57
nettok_ (~quassel@200.119.162.14) left irc: Ping timeout: 255 seconds17:59
jnoller (~jnoller@vpn.nasuni.com) left irc: Quit: Leaving...18:01
nedbat (~nedbat@python/psf/nedbat) joined #pypy.18:02
lmatosHi guys, I'm using pypy with venvs and setting --python=/usr/bin/pypy and --no-site-packages for my project is installing the packages in /opt/pypy/site-packages/. I've done some searching on pypy and venvs and the data looks to be a little dated. Any tips?18:07
apollo13 (~apollo13@unaffiliated/apollo13) left irc: Ping timeout: 268 seconds18:09
ronnylmatos: how are you running the installer?18:12
lmatosronny: virtualenv --python=/usr/bin/pypy --no-sitepackages venvname18:13
lmatos I'm running Arch with this pypy build: http://www.archlinux.org/packages/community/i686/pypy/18:14
ronnylmatos: and how do you run the installer for your package18:15
MrGecko (~LizardWiz@219.161.66.86.rev.sfr.net) joined #pypy.18:16
MrGeckoBonsoir :o)18:16
lmatosronny: you mean the arch package? or my python code? The python code is django. To make sure we aren't circling the drain here let me give some more info. I've been trying to get gunicorn working this morning and it keeps telling me no module djang. I don't know where/why that failed so I was building a new env to test. That's when I noticed it wasn't installing in the venv's site-packages but rather /opy/pypy/. Does that help?18:18
lmatosdjango**18:18
MostAwesomeDudeDoes gunicorn work with PyPy?18:18
Action: MostAwesomeDude checks18:18
ronnylmatos: so how do you invoke the installer18:19
MostAwesomeDudeHm, yep, marked as working.18:19
ronny(of the python thing you want in the env)18:20
lmatospip install thing but then it gave me permission denied to install in /opt/pypy/ so I then sudoed and realized wait, it shouldn't be installing there.18:22
ronnywhich pip did you invoke how18:23
apollo13 (~apollo13@unaffiliated/apollo13) joined #pypy.18:26
lmatosronny: ah fsck. It looks like it was "global" pip. Not my venv's pip.18:27
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) left irc: Quit: etrepum18:28
MrGeckoDo you think I can yet improve the performance of my application trying to use array.array *naive*; I just read it is planned to be coming with pypy 1.8 but it looks like it is already shipped with 1.7  ?18:35
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds18:35
lmatosronny: thanks for the lesson. xD18:36
WildChild (~wildchild@modemcable123.12-202-24.mc.videotron.ca) got netsplit.18:38
Dulak (~michael@unaffiliated/dulak) got netsplit.18:38
WildChild (~wildchild@modemcable123.12-202-24.mc.videotron.ca) returned to #pypy.18:38
Dulak (~michael@unaffiliated/dulak) returned to #pypy.18:38
Rhy0lite (dje@nat/ibm/x-vonrcckenjkanofl) left irc: Quit: Leaving18:40
DanKluev (~root@109-109-211-093-xdsl.vntc.ru) left irc: Remote host closed the connection18:43
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) joined #pypy.18:45
kalenz (~kalenz@ping.kalenz.fr) left irc: Ping timeout: 255 seconds18:51
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) left irc: Ping timeout: 276 seconds18:54
DanKluev (~root@109-109-211-093-xdsl.vntc.ru) joined #pypy.18:55
kalenz (~kalenz@ping.kalenz.fr) joined #pypy.18:57
Nick change: kalenz -> Guest151418:57
`fox` (~fox@host142-107-dynamic.45-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds18:59
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) joined #pypy.19:09
nedbat (~nedbat@python/psf/nedbat) joined #pypy.19:26
nettok_ (~quassel@190.143.184.117) joined #pypy.19:36
nettok (~quassel@200.119.151.199) left irc: Ping timeout: 276 seconds19:39
fijal (~fijal@41.53.212.9) joined #pypy.19:45
bivab (~david@ip-62-143-59-5.unitymediagroup.de) left irc: Quit: bivab19:47
fijallesshaste: ping?20:00
timotimoMrGecko: what do you mean with naive?20:00
MrGeckoi'm looking for magic tricks that will boost my perfs without much efforts :p20:01
fijalMrGecko: you can definitely try20:01
fijalarray.array should be pretty fast20:01
timotimowell, access to arrays are already jitted if your loops are good for that kind of thing20:01
fijaldepends on your usecase probably, but in general it should work20:01
lesshastefijal: hi20:02
timotimoyou can check with the jitviewer if the assembly instructions for your loops make sense20:02
MrGeckoyou know, when you're out of ideas to improve your algorithm, you're ready to try any command line and compilation options :p20:02
timotimoif they are particularly wasteful, you can probably talk to the people more involved with the jit here20:02
fijallesshaste: so, when you pack stuff into function how much faster it is?20:02
fijallesshaste: it is quite a bit for me20:02
fijalalso, I fixed the random seed so it does not depend on the run20:02
kenaan12boemmels lang-scheme 11b27460bc0d0d 15/scheme/test/test_scheme_level.py: Added a closure test20:02
kenaan12boemmels lang-scheme 1162cfba56990c 15/scheme/: Implement "append" & "append!"20:02
kenaan12boemmels lang-scheme 119eb92cb30772 15/scheme/: Move Parser helper-functions from object.py to ssparser.py20:02
lesshastefijal: oh sorry I haven't done that.. well I did pass the index as you suggested not the slice20:03
fijallesshaste: but in general, this is a pretty bad case for the JIT20:03
lesshastefijal: what do you mean by pack stuff into  a function in this case?20:03
fijalI can explain if you care :)20:03
lesshasteI do!20:03
lesshaste:)20:03
fijallesshaste: don't run things in a global environment20:03
fijalbut instead have everything in def main(): ...20:03
lesshasteoh I see20:03
fijaland if __name__ == '__main__': main()20:03
lesshasteyou can tell what a python expert I am :)20:03
fijalit's a good idea either way :)20:03
fijalso calls to functions that contain loops that execute usually few instructions are worse than they should be20:04
lesshasteah20:04
fijalit's definitely a "bug"20:04
fijalbut a hard to fix one20:04
lesshastein this case we are talking a small constant number of comparison20:04
lesshastes20:04
fijalwe can try playing with stuff20:04
fijalyes20:05
lesshasteis there a way to rewrite it to make pypy happier?20:05
fijalI found it's ~27% of time it exits20:05
fijalwell, even if there is, I don't think it's a good idea20:05
aleksi (~aleksi@176.57.74.133) joined #pypy.20:05
lesshasteright20:05
fijalI think the JIT should figure out better20:05
lesshasteI didn't understand "I found it's ~27% of time it exits"20:05
lesshastehow much of a difference did using def main make?20:06
fijalI don't know, measure20:07
fijalalso fix the seed20:07
fijalbecause it can be different between runs20:07
lesshastehmm.. calls to functions  that contain loops that execute usually few instructions.. this is where inlining would be used in C right?20:07
fijalno, pypy inlines20:07
fijalthe crux is "contain loops"20:08
lesshasteoh20:08
lesshasteso what is the overhead?20:08
lesshasteah20:08
lesshasteloop unrolling?20:08
fijalyou don't inline because traces are linear20:08
fijaland because you don't inline, there is frame creation20:08
fijal(python frame creation)20:08
fijalpacking and unpacking20:08
fijalpretty annoying stuff20:08
lesshasteok .. I think I follow20:08
lesshasteso can you do some sort of loop unrolling?20:08
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Read error: Connection reset by peer20:09
fijalthat's potentially dangerous20:09
fijalbut we should improve the case where you don't do unrolling as well20:09
fijallesshaste: anyway, I'll go to bed probably, this is something that's in my mind for a bit20:10
fijalbut it's not directly "right now"20:10
fijalbecause it's hard and complex and requires experimentation20:10
nickik (~nick@2001:67c:20a1:1018:226:9eff:feef:8369) joined #pypy.20:13
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.20:13
nickik (~nick@2001:67c:20a1:1018:226:9eff:feef:8369) left irc: Client Quit20:13
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) joined #pypy.20:16
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Ping timeout: 240 seconds20:17
MrGeckoOoh... nice Fatal RPython error: AssertionError :D20:19
MrGeckoDid I win something ?20:19
MrGeckodamn, I can't reproduce it again20:23
jerithYup. An opportunity to debug. :-D20:23
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.20:29
Vorpal (~AnMaster@unaffiliated/anmaster) left irc: Excess Flood20:30
fijalMrGecko: there were quite a few fixed since 1.720:31
fijalMrGecko: also it should be reproducible20:32
MrGeckoI use a nightly build I got a few days ago20:33
MrGeckoI really don't know how to reproduce it, i'm sorry :)20:33
fijalheh20:34
fijalok, so if you're using a nightly build, then try the release maybe :)20:34
MrGeckoIt only occured one time when I random.shuffle'd a list20:34
Nick change: Guest1514 -> kalenz20:34
fijalwhat we did was to merge quite a few experimental features20:34
fijalsince the release20:34
fijalmeh :/20:34
fijaltoo bad you don't know the seed20:35
fijalMrGecko: anyway, any luck with speed?20:35
MrGeckoNot yet.. But that's not a surprise, I didn't really expect any change.20:37
fijal (~fijal@41.53.212.9) left irc: Ping timeout: 240 seconds20:41
lizardo (~lizardo@189.2.128.130) left irc: Ping timeout: 252 seconds20:42
aleksi (~aleksi@176.57.74.133) left irc: Remote host closed the connection20:48
lmatosany psycopg2-ctypes users in here?21:04
exarkunI used pg8000 instead, it worked.21:05
lmatosexarkun: does that play nice with django? I'm not sure :/21:06
lmatosexarkun: was the speed acceptable?21:06
jerithlmatos: I used it to test that my stuff works on pypy.21:07
jerithIt doesn't play nice with Django for the same reasons psycopg2 1.4.2+ doesn't.21:08
mvt (~mvantelli@53530442.cm-6-4a.dynamic.ziggo.nl) joined #pypy.21:08
jerith(psycopg2ct, that is.)21:08
lmatosjerith: can you elaborate or provide links to such info? I'm not sure why psycopg2 1.4.2 doesn't play nice with django :/21:09
jerithEr, 2.4.221:09
jerithhttps://code.djangoproject.com/ticket/1625021:10
mvtthey should have included it in 1.3.1 ...21:11
mvtanyway afaik only the testrunner is affected21:12
jerithYeah.21:12
jerithBut still, it's nice to be able to run tests for your code.21:12
fijal (~fijal@197.109.111.224) joined #pypy.21:23
Count_Niedar (~nnscript@ip68-231-248-182.hr.hr.cox.net) joined #pypy.21:28
Niedar (~nnscript@ip68-231-248-182.hr.hr.cox.net) left irc: Ping timeout: 240 seconds21:31
fijalAlex_Gaynor: ping21:32
lmatosexarkun: was the speed acceptable?21:34
lmatosexarkun: ^ ignore this. lag issue. 21:35
lmatosjerith: I mean it's running it's just looking slow so I'm not sure if it's pyscopg2ct or something else. I'm going to do some profiling to see but I suspect it's pyscopg2ct as it seems like it would be the "weakest link"21:36
mvtlmatos: really psycopg2ct should not be that much slower then psycopg2 (i don't think the overhead is that big for 40 a 50 queries)21:36
jerithlmatos: The Django ORM is made out of slow.21:37
mvtbut you should really run django under pypy for a while before it get's faster21:37
lmatosjerith: true. 21:37
Alex_Gaynorfijal: pong21:38
lmatosIF you guys wanna help the stack is running here: www.gencyclothing.com and you'll see what I mean. (No CSS ATM as I need to adjust some paths)  What's the best way as to begin profiling this?21:39
fijalAlex_Gaynor: if you really care about the hash attack, cuckoo hashing is to be considered21:39
mvtlmatos: Try to run it with cpython 2.7 + psycopg2ct as a test21:39
mvtif that is also really slow, then psycopg2ct is the bottleneck21:40
Alex_Gaynorfijal: I think shipping a Tree based map or such is better21:40
lmatosmvt: I'm going to try that now with a test stack.21:40
fijalAlex_Gaynor: that's cpython's job, isn't it21:41
fijal"only in 3.8"21:41
gutworthtrees are so slow21:42
exarkunespecially hard wood21:42
gutworthalthough less slow than a dosed dict :)21:43
voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.21:44
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host21:44
voidspace (~anonymous@python/psf/voidspace) joined #pypy.21:44
fijallmatos: cprofile would be a good start21:45
lmatosfijal: I'm going to give that a look as soon as I test this stack. 30% there. :)21:48
lmatosthanks!21:48
chee1 (~chee1@fw2.wink.com) joined #pypy.21:49
kkris1 (~kris@93-82-45-133.adsl.highway.telekom.at) left irc: Quit: Leaving.21:58
CIA-5303cheecheeo 07roundup * 10#974/channel is not iterable: 22:02
CIA-53[new] pypy's channels are not iterable although the stackless documentation states22:02
CIA-53that they are:22:02
CIA-53http://www.disinterest.org/resource/stackless/ ... * 14https://bugs.pypy.org/issue97422:02
voidspace (~anonymous@python/psf/voidspace) left irc: Quit: voidspace22:03
fijal (~fijal@197.109.111.224) left irc: Read error: Connection reset by peer22:18
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/76022:30
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/18822:30
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/99722:30
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/1622:30
azanella (~azanella@189.6.80.207) left irc: Ping timeout: 244 seconds22:41
voidspace (~anonymous@87-194-212-65.bethere.co.uk) joined #pypy.22:52
voidspace (~anonymous@87-194-212-65.bethere.co.uk) left irc: Changing host22:52
voidspace (~anonymous@python/psf/voidspace) joined #pypy.22:52
mvtlmatos, just did some testing regarding psycopg2ct, but it really is faster then psycopg2 if you let the JIT do it's thing22:54
lmatosmvt: "let the JIT do it's thing" implies that I don't need to do anything special to get it to do it? What needs to happen to get that JIT doing thangs?22:55
mvtrendering a page with 1600 rows with a semi-complex orm query is for example ~30-40% faster running with pypy + psycopg2ct then cpython + psycopg222:55
mvtwell for every request it should get faster22:56
mvtI'm not an expert, but i thought that a loop needed to be iterated 1000 times before it gets fast22:57
lmatosAlex_Gaynor: ^ can you confirm this?22:58
Alex_Gaynor1039 iterations of a loop, yes22:59
Alex_Gaynoryou can lower it if you want, but that can have mixed effects22:59
lmatosthanks! No need to lower, just more testing.22:59
lmatosmvt: I appreciate it! I was working on my testing setup and my vm got wiped. :( I'm going to try cprofile now to see what I can learn. :D23:00
lmatosI've never profiled before and I'm trying to get an idea of the best way to profile django. Any tips?23:02
mvtthe django-command-extensions has a runprofileserver, the you simply run "python manage.py runprofileserver  --kcachegrind --prof-path=profiles"23:05
mvtand open the profile with kcachegrind23:05
mvtor runsnakerun 23:06
mvt (~mvantelli@53530442.cm-6-4a.dynamic.ziggo.nl) left irc: Quit: Leaving23:10
chee1 (chee1@fw2.wink.com) left #pypy.23:12
fschwidom (~fschwidom@2.208.38.85) joined #pypy.23:28
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 276 seconds23:31
fschwidom (~fschwidom@2.208.38.85) left irc: Remote host closed the connection23:33
nedbat (~nedbat@python/psf/nedbat) joined #pypy.23:34
MrGecko (~LizardWiz@219.161.66.86.rev.sfr.net) left irc: Remote host closed the connection23:39
Arfrever (~Arfrever@apache/committer/Arfrever) left irc: Read error: Operation timed out23:45
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds23:57
--- Fri Dec 30 201100:00

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