#pypy IRC log for Friday, 2011-10-28

aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) joined #pypy.00:08
twomashi (~Adium@201-213-122-142.net.prima.net.ar) joined #pypy.00:12
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/92100:12
caution (~caution@unaffiliated/caution) left irc: Read error: Connection reset by peer00:12
whitelynx (~whitelynx@75.110.189.230) joined #pypy.00:13
_habnabit (~habnabit@python/site-packages/habnabit) left irc: Quit: ZNC - http://znc.sourceforge.net00:13
antocuni_ (~antocuni@host193-122-dynamic.0-79-r.retail.telecomitalia.it) left irc: Ping timeout: 260 seconds00:18
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) left irc: Quit: Computer has gone to sleep.00:19
MjrTom (MjrTom@azureus/MjrTom) left irc: Disconnected by services00:23
MjrTom- (MjrTom@azureus/MjrTom) joined #pypy.00:23
Nick change: MjrTom- -> MjrTom00:23
Ademan (~yaaic@34.sub-97-159-164.myvzw.com) left irc: Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org00:26
mitchellh (~mitchellh@c-69-181-107-107.hsd1.ca.comcast.net) left irc: Quit: ZOMG BYE00:34
daniloaf (~daniloaf@187.114.217.96) left irc: Read error: Connection reset by peer00:36
_habnabit (~habnabit@python/site-packages/habnabit) joined #pypy.00:42
_pyre_1 (~stian@137.241.189.109.customer.cdi.no) joined #pypy.00:45
_pyre_ (~stian@137.241.189.109.customer.cdi.no) left irc: Ping timeout: 244 seconds00:47
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/181100:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/54500:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/59300:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/107300:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/77900:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/141100:49
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/67700:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/24300:49
ericflo (~ericflo@75.103.8.110) left irc: Quit: ericflo01:12
mcdonc (~mcdonc@cabana.palladion.com) left irc: Remote host closed the connection01:15
dmsk (dmsk@62.61.145.89.generic-hostname.arrownet.dk) left #pypy ("Leaving").01:22
whitelynx|work (~whitelynx@75.111.197.204) joined #pypy.01:31
whitelynx (~whitelynx@75.110.189.230) left irc: Ping timeout: 260 seconds01:35
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) joined #pypy.01:37
etrepum (~bob@accessnat4.mochimedia.net) left irc: Quit: etrepum01:40
etrepum (~bob@accessnat4.mochimedia.net) joined #pypy.01:44
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.01:46
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/11801:51
Moku (~John@osbk-4db0647a.pool.mediaWays.net) joined #pypy.01:53
etrepum (~bob@accessnat4.mochimedia.net) left irc: Read error: Operation timed out01:53
Nick change: Moku -> Guest3264601:53
Shinita (~John@osbk-4d087493.pool.mediaWays.net) left irc: Ping timeout: 240 seconds01:53
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/24301:57
_pyre_1 (~stian@137.241.189.109.customer.cdi.no) left irc: Quit: Leaving.02:00
nirbheek_ (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.02:04
DasIch_ (~dasich@p4FFDEBE3.dip.t-dialin.net) joined #pypy.02:08
DasIch (~dasich@p4FFDF3B2.dip.t-dialin.net) left irc: Ping timeout: 248 seconds02:11
twomashi (~Adium@201-213-122-142.net.prima.net.ar) left irc: Quit: Leaving.02:21
ernop (~ernie@114.250.112.247) joined #pypy.02:21
Nick change: DasIch_ -> DasIch02:22
nirbheek_ (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Quit: Leaving02:27
nirbheek_ (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.02:27
kvda (~kvda@124-171-36-124.dyn.iinet.net.au) left irc: Quit: -___-02:46
mat^2 (~mathias@212.130.113.35) left irc: 02:50
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/181102:58
Garen (garen@69.76.18.3) left irc: Remote host closed the connection03:07
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.03:07
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) left irc: Ping timeout: 258 seconds03:09
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 240 seconds03:11
Garen (garen@69.76.18.3) joined #pypy.03:13
lac_ (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.03:18
lac (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Ping timeout: 244 seconds03:19
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) joined #pypy.03:20
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) joined #pypy.03:20
mitchellh (~mitchellh@c-71-202-125-40.hsd1.ca.comcast.net) joined #pypy.03:36
justinpeel (~justinpee@c-76-23-44-72.hsd1.ut.comcast.net) left irc: Quit: Leaving03:41
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy03:49
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/141103:59
ernop (~ernie@114.250.112.247) left irc: Ping timeout: 258 seconds04:08
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/59304:09
gutworth (~benjamin@dhcp-128-36-89-186.central.yale.edu) left irc: Ping timeout: 260 seconds04:16
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/107304:17
ernop (~ernie@68.68.46.109) joined #pypy.04:21
EnCuKou (~encukou@a91-156-243-246.elisa-laajakaista.fi) joined #pypy.04:26
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/54504:30
nirbheek_ (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 255 seconds04:34
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/77904:40
mcdonc (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds05:10
JaredWigmore (~JaredW@118.142.17.238) left irc: Ping timeout: 245 seconds05:15
DanKluev (~root@83-232-adsl.vntc.ru) left irc: Ping timeout: 252 seconds05:16
EnCuKou (~encukou@a91-156-243-246.elisa-laajakaista.fi) left irc: Remote host closed the connection05:32
DanKluev (~root@83-232-adsl.vntc.ru) joined #pypy.05:33
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 248 seconds05:35
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.05:41
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/67705:49
asmeurer__ (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Quit: asmeurer__05:59
Nick change: zain -> m-m-m-matt06:06
Nick change: m-m-m-matt -> zain06:09
vak (3ef59c34@gateway/web/freenode/ip.62.245.156.52) left irc: Ping timeout: 265 seconds06:11
mwhudson (~mwh@linaro/mwhudson) left irc: Quit: Leaving06:18
JaRoel|4D (~jaroel|4d@sink.jaroel.nl) left irc: Remote host closed the connection06:24
mitchellh (~mitchellh@c-71-202-125-40.hsd1.ca.comcast.net) left irc: Quit: Computer has gone to sleep06:25
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) joined #pypy.06:32
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.06:39
EnCuKou (~encukou@193.167.65.94) joined #pypy.06:40
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Read error: Connection reset by peer06:42
eropple|Air (~ed@pool-74-104-187-201.bstnma.fios.verizon.net) left irc: Quit: eropple|Air06:43
k_bx (~k_bx@94.244.19.62) joined #pypy.06:46
JaRoel|4D (~jaroel|4d@office.fourdigits.nl) joined #pypy.06:56
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) joined #pypy.07:03
Taggnostr (~quassel@dyn57-215.yok.fi) left irc: Ping timeout: 258 seconds07:04
fijalmorning07:14
bgola (~bgola@c95185bc.virtua.com.br) left irc: Ping timeout: 255 seconds07:16
Taggnostr (~quassel@dyn57-215.yok.fi) joined #pypy.07:22
kenaan12brunogola default 114c56b2a60b5a 15/pypy/rlib/rsre/: [fixes issue 923] matching RegExp with optional zero-width assertion groups07:23
kenaan12brunogola default 11d4c7fe2ac048 15/pypy/rlib/rsre/rsre_core.py: test_search.py and test_zinternal.py passing07:23
kenaan12brunogola default 11caf9fa132b86 15/: merged with default branch07:23
bgola (~bgola@c95185bc.virtua.com.br) joined #pypy.07:24
fijalbgola: merged07:30
CIA-2503fijal 07roundup * 10#923/Regular expressions with optional look-ahead operations failing to match groups: [resolved] Merged, thanks! * 14https://bugs.pypy.org/issue92307:31
antocuni_ (~antocuni@host193-122-dynamic.0-79-r.retail.telecomitalia.it) joined #pypy.07:34
cfbolzAlex_Gaynor: ping?07:43
fijalmorning carl07:43
fijalI think he went to bed07:43
cfbolzok07:44
cfbolzthere *really* need to be tests for the new dict virtualization stuff in optimizeopt/test07:44
fijalhm I thought he wrote some07:44
fijalor you mean more tests?07:44
cfbolzlet me check07:44
cfbolzhm, why didn't I see them in the merge mail07:45
Action: cfbolz looks again07:45
cfbolzfijal: there is one07:45
fijalwell, I remember complaining07:45
cfbolzthere is not even one where the array escapes07:46
fijalok, then indeed07:46
fijaldid he also write test_pypy_c for the cases that are sped up?07:46
cfbolzyes, I think so07:46
Action: fijal ponders if he feels like debugging speed.pypy.org07:46
tlynn (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) joined #pypy.07:47
cfbolzfijal: but I think there should also be some about the cases that *don't* work07:47
fijalyou mean in optimizeopt or test_pypy_c?07:47
cfbolzboth07:47
fijalI agree about optimizeopt07:47
cfbolzbecause I am afraid that dict operations become slower because we try too aggressively to keep the dict virtual07:47
fijalunsure about test_pypy_c07:47
cfbolzthat could lead to bridges07:48
fijalah I see07:48
fijalwell, not completely but I see the potential problem07:48
k_bx (~k_bx@94.244.19.62) left irc: Ping timeout: 276 seconds07:50
lucian (~lucian@93-97-174-115.zone5.bethere.co.uk) joined #pypy.07:56
k_bx (~k_bx@94.244.19.62) joined #pypy.07:58
ronny*insert a day of ranting here* we cannot port pyrepl to python3, since there are bugs in curses wrt str vs bytes07:59
ronnythe bug is open since at least 2 years07:59
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) left irc: Quit: Textual IRC Client: http://www.textualapp.com/08:02
k_bx (~k_bx@94.244.19.62) left irc: Read error: Operation timed out08:03
lucian (~lucian@93-97-174-115.zone5.bethere.co.uk) left irc: Quit: Leaving08:06
fijalronny: pypy does not come with curses right?08:06
lucian (~lucian@93-97-174-115.zone5.bethere.co.uk) joined #pypy.08:06
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.08:06
teknico (~quassel@88-149-208-198.dynamic.ngi.it) joined #pypy.08:09
antocuni_ronny: do you mean that pyrepl cannot run on CPython 3?08:10
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) joined #pypy.08:16
MjrTom (MjrTom@azureus/MjrTom) left irc: Disconnected by services08:20
MjrTom- (MjrTom@azureus/MjrTom) joined #pypy.08:20
Nick change: MjrTom- -> MjrTom08:20
JaredW (~JaredW@118.142.17.238) joined #pypy.08:22
ronnyantocuni_: basically08:26
ronnyfijal: pyrepl falls back to _minimal_curses08:28
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 260 seconds08:28
fijalyes08:28
ronnyfijal: i suppose we can have acorrect _minimal_curses on pypy, but porting is tricky if the target is broken08:29
arigato (~arigo@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.08:32
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.08:32
kenaan12arigo stm 113a174d94e902 15/pypy/: Transform and compile to C code: first early version.08:33
kenaan12arigo stm 11c00c946fb09b 15/pypy/translator/: Fixes.08:33
fijalarigato: morning armin08:33
arigatomorning08:33
cfbolzarigato: can you think of a way to get at a config object in the jit optimizer?08:37
cfbolzok, got it08:39
cfbolzself.optimizer.metainterp_sd.config08:39
arigato:-_/08:40
arigatofwiw I'm fighting in the stm branch with, precisely, getting my hands on a config object08:40
khs (~khs@2001:700:300:2120:725a:b6ff:fee5:a44) joined #pypy.08:40
arigatothe whole driver.py is kind of nice and unusable because e.g. there is no way to control with a config object what steps to do08:40
arigatoat least I couldn't find any way08:41
cfbolzthere is none, I think08:41
arigatoI've stuck the stm transformation at the start of the "database" step in genc.py, and that "works"08:43
ojii (~ojii@40-34.60-188.cust.bluewin.ch) joined #pypy.08:43
fijalarigato: did you notice I found a bug in rpython's str.split?08:43
fijaler08:43
fijalstr.strip08:43
arigatono08:43
fijalcame as a surprise08:43
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.08:44
arigatoanyone has a clue why https://bitbucket.org/pypy/pypy/pull-request/13/fixes-issue-923#comment-673 is marked as "Accepted by arigo 9 minutes ago" when I only viewed this page 1 minute ago?08:44
fijalarigato: I think bitbucket's bug08:45
fijalI clicked "accept"08:45
arigatoI suppose we can find an obscure reason for why it's "arigo"08:46
ronnyreport a bug?08:48
arigatofijal: I suppose you have understood precisely why the bug was caused and why the fix is a good idea? :-/08:49
fijalarigato: I understood the issue and tests08:49
fijalI claim08:50
arigatoso basically no08:50
kenaan12cfbolz int-tag-untag-as-operations 11c0429bb24dd1 15/pypy/: use int_untag for untagging too (duh)08:50
kenaan12cfbolz int-tag-untag-as-operations 111bab583b7cdb 15/pypy/jit/metainterp/optimizeopt/: when the optimizer sees an int_tag operation, it can later remove the int_and(x, 1) operation ...08:50
fijalarigato: :)08:51
tlynn (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) left irc: Ping timeout: 240 seconds08:53
ronnyhmm08:57
ronnyseems like the _minimal_curses in lib_pypy is actually broken/unable to find curses08:58
EnCuKou (~encukou@193.167.65.94) left irc: Read error: Connection reset by peer09:00
cfbolzarigato: I looked at strip, and it seems pretty clear that the implementation did not take into account that the result could be empty09:01
cfbolzthat's what maciek's fix is about09:01
fijalcfbolz: I think he's talking about pull request with res09:02
cfbolzah09:03
kenaan12fijal numpy-minilang 117d18eb124a06 15/pypy/module/micronumpy/: function call and move one more test09:03
cfbolzfijal: anyway, the fix should go to trunk09:06
fijalcfbolz: yes, but I'm planning to merge my branch by today09:06
cfbolzadvanced09:06
zain (~zain@69.38.221.130) left irc: Ping timeout: 258 seconds09:07
Nick change: zain__ -> zain09:07
mvt (~mvantelli@87.213.45.85) joined #pypy.09:09
Action: fijal is happy with test_zjit09:09
kenaan12RonnyPfannschmidt pyrepl[py3ksupport] 1119d9ded8abdc 15/pyrepl/: add a modified version of _minimal_curses, get to the point of kind of running pythoni on p...09:12
ronnyit works, but dont try backspae :P09:12
ronnyor up/down :P09:13
pedronisarigato, I suppose driver could be refactored to do that, it's a bit beyond what was designed for at the time09:18
stakkars (~tismer@p5DDB5E98.dip.t-dialin.net) left irc: Read error: Connection reset by peer09:20
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood09:21
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.09:22
stakkars (~tismer@p5DDB5E98.dip.t-dialin.net) joined #pypy.09:22
whitelynx|work (~whitelynx@75.111.197.204) left irc: Quit: Ex-Chat09:25
arigatopedronis: yes, I imagine09:26
arigatoa bit hard to find motivation, given that with 3 lines I could just hack it into the "database" step09:27
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 255 seconds09:30
fijal*cough*09:30
fijalarigato: http://xkcd.com/292/09:30
arigato:-)09:30
cfbolzso, the world keeps being alright09:34
cfbolztagged pointers are still slow in pypy09:34
arigato:-/09:35
cfbolzor at least not faster09:35
arigatook09:35
arigatobut they are with prolog?09:35
cfbolzhaven't dared to measure09:35
fijalAlex_Gaynor: ping09:37
fijalcfbolz: paper!09:37
EnCuKou (~encukou@193.167.65.94) joined #pypy.09:38
cfbolzfijal: need more data09:40
cfbolzfijal: but yes, that's the plan09:40
kenaan12RonnyPfannschmidt pyrepl[py3ksupport] 1161d643893b64 15/pyrepl/python_reader.py: use lines of unicode_escaped strings for history - its multi-python safe and more readable09:47
Arach (~arach@2.95.138.114) left irc: Remote host closed the connection09:48
Arach (~arach@2.95.138.114) joined #pypy.09:53
Action: fijal finds more bugs in rpython09:55
verte (~verte@python/site-packages/verte) joined #pypy.09:56
Trundle (~andy@p578bfdcf.dip0.t-ipconnect.de) joined #pypy.09:59
Trundle (~andy@p578bfdcf.dip0.t-ipconnect.de) left irc: Changing host09:59
Trundle (~andy@python/site-packages/trundle) joined #pypy.09:59
fijalgood10:03
kenaan12fijal numpy-minilang 1175b8ab99578c 15/pypy/rpython/: more bugs in strip10:04
fijalnow I found a real performance issue with our handling of max10:04
kenaan12fijal numpy-minilang 11312b01650172 15/pypy/module/micronumpy/: work more on test_zjit. now it shows a problem in our handling!!! (wow)10:04
fijalgood10:05
kenaan12arigo stm 11cf109f42d6ab 15/pypy/translator/stm/: 'transaction_boundary' is better for the transformer.10:05
kenaan12arigo stm 11346f0c8a0ed7 15/pypy/translator/stm/: Progress10:05
kenaan12fijal numpy-minilang 112518fc31f52d 15/pypy/module/micronumpy/test/test_zjit.py: remove ported tests10:06
Guest32646 (~John@osbk-4db0647a.pool.mediaWays.net) left irc: Read error: Connection reset by peer10:06
Shanita (~John@osbk-4d087ff6.pool.mediaWays.net) joined #pypy.10:09
\toothrot (~mux@unaffiliated/htoothrot) left irc: Ping timeout: 260 seconds10:13
htoothrot (~mux@unaffiliated/htoothrot) joined #pypy.10:16
arigato (~arigo@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Ping timeout: 244 seconds10:16
k_bx (~k_bx@195.20.130.1) joined #pypy.10:25
lizardo (~lizardo@189.2.128.130) joined #pypy.10:36
aleksi (~aleksi@85.235.191.82) joined #pypy.11:04
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) left irc: Quit: Computer has gone to sleep.11:14
ernop (~ernie@68.68.46.109) left irc: Ping timeout: 258 seconds11:15
antocuni_ (~antocuni@host193-122-dynamic.0-79-r.retail.telecomitalia.it) left irc: Ping timeout: 260 seconds11:19
Vorpal (~AnMaster@unaffiliated/anmaster) joined #pypy.11:20
kenaan12fijal numpy-minilang 11409b83704acb 15/pypy/: minor tweaks and port more tests11:23
antocuni_ (~antocuni@host130-123-dynamic.2-87-r.retail.telecomitalia.it) joined #pypy.11:31
DanKluev (~root@83-232-adsl.vntc.ru) left irc: Ping timeout: 255 seconds11:35
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 240 seconds11:49
Furao (~Furao@pc-108-250-30-200.cm.vtr.net) joined #pypy.11:50
DanKluev (~root@78-24-231-218-udsl.vntc.ru) joined #pypy.11:50
Furaohello, I just installed pypy from ubuntu ppa, and when I run buildout to setup my project, I get many unknown url type: https -- Some packages may not be found11:51
Furaolike if SSL is not enable and make https:// unusable11:52
Arach (~arach@2.95.138.114) left irc: Ping timeout: 256 seconds11:53
ronnyyou might want to ask the maintainer of the ppa11:53
bgolaFurao: are you using oneiric ?11:54
Furaoyes11:55
Furaobgola, yes i'm in oneiric11:55
Furaoldd /usr/bin/pypy show it's dynamically linked to openssl11:55
bgolaFurao: and it is, indeed. does it work with the official binary release ?11:57
Furaodid not try :)11:57
bgolacan you try it ? or maybe help me to reproduce the error so I can try 11:59
Furao/usr/local/bin/pypy: error while loading shared libraries: libssl.so.0.9.8: cannot open shared object file: No such file or directory12:00
Furaothis is with official linux bin12:00
whyking (~quassel@68.1.185.182) joined #pypy.12:00
Furaolucky, there is a libssl0.9.8 package12:00
Furaothey packaged older version as well12:01
gutworth (~benjamin@dhcp-128-36-89-186.central.yale.edu) joined #pypy.12:01
bgolaFurao: hm, the ppa package is linked to the new version probably because there is no libssl0.9.8-dev 12:02
Furaothere is not https errors with official bin12:03
bgolaand the new version removes support for SSLv2_method and that should be the only difference between the official binary package and the ppa one12:03
Furaobgola: you are package maintainer?12:04
Furaoppa I mean12:04
Furaoah yes12:05
Arach (~arach@2.95.27.244) joined #pypy.12:06
fijalbgola: thx for ppa btw12:06
FuraoSSLv2_method is unsafe with pypy?12:07
bgolaFurao: it's unsafe, they removed in libopenssl1.0.0 12:08
bgolafijal: :)12:08
fijalFurao: pypy has nothing to do with it12:08
fijalFurao: it's debians/ubuntus policy12:08
fijaland it's unsafe no matter what really12:08
Furaook12:09
Furaosomething else rely on this to handle https :)12:09
fijalbgola: are there download stats btw?12:12
bgolaFurao: how can i reproduce the message?12:12
bgolafijal: only using the launchpad API12:12
fijalnot on the website?12:12
bgolafijal: no :(12:12
antocuni_ (~antocuni@host130-123-dynamic.2-87-r.retail.telecomitalia.it) left irc: Quit: Leaving12:12
bgolafijal: http://blog.launchpad.net/cool-new-stuff/tracking-ppa-download-statistics12:13
Furaobgola: in my case, I set a virtualenv, installed zc.build and ran buildout and as soon as it get https:// url to download package source it fail12:14
bgolahm, simply trying to urllib2.urlopen("https://google.com/") fails too12:16
senyai (3ead87cd@gateway/web/freenode/ip.62.173.135.205) joined #pypy.12:16
fijalbgola: do you feel like digging there?12:16
Action: fijal is busy killing zombies12:16
bgolafijal: yes, will take a look12:16
fijalnot really actually12:17
lmoura (~lmoura@187.58.100.139) left irc: Ping timeout: 240 seconds12:17
bgolafijal: meh, it was easy (and already fixed on newer pypy)..12:25
bgolaFurao: tks for the report12:26
witulski (~stupsi@fwstups.cs.uni-duesseldorf.de) joined #pypy.12:27
witulski (stupsi@fwstups.cs.uni-duesseldorf.de) left #pypy.12:27
fijalbgola: ?12:27
fijalyou mean to have SSL at all without v2?12:27
kenaan12fijal numpy-minilang 11c1f7d371f290 15/pypy/module/micronumpy/: make old tests pass (just because).12:27
fijalmeh, we should release a newer pypy :/12:27
lmoura (~lmoura@187.113.101.120) joined #pypy.12:30
timonatorfijal: i'll write those docstrings at once, so that you can release immediately ;)12:30
Furao (~Furao@pc-108-250-30-200.cm.vtr.net) left irc: Ping timeout: 240 seconds12:30
Furao (~Furao@pc-108-250-30-200.cm.vtr.net) joined #pypy.12:31
bgolafijal: yes, with v23 or v3 12:31
fijalbgola: ok12:31
fijalFurao: get a nightly then12:31
bgolai'll patch the ppa package and upload a new version12:33
Furaogreat thanks12:33
timonatorany preferred docstring style?12:34
fijaltimonator: copy the numpy ones?12:36
timonatorwill do12:37
timonatoroh, they are looooong12:38
timonatoralso, for average - for instance - we don't currently implement all of the functionality. how do i handle those cases?12:39
fijaltimonator: clear error would be good12:39
fijalraise NotImplementedError("stuff")12:39
timonatorOK. full docstring copy and errors.12:40
timonatormaybe a note in the docstring about what's not implemented?12:40
fijalif it raises that's fine I guess12:40
fijaldocstrings tend to fall out of favor12:41
timonatori hope doctests are not run as part of the pypy test suite12:41
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.12:43
JaRoel|4D (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection12:43
tomh (u501@gateway/web/irccloud.com/x-kizayexssboeecpg) left irc: Excess Flood12:43
tomh (u501@gateway/web/irccloud.com/x-veldheecfgpivexj) joined #pypy.12:43
timonatoroh, my locally installed numpy, where i was stealing the docstrings from, is older than the current documentation - what numpy do we target? trunk?12:48
fijalyes12:49
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.12:49
timonatorgood12:49
kenaan12fijal default 11c91b4f3c204f 15/pypy/: Merge numpy-minilang, this refactors test_zjit to use new cool mini-language instead of hardcoding things.12:51
kenaan12fijal numpy-minilang 117a58d06641f3 15/: merged12:51
santagada (~leonardo@201.47.210.238.dynamic.adsl.gvt.net.br) joined #pypy.12:53
santagada (leonardo@201.47.210.238.dynamic.adsl.gvt.net.br) left #pypy.12:54
EnCuKou (~encukou@193.167.65.94) left irc: Remote host closed the connection13:09
Furao (Furao@pc-108-250-30-200.cm.vtr.net) left #pypy ("Leaving").13:15
fprimex (~fprimex@brent-macbook.sc.fsu.edu) joined #pypy.13:16
fprimexhi all13:16
fprimexgiving a talk on python today and wanted to mention pypy. I know this will come up, and I've read all of the info I can find on it...13:16
fprimexcan micronumpy be installed into an existing pypy 1.6 install, or does it require the branch?13:17
lucianfprimex: the latter13:17
fprimexlucian: thanks13:17
cfbolzlucian: no, wait13:17
luciancfbolz: really?13:17
cfbolzfprimex: do you have a 1.6 around?13:17
masqueradeThe branch was merged upstream I thought13:18
cfbolzfprimex: I think you can just say "import numpy" in it13:18
luciancfbolz: oh, you can13:18
fprimexcfbolz: just compiled and installed it last night from a fedora src.rpm13:18
cfbolzbut at that point things were not too good yet13:18
lucianstupid me. sorry13:18
fprimexcfbolz: well, how about that...13:20
fprimex:D13:20
Rhy0lite (dje@nat/ibm/x-uwclkuvdcffdjxal) joined #pypy.13:20
Action: lucian crawls into a dark hole13:20
cfbolzlucian: no worries13:21
fprimexnow to figure out what I can and can't really do with it and run this sample code. I already see that it doesn't have some ways of declaring arrays, but I'm not sure how old the source packaged in with this srpm was13:21
fprimexcfbolz: thanks!13:21
cfbolzfprimex: for more features try the nightly13:21
fprimexcfbolz: awesome, I will see about building that after my talk, which is in 2 hours13:22
fprimexcfbolz: oh, also, do you know how much disk space a pypy build takes?13:22
fprimexI build in a RAM disk on a node in our cluster and it filled up the 3.2GB that were free there13:22
lucianfprimex: the nightly is usually enough13:22
fprimexof course, my src.rpm is using python to build, and not pypy13:22
fijal fprimex bad ide13:27
fijalalso, why not download nightly?13:27
fijalthe nightly has by far more features, like dtypes13:27
fprimexfijal: if the nightly also corrects the libssl/libcrypto linking issue I can use it13:28
lmoura (~lmoura@187.113.101.120) left irc: Read error: Connection reset by peer13:28
fijalfprimex: as in uses your .so names and not mine .so names?13:29
fprimexfijal: yeah, I think. I got complaints about needing libssl.0.9.8, when on RHEL6 I had installed the libssl.0.9.8e compat package13:29
fprimexwhen I tried to finagle a symlink in my home directory and manipulate LD_LIBRARY_PATH, it then complained about missing version information13:30
fijalI guess indeed you can't use nightly then13:30
fprimexI can rebuild though13:30
fprimexit's no big deal. kind of my job around here13:30
fijalbinary releases on linux are a mess :(13:30
fijalok13:30
canta (~canta@77-20-123-240-dynip.superkabel.de) joined #pypy.13:31
fprimexit would help, though, if I knew how much space the build ultimately would need13:31
fprimexI can grow our build node or clean up space on it13:31
cfbolzfprimex: a few gig13:31
fprimex>3.2GB is what I'm finding13:31
fprimexfilled up the RAM disk13:31
fijalon /tmp though13:32
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.13:33
cfbolzit's all a bit extreme :-(13:33
fijalcfbolz: pypy requirements?13:34
cfbolzyes13:34
fijalyes, totally13:34
fprimexI will see about building on disk and get it figured out13:34
fijalnot much we can do now though13:35
fprimexhas anyone been doing parallel computing with pypy? as in, with ipython or mpi4py?13:35
fijalnot that I know about13:35
fijalwe focused on single-core performance so far13:35
fijalsince we're getting ~30x speedups on numeric stuff, there is still a way to go on that front ;-)13:36
fprimexsure :)13:36
fprimexwell, I would be happy to test anything in that direction if I can get pypy going on our cluster13:36
cfbolzfprimex: cool13:36
fprimexand since I already have a working 1.6 build it is looking promising13:36
fijalbut yeah, we were even thinking about doing multiple-thread parallelism for numpy13:37
fijalfprimex: cool!13:37
fprimexgetting mpi4py working would probably be plenty13:37
fijalfprimex: if .so names are a problem, I can probably ship you C sources13:37
fprimexalmost anyone here doing threading or openmp are dropping to C through cython13:37
fijalthat does not come as a surprise13:38
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) joined #pypy.13:39
lmoura (~lmoura@187.113.110.9) joined #pypy.13:45
kenaan (~kenaan@cube.twistedmatrix.com) left irc: Remote host closed the connection13:45
derdon (~derdon@pD9E1D83B.dip.t-dialin.net) joined #pypy.13:48
lac (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.13:48
lac_ (~quassel@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Ping timeout: 244 seconds13:48
whitelynx (~whitelynx@li117-47.members.linode.com) joined #pypy.13:53
fprimex (~fprimex@brent-macbook.sc.fsu.edu) left irc: Quit: back after presentation13:55
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection13:58
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.14:00
whyking (~quassel@68.1.185.182) left irc: Ping timeout: 276 seconds14:02
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection14:07
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) joined #pypy.14:18
amaury (~amaury_@74.125.57.34) joined #pypy.14:22
mcdonc (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds14:35
kenaan (~kenaan@cube.twistedmatrix.com) joined #pypy.14:47
kenaan12cfbolz int-tag-untag-as-operations 114c4642283aff 15/pypy/jit/metainterp/blackhole.py: move an ll_assert around: cast_ptr_to_int *can* be used on an even pointer (to check whether i...14:48
verte (~verte@python/site-packages/verte) left irc: Quit: ~~~ Crash in JIT!14:51
Alex_Gaynorcfbolz, fijal: pong14:54
EnCuKou (~encukou@a91-156-243-246.elisa-laajakaista.fi) joined #pypy.14:54
cfbolzAlex_Gaynor: I think there need to be a few more tests about the new virtualization of arrays of structs in optimizeopt/test14:57
cfbolzat least one where the array actually escapes14:57
Alex_Gaynorok, will do14:57
cfbolzAlex_Gaynor: and maybe a test_pypy_c test that shows that normal dict creation (ie without constant keys) still works as expected15:00
Alex_Gaynorcfbolz: you mean a test like for i in xrange(1000): d = {}; d[i] = str(i) or something?15:01
cfbolzAlex_Gaynor: yes15:01
Alex_Gaynorok, I can do that, thanks15:01
cfbolzAlex_Gaynor: e.g. if you had gotten the unroll_safe wrong, would should notice then15:02
amaury (~amaury_@74.125.57.34) left irc: Ping timeout: 240 seconds15:04
Alex_Gaynorfijal: why are we using -> as a subscript operator?15:05
fijalwhy not?15:05
Alex_Gaynorit looks silly15:05
Alex_Gaynorand it's inconsistant with setitem15:05
Alex_Gaynorfijal: why are tehre still py.test.skip() broken in the merged stuff?15:06
fijaluh15:07
fijalit seems I broke few things :/15:07
Alex_GaynorI suppose branches shouldn't break things ;)15:07
fijaleh only this one15:08
Alex_Gaynor:)15:08
fijalAlex_Gaynor: ah I wanted one more thing15:08
fijalmax/min are broken15:08
Alex_Gaynorthe code or the tests?15:09
fijalthe code15:09
Alex_Gaynorhow?15:09
fijalat least test_zjit was not testing what it was supposed to test15:09
fijallook at skipped test jit15:09
fijaland run them with --viewloops15:09
kenaan12fijal numpy-multidim 1164aadf44e926 15/: merge default - tests still broken, working on it15:10
kenaan12fijal default 118b3dc9ba02ca 15/pypy/module/micronumpy/test/test_compile.py: forgotten import15:10
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.15:10
Alex_Gaynorthe tests look basically sane before15:11
asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy.15:12
fijalno15:13
mvt (~mvantelli@87.213.45.85) left irc: Quit: Leaving15:13
fijalthey were just not testing what they were supposed to15:13
Alex_Gaynorfijal: btw, what is arraygetsubstruct?15:14
fijalno idea15:14
Alex_Gaynorit's some SpaceOp, it seems to be generated by my_raw_array_of_structs[2].val15:14
khs (~khs@2001:700:300:2120:725a:b6ff:fee5:a44) left irc: Quit: Leaving15:17
Alex_Gaynorcfbolz: do you know what arraygetsubstruct is?15:17
cfbolzwhere do you see it?15:18
aleksi (~aleksi@85.235.191.82) left irc: Remote host closed the connection15:19
Alex_Gaynorcfbolz: In my numpy-complex branch, it seems to come from my_raw_array_of_structs[2].field15:20
cfbolzAlex_Gaynor: grepping for it on trunk yields nothing15:20
cfbolzso how would I know where it comes from?15:20
Alex_Gaynorerr, sorry getarraysubstruct15:20
cfbolz"    # the diff between op_getarrayitem and op_getarraysubstruct15:21
cfbolz    # is the same as between op_getfield and op_getsubstruct"15:21
Alex_GaynorI got that far, do you know the difference ebtwee getfield and getsubstruct?15:22
cfbolzyou need it when array elements aren't ptrs to structs, but structs15:22
cfbolzfor getfield it's the same thing15:22
cfbolzbut I am confused, I thought they are both superseded by getinteriorfield15:22
Alex_Gaynoris getinteriorfield only for GC ptrs? and if so, is tehre a reason not to make it so?15:24
arigato (~arigo@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.15:24
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.15:24
lucian (~lucian@93-97-174-115.zone5.bethere.co.uk) left irc: Read error: Connection reset by peer15:25
JaRoel (~jaroel|4d@office.fourdigits.nl) joined #pypy.15:26
cfbolzAlex_Gaynor: no clue right now, need to catch a train15:26
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) left irc: Quit: Leaving15:26
Alex_Gaynorbye15:26
fijalarigato: any idea why stacklets require some weird stuff from C that's not under windows?15:26
arigatofirst time I hear that15:27
fijalhttp://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/240/steps/translate/logs/stdio15:27
fijalmodule_0.obj : error LNK2019: unresolved external symbol _slp_switch_raw referenced in function _slp_switch15:27
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.15:27
fijalbecause importing pypy.rlib.rstacklet raised CompilationError15:28
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 260 seconds15:29
arigatobah, it's just because we need to compile and link another file on windows15:29
arigatothere is a separate switch_x86_msvc.asm15:29
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.15:31
Action: arigato doesn't feel like finding out how to link a .asm file on Windows15:31
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 240 seconds15:32
arigatohelp, anyone? :-)15:32
kenaan12alex_gaynor numpy-complex 118f899b94cb87 15/: merged default15:34
kenaan12alex_gaynor numpy-complex 110c5959cc2de4 15/pypy/: more tests for array of structs15:34
Alex_Gaynorsigh, did I put this on the wrong branch...15:34
kenaan12alex_gaynor default 1111d7ea62944e 15/pypy/: more tests for array of structs15:35
kenaan12alex_gaynor default 11d2376dc9d29a 15/pypy/: merged upstream15:35
Alex_Gaynorarigato: do you know what getarraysubstruct is?15:35
arigatoyes, it returns an interior pointer to an array item (for arrays of structs)15:36
arigatoit should not really be used any more, for GcArrays15:36
arigato{get,set}interiorfield should be used instead15:36
Alex_Gaynoris there a reason getinteriorfield_raw isn't used instead?15:36
arigatofor raw stuff it's still useful15:37
fijalhm15:37
fijalshould space.issequence_w(reversedrangeiterator) return True?15:37
Alex_Gaynorit isn't subscriptable, so I claim no15:37
Alex_Gaynorarigato: hmm, supporting it in the JIT will be a mess I suppose15:38
arigatonot really15:38
fijalI thought we killed reversedrangeiterator15:38
fijalthen how am I supposed to determine whether something is a sequence or not?15:38
arigatoAlex_Gaynor: if it's also only about raw stuff, then getarraysubstruct_raw is just some simple arithmetic15:38
fijalarray(iterator)15:38
fijalshould return array of a list15:38
Alex_Gaynorfijal: no it doesn't, not on numpy at least15:39
fijalpfff15:39
fijalthen why the fuck we have a test that checks it?15:39
fijalAlex_Gaynor: that's what I meant by -A15:39
Alex_Gaynorfijal: you wrote that test, that's all I'm saying15:39
fijalpfff15:39
Alex_Gaynorfijal: give me two minutes and then I'll fix the True/False ones, I want to finish speaking with arigato about something15:40
fijalyes, but if they all passed with -A15:40
fijal...15:40
Alex_Gaynorarigato: ah, it just returns a ptr, I suppose that's easy enough15:40
arigatoAlex_Gaynor: yes, a pointer to the start of the Nth substructure in the array15:40
Alex_Gaynorarigato: and then I suppose there is some substructure_getfield or something?15:40
arigatono, a regular getfield_raw15:41
Alex_Gaynoroh, this sounds almost easy15:41
Alex_Gaynor:)15:41
arigatoyes :-)15:41
Alex_GaynorI suppose you can be more efficient if you combine the two operations15:41
Alex_Gaynorthe precise efficiency of numpy complex arrays isn't my biggest concern ATM though15:42
arigatoI suppose, but just wondering, do we even have such non-gc arrays of structs?15:42
arigatoah I see15:42
Alex_Gaynornumpy complexes will be15:42
Alex_Gaynornumpy is the mother of all raw-memory innovations :)15:42
arigatook :-)15:42
Alex_Gaynorok time to fix -A tests for fijal 15:43
Action: fijal fights with various random things15:43
fijalI guess dinner & cleaning though first15:44
Alex_Gaynorfijal: "random bugs in random [not a funny joke]"15:44
amaury (~amaury_@74.125.57.34) joined #pypy.15:47
G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving.15:51
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Read error: Operation timed out15:55
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) joined #pypy.15:56
Trundle (~andy@python/site-packages/trundle) left irc: Remote host closed the connection15:57
kenaan12alex_gaynor default 111847537fd4b5 15/pypy/module/micronumpy/: make more of the micronumpy tests pass under -A16:01
Alex_Gaynorfijal: ok the situation is improved, the remaining tests are either a) str/repr issues, b) trying to use numpy.array in isinstance, on real numpy array is just a function which returns a numpy.ndarray.  I suggest we copy that and switch them to be ndarray.  Shall we do that on your branch?16:01
lmoura (~lmoura@187.113.110.9) left irc: Ping timeout: 258 seconds16:13
amaury (~amaury_@74.125.57.34) left irc: Ping timeout: 240 seconds16:17
fijalAlex_Gaynor: yeah I guess so16:18
fijallet me finish fixing tests16:18
Alex_GaynorI'll never get in the way of that16:18
lmoura (~lmoura@186.215.206.130) joined #pypy.16:19
kenaan12arigo stm 11f0ea7da7446e 15/pypy/translator/: begin_inevitable_transaction.16:20
kenaan12arigo stm 113ec86f341cf6 15/pypy/translator/: Progress.16:20
kenaan12arigo stm 11451f6144a150 15/pypy/translator/: Random progress.16:20
kenaan12arigo stm 11f647b24f79f3 15/pypy/: Add the missing operations.16:20
kenaan12arigo stm 11c30efe18be88 15/pypy/translator/stm/: bug fix16:20
kenaan12arigo stm 113126dbb549ce 15/pypy/translator/stm/: Test that the stm mode falls back to "inevitable_transaction" when we see an unsupported operation like raw_malloc.16:20
k_bx (~k_bx@195.20.130.1) left irc: Write error: Broken pipe16:21
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Quit: Leaving16:23
lmoura (~lmoura@186.215.206.130) left irc: Ping timeout: 248 seconds16:24
fschwidom (~fschwidom@92.117.90.6) joined #pypy.16:29
fprimex (~fprimex@brent-macbook.sc.fsu.edu) joined #pypy.16:30
kenaan12fijal numpy-multidim 11cfc66011fc77 15/pypy/module/micronumpy/: fix tests broken during the merge16:31
fijalfprimex: how did the talk go?16:31
fijalAlex_Gaynor: fixed, modulo test_zjit16:31
fprimexfijal: good. small but very interested group. most of them actually friends of mine from the math dept here16:31
TkTech (TkTech@tkte.ch) left #pypy.16:31
fprimexon person in particular has some code that takes a week to execute. I'm going to help her with it some, probably16:31
Alex_Gaynorfprimex: that's a lot of time playing with swords (http://xkcd.com/303/)16:32
fprimexAlex_Gaynor: I know, right? I think she spends it teaching and researching, but I don't doubt she would love some swords16:34
Alex_Gaynoreveryone loves swords!16:34
fijalfprimex: if you want to run it on pypy and want some advice don't hesitate to come here and ask16:34
fprimexfijal: that is the direction I'm heading, but I'm not sure if she's ready for all this yet16:35
fprimexstill figuring out synaptic and how to install numpy :P16:35
Action: fijal is also still figuring out synaptic and how to install numpy16:37
Alex_Gaynorpip!16:38
fijalAlex_Gaynor: pip won't install blast for me16:38
Alex_Gaynorsudo apt-get install libblas16:38
lmoura (~lmoura@187.113.67.49) joined #pypy.16:39
Alex_Gaynorfijal: where would I put a test for supporting getarraysubstruct in the JIT?16:39
Alex_GaynorI guess I should at least have a jtransform test to make it getarraysubstract_raw16:40
fijalAlex_Gaynor: have you ever tried installing numpy on a mac?16:40
Alex_Gaynornope16:40
fijalAlex_Gaynor: you also need a bunch of other tests16:41
fijallike test_runner in backend16:41
Alex_Gaynorfijal: this is where I'll start though, one at a time :)16:41
fijalyeah16:41
fijalthis is silly :(16:41
fijalfor all real backends you would emit a bunch of stuff in jtransform and be happy16:42
fijalbut that won't work with llgraph16:42
Alex_GaynorI figured I'd make it a real resop16:42
kenaan12arigo stm 111cffa0d605b4 15/pypy/translator/stm/: - mallocs of GC objects is supported by the STM system (for now   with Boehm)  - fix: don't store ophandler on self.__c...16:42
Alex_Gaynoris there a reason not to?16:42
fijalother than editing 7 levels instead of 1 no16:42
fijalit's also worse for the optimizer maybe16:42
fijalbut maybe not16:42
Alex_Gaynorit's raw memory, does the optimizer ever do anything with it?16:43
fijalnot yet16:43
arigatoyes, it would, because it's just integer additions16:43
arigatoraw memory = integer16:43
arigatofor the optimizer16:43
fijalthen maybe indeed not make it a resop16:44
fijalbecause it's nonsense16:44
arigatobut that's not supportable on llgraph16:44
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) left irc: Ping timeout: 258 seconds16:44
Alex_Gaynorpff, stupid llgraph :/16:44
Alex_Gaynorso I guess I need a resop after all16:44
arigatoor maybe it is16:44
arigatoif you feel like playing half a hour with random stuff like llmemory.sizeof() returning symbolic integers16:45
fijalif you're going to make a resop just for llgraph then I claim it's a dumb idea16:45
fijalllgraph should play along16:45
Alex_GaynorI'll try it without a resop for now, it should basically just be v1, v2: return v1 + (v2 * sizeof_element)?16:45
arigatoAlex_Gaynor: yes, assuming a nolength raw array16:46
Alex_Gaynorarigato: those are the only kind the JIT supports so far for raw arrays, so "good"16:46
fijalarigato: does the JIT know about raw arrays *with* length?16:46
arigatoyes16:46
arigatofijal: no16:46
fijalE                   TypeError: <GcStruct tuple2 { item0, item1 }> instance field 'item1':16:47
fijalE                   expects <* GcStruct object { typeptr }>16:47
fijalE                       got <* GcStruct pypy.interpreter.baseobjspace.W_Root { super }>16:47
fijalfun16:47
fijalI'm sure someobjects are not allowed by llinterp16:47
fijalso why did I get such a generic thing?16:47
fijalarigato: feel like helping?16:48
arigatofijal: maybe, if you feel like explaining at least what's the context16:49
fijalarigato: hg up numpy-multidim16:49
fijalpy.test pypy/module/micronumpy/test/test_zjit.py16:49
gtaylor (~gtaylor@108-196-160-120.lightspeed.tukrga.sbcglobal.net) joined #pypy.16:49
arigatothe context of the error16:49
fijalI'm faking space16:49
fijaland somehow llinterp explodes16:49
arigatothe faking is maybe not good enough16:50
fijalhttp://paste.pocoo.org/show/499638/16:50
fijalwell sure16:50
fijalbut how could it ever translate and end up in such an error?16:50
fijalit seems like a perfectly fine RPython to me16:50
arigatoah, dinner time16:50
arigatosorry16:50
arigatonote that llinterp is more picky than a C compiler about the types16:51
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) left irc: Quit: fzzzy16:51
arigatothe C compiler will just emit a warning, or maybe not even16:51
arigatofor the case you pasted above16:51
fijalok16:51
fijalbon appetit I gues16:51
arigatothanks16:52
fijalor apettit I never remember16:52
fijalah16:54
fijalAlex_Gaynor: it's you!\16:54
Alex_Gaynorfijal: what'd I do?16:54
fijalyou implemented enumerate wrongly16:54
fijalsomehow :)16:55
fijalthere is a missing cast at the very least16:56
Alex_Gaynoreh, I didn't implement enumerate at all, AFAIK16:56
fijal*somewhere*16:56
fijalso who did it?16:56
Alex_GaynorRPython enumerate?16:56
Alex_Gaynoreh, what do I need to do to get an integer size for a struct?16:57
fijalrffi.sizeof16:57
fijaldepends on the level ;-)16:57
kenaan12fijal numpy-multidim 11488dbd72cd7b 15/pypy/module/micronumpy/: make test_zjit pass (up to the point anyway)16:57
fijalwow16:58
fijalI'm impressed that they pass16:58
Alex_Gaynorfijal: rffi.sizeof isn't enough16:58
Alex_Gaynorfijal: this is in jtransform16:58
fijalno?16:58
fijalah yes16:58
fijalin jtransform there are things16:58
Alex_Gaynorno, it returns a symbolic still16:58
fijalwhy do you need an int?16:58
fijalno16:58
fijalrffi.sizeof always returns an integer16:58
fijalit's not llmemory.sizeof16:59
Alex_Gaynorno, it really doesn't16:59
Alex_Gaynorhttp://paste.pocoo.org/show/499642/16:59
fijal"""Similar to llmemory.sizeof() but tries hard to return a integer16:59
fijal    instead of a symbolic value.16:59
fijal    """16:59
fijalwell, it seems it tried17:00
Alex_Gaynorand failed17:00
Alex_Gaynorlooking at the code, it doesn't try very hard for structs17:00
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.17:00
fijalwhat's op.args[0].concretetype.TO.OF17:00
fijal?17:00
Alex_Gaynorlltype.Struct("S", ("x", lltype.Float), ("y", lltype.Float))17:00
fijalindeed it didn't try very hard17:00
fijalbut in jtransform there are other ways17:00
Alex_Gaynorhow?17:00
fijalwhy do you need an integer anyway?17:00
fijalbtw, have you written test in test_optimizeopt?17:01
Alex_Gaynorno, so far just a jtransform test17:01
Sho_ (~EHS1@kde/hein) joined #pypy.17:01
Alex_GaynorI'm not sure how you write an optimizeopt test for this anyways17:01
Alex_Gaynorthere's nothing in the optimizers17:01
dimazest (~dimazest@balticom-202-59.balticom.lv) left irc: Read error: Connection reset by peer17:01
fijalno, for virtual dicts17:02
fijalwhy do you need an integer?17:02
fijalcan't you pass a symbolic?17:02
Alex_Gaynoryes, I wrote the tests carl suggested for that17:03
Alex_Gaynorfijal: I dunno, can you pass symbolic?17:03
fijalI think so17:04
fijaleveryone should be relatively happy17:04
Alex_Gaynorpff, how do you test that it returned the right thing17:04
Alex_Gaynoryou can't compare a symbolic17:04
fijalyou can look into it17:05
Alex_Gaynorfijal: http://paste.pocoo.org/show/499646/ look correct?17:06
Alex_Gaynornow I write a test_ajit I suppose?17:06
fijalI dunno write some test in test_ajit17:06
fijalyes, precisely17:06
fijalI never know if test_jtransform is correct until I run the thing17:06
fijalbut then you don't need a test_runner at least17:06
fijalmaking the llgraph backend happy might be fun :/17:07
Alex_Gaynorfijal: pff, you cannot pass an llmemory.ItemOffset as an arg to SpaceOperation in jitcode17:13
Alex_Gaynorso I guess I'm back to "how do you convert this to an integer"17:14
fijalpff17:15
fijalteach rffi.sizeof how to deal with structs I gues17:15
fijala bit annoying17:15
Alex_GaynorI wonder if that isn't the wrong level :/17:15
tilgovi (~randall@adsl-71-138-140-255.dsl.pltn13.pacbell.net) joined #pypy.17:15
tilgovi (~randall@adsl-71-138-140-255.dsl.pltn13.pacbell.net) left irc: Changing host17:15
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.17:15
Alex_Gaynorhow do you force it to compute the result anyways?17:16
ojii (~ojii@40-34.60-188.cust.bluewin.ch) left irc: Remote host closed the connection17:16
mtigas (~Adium@users.spokesman.com) joined #pypy.17:17
fijalit is at the wrong level17:18
fijalforce who to compute the result?17:18
fijalwell17:18
fijalit's not17:18
Alex_Gaynoranyone17:18
fijalsince those are raw structs17:18
fijalteach rffi.sizeof17:18
Alex_Gaynoris it just sum(rffi.sizeof(x) for x in STRUCT._flds.itervalues())?17:18
mtigas (~Adium@users.spokesman.com) left irc: Client Quit17:19
fijalno no no no17:19
fijalalignment17:19
Alex_Gaynorsee, this is why I didn't want to compute it myself ;)17:19
Alex_Gaynorsurely there's a funciton somewhere already which does this?17:19
Alex_Gaynorhow do you get integer offsets for descrs?17:19
fijalsomewhere in jit/backend/llsupport?17:20
fijalit uses ctypes17:21
fijaleven more fun uses ll2ctypes17:21
fijal!17:21
mtigas (~Adium@users.spokesman.com) joined #pypy.17:21
Alex_Gaynorctypes.sizeof(ll2ctypes.get_ctypes_type(TYPE))17:22
Alex_Gaynorhow wrong would it be to just copy this?17:22
Alex_Gaynorwell, import it, but yes17:22
fijalvery wrong17:25
fijalI would be a happier man if ll2ctypes is not imported during translation17:25
fijalfor a fact, I have ll2ctypes report broken offsets to me17:25
Alex_Gaynorfijal: so what do you suggest?17:27
fijalgood question17:27
Alex_Gaynorsilly yaks17:28
fijalI guess if you translate support code things are different?17:28
Alex_Gaynorwhat is that even?17:28
fijalwe don't use ll2ctypes is everything is translated17:29
fijalas opposed to just functions we run17:29
fijaland not the JIT17:29
Alex_Gaynorok, and?17:29
fzzzy (~donovan@nat/mozilla/x-wkurdvkrrfochpcu) joined #pypy.17:30
kenaan12hager ppc-jit-backend 1187920fe776f3 15/pypy/jit/backend/: Started implementation of CALL.17:31
k_bx (~k_bx@94.244.19.62) joined #pypy.17:33
fijala bit no idea17:34
fijaland then we don't use ll2ctypes for translation17:34
fijalwell17:36
fijalleave a comment and use ctypes for now17:36
Alex_Gaynorok17:36
kenaan12fijal numpy-multidim 11252e03277a09 15/pypy/module/micronumpy/: basic slice support17:36
Alex_Gaynorfijal: does it seem correct to add __add__ to AddressAsInt?17:36
fijalno17:37
fijalwhy?17:37
fijalAddressAsInt can't be added17:37
teknico (~quassel@88-149-208-198.dynamic.ngi.it) left irc: Remote host closed the connection17:38
fijal@fijall pypy-c-jit-48400 crashed on windows. Haven't even started the test.17:38
fijalI love twitter17:38
fijalhow is this a bug report?17:38
Alex_Gaynorfijal: so what od you suggest for http://paste.pocoo.org/show/499664/ ?17:38
fijalpfff17:39
fijalwell17:39
fijalAddressAsInt would be ok to have __add__17:39
fijalbut it should return something that would explode if you try to translate it17:39
Alex_Gaynoranother AddressAsInt I would hitnk17:39
fijalok17:40
fijalbut then you need to do correct addition on self.adr17:40
fijalwith a symbolic!17:40
fijalsee why this is a mess?17:40
Alex_Gaynorpff, yes17:41
Alex_Gaynorand all I wanted was array[idx].field :/17:41
fijalwhy can't you pass a symbolic to space operation?17:41
fijalmaybe you should start with doing that?17:41
fijaland then addressasint can have __add__17:41
Alex_Gaynorit blows when up when trying to flatten the jitcode17:41
fijalwell, look how it flattens AddressAsInt maybe17:42
fijalyou know what's the problem right?17:42
Alex_Gaynorintegers and pointers are evil17:42
fijalor why we have address as int in the first place17:42
Alex_Gaynorno idea17:44
fijalyou can't render ints for pointers in C code17:44
fijalbecause they're invalid17:44
fijalthey have to be &function_name17:44
fijalbut if you add stuff to them, you still have to keep track17:44
fijalit'll be &functionname + sizeof(stuff)17:45
Alex_Gaynorfun17:47
ramusara (~ramusara@220.156.210.236.user.e-catv.ne.jp) left irc: Quit: Leaving...17:47
fijalyes17:48
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection17:48
Alex_Gaynorfijal: ok I have the issue now, if you have a symbolic there it blows up in assembler:emit_const17:58
etrepum (~bob@75-101-96-144.dsl.static.sonic.net) left irc: Quit: etrepum17:58
rekamso (~textual@216.190.29.42) joined #pypy.17:58
fijaldunno, try to sneak it in?17:59
fijalit'll work for llgraph backend, unsure about x86 ;-)17:59
Alex_Gaynorwell at this poitn you have to convert it to an int I think18:00
Alex_Gaynorthere's no other way to emit a const18:00
fijalAddressAsInt is emitted how?18:01
Alex_Gaynorit's not emitted, it just happens to be around form somewhere when run on llgraph18:01
Alex_Gaynoryou don't have an AddressAsInt const18:01
fzzzy (~donovan@nat/mozilla/x-wkurdvkrrfochpcu) left irc: Read error: Connection reset by peer18:02
fzzzy (~donovan@nat/mozilla/x-geozbagljmsmlwhf) joined #pypy.18:02
fijalAlex_Gaynor: what happens if you have one?18:07
Alex_Gaynorno idea, heaptracker.adr2int perhaps?18:07
senyai (3ead87cd@gateway/web/freenode/ip.62.173.135.205) left irc: Ping timeout: 265 seconds18:09
tos9 (~tos9@unaffiliated/tos9) left irc: Quit: leaving18:09
Alex_Gaynorfijal: err no, not that, TBH I'm not sure I follow wht happens, it seems to emit an index into a list of constants, I'm not sure what that constant is thoguh18:10
fijalsee?18:10
Alex_Gaynoroh, it just lets the constant sit around, and teh translator converts it I suppose18:10
derdon (~derdon@pD9E1D83B.dip.t-dialin.net) left irc: Remote host closed the connection18:10
fijalsee?18:11
Alex_Gaynorfijal: see what, what's your conclusion from that?18:11
fschwidom (~fschwidom@92.117.90.6) left irc: Remote host closed the connection18:11
fijalit's possible to squeeze ItemOffset there18:11
Alex_Gaynorsqueeze?18:11
fijalas in pass further18:11
Alex_GaynorI'll try it18:12
Alex_Gaynorfijal: ok now what do I do for AddressAsInt + ItemOffset?18:13
fijalmake __add__18:13
Alex_Gaynorand what should it do?18:14
fijalthat would return another AddressAsInt with self.adr + <rhs>18:14
Alex_Gaynorhow do you conver the RHS to an integer?18:14
fijalyou don't18:14
Alex_Gaynorwe're back to the same problem we started wth18:14
fijalself.adr is an address18:14
Alex_Gaynorfijal: blows up anyways http://paste.pocoo.org/show/499678/18:16
fijalgood18:17
fijalare you adding correct stuff?18:17
fijalseems not :)18:17
Alex_Gaynorpff, who knows18:17
fijal* should be first no?18:17
Alex_Gaynorhttp://paste.pocoo.org/show/499681/18:18
tos9 (~tos9@unaffiliated/tos9) joined #pypy.18:20
Alex_Gaynorfijal: ^ any idea?18:22
fijalindeed sounds good18:22
Alex_Gaynorso why it blows up?18:22
Alex_GaynorI've rwitten like 4 lines of code and twenty of tests :/18:22
fijalcommit it to a branch and I'll have a look?18:23
kenaan12alex_gaynor jit-raw-array-of-struct 11b968a2c1909c 15/pypy/: progress18:23
Alex_Gaynordone18:23
fijalsec18:26
fijalfixing tests18:26
Alex_Gaynormy tests?18:27
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Quit: reboot18:29
kenaan12fijal numpy-multidim 11a2490f74710a 15/pypy/module/micronumpy/: better support for slices18:29
fijalno18:29
Alex_GaynorAHHH, I think there's a bug in my test, the array should be nolength18:29
Action: Alex_Gaynor goes to find another asser to add18:29
fijal:]18:29
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) joined #pypy.18:30
kenaan12alex_gaynor jit-raw-array-of-struct 11122a250914a5 15/pypy/jit/metainterp/test/test_ajit.py: test fix18:30
Alex_Gaynorfijal: don't worry, there's still another bug18:30
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.18:31
Alex_Gaynorfijal: http://paste.pocoo.org/show/499691 is the new fail btw18:34
kkris (~kris@80-123-42-216.adsl.highway.telekom.at) joined #pypy.18:36
fijalyou need to teach it how to cast this correctly to int18:37
Alex_Gaynorit's the opposite I think18:38
fijalor the opposite18:39
fijaltypes are incorrect18:39
fijalit seems18:39
Alex_Gaynorwell, they're sort of correct18:39
Alex_Gaynorthe issue is the add returns a pointer to an array of structs18:40
fijalemit a cast?18:40
fijalin jtransform18:40
Alex_Gaynorwell, you don't need to in x86, only for llgraph18:41
fijalyes18:41
fijalbut cases in x86 are moot in general18:41
fijalcasts18:41
Alex_Gaynorfijal: well, is it correct that the type of adding an offset to an array is another array?18:43
Alex_GaynorI claim it should be an element18:43
fijalit's a subarray18:44
etrepum (~bob@accessnat4.mochimedia.net) joined #pypy.18:44
fijalwhat is box.wrap returning?18:45
fijala W_SomethingObject18:45
fijalright?18:45
fijaletrepum: hi18:45
etrepumfijal: hey18:46
Alex_Gaynorit doesn't seem to be18:46
Alex_Gaynorfijal: yes18:46
fijalhttp://paste.pocoo.org/show/499699/18:46
fijalUnsigned seems to claim to return int?18:46
Alex_Gaynoryou broke it!18:47
fijalthat's a weird annotator error...18:47
Alex_Gaynorwhat does _single_item_at_index do?18:47
fijalwtf18:48
fijalI'm a moron18:48
fijalAlex_Gaynor: returns index under which to write/read a single item to an array18:48
Alex_Gaynorfijal: that's a terrible name ;)18:49
fijalso say a[4,3,2] = ....18:49
fijalwould call it18:49
fijal_index_of_single_item?18:49
Alex_Gaynoryes18:49
kenaan12fijal numpy-multidim 11d5d0e247d026 15/pypy/module/micronumpy/: write another test, I'm a moron18:52
kenaan12fijal numpy-multidim 1114002f5da120 15/pypy/module/micronumpy/interp_numarray.py: rename a horrible name, this one is for alex18:52
Arach (~arach@2.95.27.244) left irc: Ping timeout: 240 seconds18:52
Alex_Gaynorfijal: btw how fast are these tests compared to the old ones18:53
fijaldepends :/18:53
Alex_Gaynorok18:54
fijalif you run all of them faster, but a single one is slower18:54
Alex_Gaynorsounds ok to me18:54
fijalbecause you compile only once, but you have to pull the entire parser18:54
fijalon the flipside, no more NonConstant and annotator hacks18:54
fijal(at least)18:54
fijalit's ~20s to build18:54
fijaland then each one takes a second or two or three18:54
Alex_Gaynorah, so quite a bit slower for one18:55
Alex_Gaynorbut ok, that's life18:55
fijalyes18:56
fijalone is quite a bit slower18:56
fijalif you were to run one, you can prebuild the code18:56
fijaland then it's quite a bit faster18:56
fijalbut I didn't feel like adding a layer of trickery there18:56
Alex_Gaynorthat's fine18:56
fijalyou could probably do the same thing as test_newgc or so18:57
fijalso you define functions, gather them and then run tests18:57
fijalbut well18:57
fijalpffff18:58
fijalAlex_Gaynor: why does list.reverse() for the list?18:58
Alex_Gaynorwhat?18:58
fijalI have a virtual18:58
fijaland I call l.reverse()18:58
fijalit's forced18:58
fijalforce18:58
Alex_Gaynorso throw @jit.unroll_iff(lambda lst: jit.isvirtual(lst)) on ll_list_reverse18:59
fijalnice :)18:59
Alex_Gaynorwe have good infrastructure18:59
fijaldoes oopspec interfere?19:00
fijalit has a loop19:00
fijalnot like we need the oopspec19:00
Alex_Gaynoryaeh remove the oopspec19:00
Alex_Gaynorthere are tons of them in rpython/ from who knows how long ago19:00
Alex_Gaynorprobably since JIT #2 or somthing19:00
fijalwhere is unroll_iff?19:01
arigato (~arigo@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Ping timeout: 244 seconds19:01
fijalJIT #119:01
fijalthat supported resizable arrays19:01
fijalI don't see it defined in rlib/jit.py19:01
Alex_Gaynorfijal: sorry, looik_inside_iff19:02
Alex_Gaynorcarl made me name it that :)19:02
fijaldo you have any idea why line numbers are completely of?19:03
Action: fijal is impressed19:04
fijalAlex_Gaynor: have you seen how calc_index looks like?19:04
Alex_Gaynorno, should I have a gander on your branch?19:04
fijalI just wanted to tell you how awesome the JIT is19:04
fijalso as of now it gets compiled to one int_mul and one int_add if self.shape = [number]19:05
Alex_Gaynorthe just turns "%(i)d %(i)d" % {"i": i} into 0 dict operations19:05
Alex_Gaynorthat's kind of good19:05
Alex_Gaynorwow, tat's pretty good19:05
fijalwith all the lists and whatnot19:06
fijalso there is no point of a fast path there for 1 DIM19:06
Arach (~arach@2.95.26.12) joined #pypy.19:06
Alex_Gaynorfijal: the only thing that would be cooler is if we could move this to be pure python, and only the dtype API was in RPython or something19:06
Alex_Gaynorwe need a few more JIT hints for that19:06
fijalyes, probably more than a few19:07
Alex_Gaynorprobably, something for the future :)19:07
lac:-)19:07
Alex_Gaynorfijal: well, push the unrolling commit already :)19:07
fijalAlex_Gaynor: anyway, this is 2 int_adds too many19:08
Alex_Gaynorfijal: you said it was one int_add and one int_mul19:08
Alex_Gaynorhow can that be 2 int_adds too many19:09
fijalthere are 2 calls to calc_index19:09
kenaan12fijal numpy-multidim 1146409ef7c41a 15/pypy/: another test. make it work19:09
Alex_Gaynorand something doesn't get folded?19:09
Alex_Gaynorcan you paste the jit-klog?19:10
fijalshape is not considered a loop constant19:16
fijaleven if I make it an immutable field19:16
fijalhm19:16
fijalthere is a missing optimization19:16
fijala read out of an array is loop invariant if the array is immutable19:17
fijalmaybe?19:17
Alex_Gaynorfijal: did you do _immutable_fields_ = "foo" or "foo[]"19:17
Alex_Gaynorfopo[*]19:17
Alex_Gaynorrather19:17
fijalstorage[*]19:17
fijalbut the array is not a constant19:17
fijalbut maybe we can have an immutable field shape?19:18
Alex_Gaynorshape should be part of the signature perhaps?19:18
fijalno19:18
fijalonly shape of shape :)19:18
Alex_Gaynorat leas the dimensions?19:18
fijalyes19:18
fijallater though19:18
fijalfighting with 1D so far19:18
Alex_Gaynorindeed19:18
fijalwe have to make it a signature, otherwise we end up with bridges19:18
fijaland they suck19:18
fijalhm19:19
Alex_Gaynorsuck-ish19:19
fijalI don't want to promote shape, but I want to make it loop invariant19:19
Alex_Gaynorthat should happen automatically19:19
Alex_Gaynorwhy doesn't it?19:20
Alex_Gaynor(afk for 20 minutes or so, back to discuss more then)19:20
fijalok19:20
fijalAlex_Gaynor: a random observation19:29
fijalif there is just one guard, it'll never come back to the same loop19:29
fijalso you can assume there are no bridges ever19:29
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.19:31
Rhy0lite (dje@nat/ibm/x-uwclkuvdcffdjxal) left irc: Quit: Leaving19:33
dmsk (~dmsk@62.61.145.89.generic-hostname.arrownet.dk) joined #pypy.19:36
dmsk (dmsk@62.61.145.89.generic-hostname.arrownet.dk) left #pypy.19:37
fijalAlex_Gaynor: I think it's a test artifact19:40
zain2 (~zain@69.38.221.130) joined #pypy.19:41
MjrTom (MjrTom@azureus/MjrTom) left irc: Quit: Planet Earth is blue and there's nothing I can do19:46
Alex_Gaynorfijal: what?19:47
fijalthe adds19:48
fijalthe original test was using 3 as a constant19:48
fijalso it could fold stuff together19:48
fijalbut now it's a very generic "step"19:48
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 260 seconds19:49
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.19:49
Alex_Gaynoroh, don't change the test then ;)19:49
fijalso I guess it's fine19:50
fijalin general we want loops to work with any step not just 319:51
fijaland this is how it got compiled :)19:51
Alex_Gaynorok, and if it's 3 it gets constant folded?19:51
mabbikeel (~mabbikeel@cpc4-duns7-2-0-cust218.9-3.cable.virginmedia.com) left irc: Ping timeout: 252 seconds19:51
fijalno19:51
fijalbecause step got annotated as constant 319:51
fijalwe can't have it in new tests19:52
Alex_Gaynorah, so you're saying it was an accident before19:52
Alex_Gaynorthat only worked in tests19:52
fijalno19:52
fijalit only worked in this particular test19:52
Alex_Gaynorright, because of the annotation19:54
MjrTom (MjrTom@azureus/MjrTom) joined #pypy.19:54
fijalyes19:54
fijalbecause the compiled assembler only worked on slices with step 319:54
Alex_Gaynorright19:54
fijalso I guess we're fine ;-)19:56
fijalgood19:57
fijalso the new code does not perform any worse at least19:57
Alex_Gaynoryes19:57
Alex_Gaynorfijal: btw, fakeaddress + itemoffset does not return a subarray if the fakeaddress is a C-like array19:57
Alex_Gaynorit just returns an array19:57
Alex_GaynorI wonder, why doesn't lltype support casting an array to point to an element of that array19:58
fprimexis there a nightly src tarball I could use for an RPM or should I get it off bitbucket?19:58
Alex_Gaynorfprimex: no, I don't think we have a nightly src tarball19:59
kenaan12fijal numpy-multidim 11d790f132a006 15/pypy/module/micronumpy/interp_numarray.py: make shape and chunks non-resizable lists19:59
fijalyes there are19:59
kenaan12fijal numpy-multidim 11cd07f23f15fc 15/pypy/module/micronumpy/interp_numarray.py: make shape and chunk immutable. no real benefits unless some array is constant.19:59
fijalbitbucket has .tar.bz2 links19:59
Alex_Gaynorfijal: yes, but not on buildbot.pypy.org19:59
fijalfprimex: if you get a checkout however, it'll have correct hg version numbers19:59
fijalok20:00
fijalnow multi dim test_zjit20:00
kenaan12fijal numpy-multidim 114fde3633c155 15/pypy/module/micronumpy/test/test_zjit.py: fix test, skip the old ones20:00
jonvk (~jonathan@bas1-montreal29-1279650546.dsl.bell.ca) joined #pypy.20:00
fprimexfijal: is there any neat way to trim the nightly off the checkout? I'd like to avoid having a 226.9 SRPM20:00
fprimex226.9MB SRPM, that is20:01
fijalfprimex: if all you want is just source, take this https://bitbucket.org/pypy/pypy/get/1847537fd4b5.tar.bz220:01
fijalif you however want to just package it after building20:01
fijalthen there is a script called package.py20:01
fijalpypy/tool/release/package.py I think20:01
fprimexthanks20:02
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 260 seconds20:02
fijalhm20:02
fijalAlex_Gaynor: our architecture is drastically different from numpy's20:02
Alex_Gaynorfijal: is it better?20:03
fijalI hope so20:03
fijalbut chiefly it's different20:03
fijalwell, we couldn't do all the jit tricks without it20:03
fijalbut it's also impossible without RPython20:03
fijalso I guess it is better20:03
Alex_Gaynorhow are we different, besides the fact that we combine operations?20:03
fijalwe have boxes20:04
stakkars_ (~tismer@212.255.22.92) joined #pypy.20:04
Alex_Gaynorthey have what, unions?20:04
fijalwell yes, we combine operations20:04
Alex_Gaynoroh of course, you don't need boxes if you don't combine operations20:04
fijalyes20:05
fijaland that changes a lot20:05
Alex_Gaynorwell we can reuse these boxes eventually though20:05
Alex_Gaynorto be the applevel classes as well20:06
fijalthat would simplify a bit20:07
Alex_Gaynoryes, it'll be good I tihink20:07
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.20:08
Alex_GaynorI need to figure out how you do multiple inheritance of a typedef20:08
fijalhahaha :)20:08
fijalwhile you're at it, remove multimethods ;-)20:08
Alex_Gaynoryeah , i"ll get right on that20:09
asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) left irc: Quit: asmeurer20:12
Alex_Gaynorfijal: why can't I cast a pointer to an array of structs to a pointer to a struct?20:17
fprimexok, got a nightly building into an RPM with 6GB of free space on the RAM disk20:21
zain (~textual@c-67-160-201-63.hsd1.ca.comcast.net) left irc: Disconnected by services20:21
Nick change: zain2 -> zain20:21
fprimexI'm having it not clean up afterward so I can see how much it used20:21
lacAlex_Gaynor: maybe because the fields are not guaranteed to be aligned properly?20:21
zain__ (~textual@c-67-160-201-63.hsd1.ca.comcast.net) joined #pypy.20:22
Da_Blitz (~Da_Blitz@203.56.250.63) left irc: Ping timeout: 240 seconds20:24
Shanita (~John@osbk-4d087ff6.pool.mediaWays.net) left irc: Read error: Connection reset by peer20:26
Da_Blitz (~Da_Blitz@203.56.250.63) joined #pypy.20:27
Moku (~John@osbk-4d087ff6.pool.mediaWays.net) joined #pypy.20:29
Nick change: Moku -> Guest3216720:29
Nick change: Guest32167 -> Shinta20:29
Nick change: Shinta -> Shinita20:29
zooko (~user@c-67-190-86-73.hsd1.co.comcast.net) joined #pypy.20:30
zookofijal: I want a blog entry or article from you or someone which is all about "so you want to write 100% pure Python code, but make it possible for PyPy to JIT it into very efficient native code"...20:32
zookoKnowing you, you would wisely emphasize the big nonos and the big wins and not waste everyone's times with nano-optimizations.20:32
zooko 20:32
zookoMaybe the first pass that this hypothetical writer is going to write could concentrate on things that are *both* unidiomatic or confusing or error-prone or whatever, *and* make it hard for the JIT to do its thing. :-)20:32
zookoMaybe these hypothetical hackers who are going to cooperate with this hypothetical writer to write this thing which I want to exist20:32
zookomaybe they will treat this document as a list of warts that ought to be fixed.20:32
zookoIf it is important enough that normal users need to learn it (as is the case with accumulator += nextstring), then it probably deserves to be high on the list of things that we would fix if we could.20:32
RiaanFV (~riaanfv@41-132-169-212.dsl.mweb.co.za) joined #pypy.20:36
tilgovi (~randall@199.36.158.16) joined #pypy.20:37
tilgovi (~randall@199.36.158.16) left irc: Changing host20:37
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.20:37
fprimex (~fprimex@brent-macbook.sc.fsu.edu) left irc: Quit: blast off!20:41
RiaanFV (riaanfv@41-132-169-212.dsl.mweb.co.za) left #pypy ("Leaving").20:59
lizardo (~lizardo@189.2.128.130) left irc: Quit: Leaving20:59
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) left irc: Quit: unbit21:13
unbit (~unbit@93-58-162-185.ip159.fastwebnet.it) joined #pypy.21:14
stakkars_ (~tismer@212.255.22.92) left irc: Read error: Connection reset by peer21:30
mcdonc (~mcdonc@cabana.palladion.com) left irc: Ping timeout: 258 seconds21:31
Alex_Gaynorlac: I think they are, this is an Array with nolength=True, so it's the same as a C array21:32
felipejcruz (~felipecru@187.102.160.14) joined #pypy.21:42
mtigas (~Adium@users.spokesman.com) left irc: Quit: Leaving.21:45
tlynn (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) joined #pypy.21:47
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/69521:49
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/11921:49
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/92221:49
k_bxjitviewer is broken a little21:49
k_bxjitviewer.py:195 should contain __main__, not __name__21:50
k_bx(I assume)21:50
kkris (~kris@80-123-42-216.adsl.highway.telekom.at) left irc: Remote host closed the connection21:53
felipejcruz (~felipecru@187.102.160.14) left irc: Quit: felipejcruz21:53
mrsipan_ (~ben@lb1.att.zope.net) joined #pypy.21:59
k_bx (~k_bx@94.244.19.62) left irc: Ping timeout: 260 seconds22:15
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 245 seconds22:23
zooko (~user@c-67-190-86-73.hsd1.co.comcast.net) left irc: Ping timeout: 276 seconds22:25
mcdonc (~mcdonc@cabana.palladion.com) joined #pypy.22:31
tlynn_ (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) joined #pypy.22:32
tlynn (~tlynn@cpc6-cmbg14-2-0-cust121.5-4.cable.virginmedia.com) left irc: Ping timeout: 240 seconds22:35
stakkars_ (~tismer@i59F6BB85.versanet.de) joined #pypy.22:44
zooko (~user@97-118-92-53.hlrn.qwest.net) joined #pypy.22:52
zooko (~user@97-118-92-53.hlrn.qwest.net) left irc: Ping timeout: 260 seconds22:58
whitelynx (~whitelynx@li117-47.members.linode.com) left irc: Quit: Ex-Chat23:02
aat (~aat@cpe-72-225-174-173.nyc.res.rr.com) joined #pypy.23:06
mrsipan_ (~ben@lb1.att.zope.net) left irc: Ping timeout: 240 seconds23:12
kenaan12gutworth default 11b9587add901b 15/pypy/objspace/std/: add float.is_integer()23:20
kenaan12gutworth default 11b9fb5a2b414e 15/pypy/: merge heads23:20
Alex_Gaynorgutworth: write test for the infs and nan please ;)23:21
bgolathat was fast :p23:22
Alex_Gaynorand it has more tests than CPythons23:22
bgolatrue, haha23:22
kenaan12gutworth default 117cfece2a13d1 15/pypy/objspace/std/: handle special values23:27
Alex_Gaynoryay, thanks gutworth 23:28
gutworthwho knows what is_integer is good for23:29
Alex_Gaynornot me23:30
Alex_Gaynornot writing tests apparently23:30
Alex_Gaynorgutworth: I guess we should be good people and go add the tests to CPython?23:30
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/69523:34
fzzzy (~donovan@nat/mozilla/x-geozbagljmsmlwhf) left irc: Quit: fzzzy23:35
gutworthindeed23:38
asmeurer (~asmeurer@dhcp-baca-230.resnet.nmt.edu) joined #pypy.23:41
stakkars_ (~tismer@i59F6BB85.versanet.de) left irc: Quit: schnarch23:48
fzzzy (~donovan@76-198-130-19.lightspeed.mtvwca.sbcglobal.net) joined #pypy.23:54
JaRoel (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection23:59
--- Sat Oct 29 201100:00

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