#pypy IRC log for Thursday, 2009-06-25

gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) left irc: Remote closed the connection00:03
lucian_ (n=lucian@ joined #pypy.00:05
kenaan03benjamin 10r6595014 pypy/branch/pyjitpl5/pypy 07M(pycode.py objects.py pycode.py serializer.py test_jit.py): make SPLI code objects SPLIObjects ...00:06
gutworthgrr. s_ImpossibleValue00:08
lucian (n=lucian@ left irc: Read error: 60 (Operation timed out)00:10
fijalgutworth: you killed my clever hack :)00:10
fijalgutworth: ah00:10
fijalgutworth: now you need to add storing/retrieving names from serialize/deserialize00:11
fijalit was not needed00:11
fijalbut now it's needed00:11
gutworththat's exactly what I'm doing :)00:11
lucian (n=lucian@ joined #pypy.00:17
fijalantocuni: are you using translations on bigdog-vm2?00:18
antocuniI have one that have been forked-before00:18
antocunido you want me to kill it?00:18
fijalin case you don't necesarilly need them, we're low on mem there00:18
fijalI'm debugging fully blown translation00:19
fijaland it eats quite a bit00:19
fijalof mem00:19
fijalI wonder if we can get bigger swap00:20
fijalbigdog: hey?00:20
fijalfound a bug :-/00:24
[mathias] (n=mathias@0x573e86a2.esnqu2.dynamic.dsl.tele.dk) left irc: 00:24
Action: gutworth curses binary data00:25
antocuni (n=antocuni@host146-120-dynamic.11-79-r.retail.telecomitalia.it) left irc: "Leaving"00:26
fijalgutworth: we can have a text format instead00:27
gutworthhow about xml? :)00:27
fijals say no to xml!00:28
lucian_ (n=lucian@ left irc: Read error: 60 (Operation timed out)00:28
kenaan03fijal 10r6595114 07M(pypy/branch/pyjitpl5/pypy/jit/metainterp/simple_optimize.py): Copy operations, even in simple_optimize ...00:30
kenaan03fijal 10r6595214 07M(pypy/branch/pyjitpl5/pypy/jit/metainterp/simple_optimize.py): improve comment ...00:30
lsoto (n=lsoto@ joined #pypy.00:41
victorpoluceno_ (n=Victot@ joined #pypy.00:45
voidspace (n=fuzzyman@87-194-212-65.bethere.co.uk) left irc: 00:52
victorpoluceno_ (n=Victot@ left irc: "Leaving"00:55
lucian_ (n=lucian@ joined #pypy.00:56
lucian__ (n=lucian@ joined #pypy.01:01
victorpoluceno (n=Victot@ left irc: Read error: 110 (Connection timed out)01:03
Trundle (n=andy@p5B134DB1.dip.t-dialin.net) left irc: "..."01:04
lucian___ (n=lucian@ joined #pypy.01:06
lucian (n=lucian@ left irc: Read error: 110 (Connection timed out)01:21
lucian (n=lucian@ joined #pypy.01:25
lucian_ (n=lucian@ left irc: Read error: 110 (Connection timed out)01:25
kenaan03benjamin 10r6595314 pypy/branch/pyjitpl5/pypy/jit/tl/spli 07M(pycode.py serializer.py test_serializer.py test_translated.py): serialize co_names ...01:27
fijal (n=fijal@rtr-folsom.wireless.indra.com) left irc: "Leaving"01:27
kenaan03benjamin 10r6595414 07M(pypy/branch/pyjitpl5/pypy/jit/tl/spli/interpreter.py): co_names_w -> co_names ...01:28
gutworthhah! all tests pass now :)01:28
kenaan03benjamin 10r6595514 07M(pypy/branch/pyjitpl5/pypy/jit/tl/spli/interpreter.py): implement STORE_GLOBAL to get the annotator to generalize the global dict ...01:28
gutworthit would be nice sometimes to have hint(obj, initial_generalization=SomeObject())01:29
gutworthseems like less of a hack :P01:30
lucian__ (n=lucian@ left irc: Read error: 110 (Connection timed out)01:31
kenaan03benjamin 10r6595614 07M(pypy/branch/pyjitpl5/pypy/jit/tl/spli/targetspli.py): fix translation of target ...01:36
AirCastle (i=AirCastl@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: 01:38
fijal (n=fijal@c-75-71-66-41.hsd1.co.comcast.net) joined #pypy.01:41
fijalgutworth: ping01:42
Action: gutworth eating now01:42
fijalwould be cool if we replicate the framestack thingie01:42
fijal(would be like-python)01:42
fijalso if you have an exception, you print a traceback out of framestack01:43
lucian___ (n=lucian@ left irc: Read error: 110 (Connection timed out)01:48
lucian_ (n=lucian@ joined #pypy.01:49
lucian__ (n=lucian@ joined #pypy.01:52
lsoto (n=lsoto@ left irc: Read error: 60 (Operation timed out)01:53
ehmo (n=nope@chello089173023218.chello.sk) left irc: "http://blog.synopsi.com"01:57
ehmo (n=nope@chello089173023218.chello.sk) joined #pypy.01:58
gutworthfijal: good idea. I suppose we should just make a miny execution context02:06
fijalthat would also put us closer to what python does02:07
kenaan03benjamin 10r6595714 07M(pypy/trunk/pypy/translator/goal/translate.py): only start pdb on successful translation if it is requested ...02:07
lucian_ (n=lucian@ left irc: Read error: 110 (Connection timed out)02:08
lucian (n=lucian@ left irc: Read error: 110 (Connection timed out)02:15
lucian (n=lucian@ joined #pypy.02:29
lucian_ (n=lucian@ joined #pypy.02:42
gutworthfijal: the virtualizable document should be updated02:47
lucian__ (n=lucian@ left irc: Read error: 110 (Connection timed out)02:51
lucian (n=lucian@ left irc: Read error: 60 (Operation timed out)02:51
AirCastle (i=AirCastl@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.02:54
yangkun (n=wayhome@ joined #pypy.02:56
dialtone_ (n=dialtone@adsl-67-125-22-195.dsl.pltn13.pacbell.net) joined #pypy.03:12
dialtone (n=dialtone@unaffiliated/dialtone) left irc: Read error: 110 (Connection timed out)03:12
lucian (n=lucian@ joined #pypy.03:16
lucian__ (n=lucian@ joined #pypy.03:22
lucian___ (n=lucian@ joined #pypy.03:24
lucia____ (n=lucian@ joined #pypy.03:26
luci_____ (n=lucian@ joined #pypy.03:28
lucian_ (n=lucian@ left irc: Read error: 110 (Connection timed out)03:40
lucian (n=lucian@ left irc: Read error: 110 (Connection timed out)03:44
lucian__ (n=lucian@ left irc: Read error: 110 (Connection timed out)03:45
lucia____ (n=lucian@ left irc: Read error: 110 (Connection timed out)03:45
lucian___ (n=lucian@ left irc: Read error: 110 (Connection timed out)03:46
lsoto (n=lsoto@pc-148-86-104-200.cm.vtr.net) joined #pypy.04:03
luci_____ (n=lucian@ left irc: "Leaving..."04:07
gutworth (n=bpeterso@64-131-8-201.usfamily.net) left irc: Read error: 60 (Operation timed out)04:49
derfel (n=derfel@host104-3-dynamic.180-80-r.retail.telecomitalia.it) left irc: Read error: 54 (Connection reset by peer)05:09
mwhudson (n=mwh@118-92-161-120.dsl.dyn.ihug.co.nz) left irc: Read error: 104 (Connection reset by peer)06:18
yangkun (n=wayhome@ left irc: Read error: 110 (Connection timed out)06:19
mwhudson (n=mwh@canonical/launchpad/mwhudson) joined #pypy.06:23
yangkun (n=wayhome@ joined #pypy.07:32
lsoto (n=lsoto@pc-148-86-104-200.cm.vtr.net) left irc: Read error: 110 (Connection timed out)08:01
RealNitro (n=jens@nat/sun/x-01589e91bccb08eb) joined #pypy.08:36
FMJagCave (n=na@ip68-96-169-94.lv.lv.cox.net) left irc: Client Quit09:02
fuzzyman (n=fuzzyman@87-194-212-65.bethere.co.uk) joined #pypy.09:36
cfbolz (n=cfbolz@fwstups.cs.uni-duesseldorf.de) joined #pypy.09:55
jewel (n=jewel@dsl-242-129-65.telkomadsl.co.za) joined #pypy.10:08
kenaan03cfbolz 10r6595814 07M(pypy/extradoc/sprintinfo/ep2009/people.txt): will definitely arrive later (if at all). ...10:12
cfbolz (n=cfbolz@fwstups.cs.uni-duesseldorf.de) left irc: "Leaving"10:26
arigato (n=arigo@c-c348e353.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.10:37
cami (n=cami@217-138.2-85.cust.bluewin.ch) joined #pypy.10:38
kenaan03arigo 10r6595914 07M(pypy/extradoc/talk/ep2009/status/status.txt): Be more vague about it, as recommended by Laura. ...10:40
verte (n=verte@dsl-220-235-228-125.vic.westnet.com.au) joined #pypy.10:46
verte_ (n=verte@dsl-220-235-228-125.vic.westnet.com.au) joined #pypy.11:11
verte (n=verte@dsl-220-235-228-125.vic.westnet.com.au) left irc: Nick collision from services.11:11
Nick change: verte_ -> verte11:11
cami (n=cami@217-138.2-85.cust.bluewin.ch) left #pypy.11:24
jewel (n=jewel@dsl-242-129-65.telkomadsl.co.za) left irc: Read error: 113 (No route to host)11:46
yangkun (n=wayhome@ left irc: "Leaving."11:50
lizardo (n=lizardo@ joined #pypy.11:52
kenaan03arigo 10r6596014 07M(pypy/extradoc/talk/ep2009/status/pypy-status.pdf): Update the pdf. ...11:59
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.12:00
pjenvey_ (n=pjenvey@astound-69-42-4-166.ca.astound.net) left irc: 12:13
Trundle (n=andy@p5B134B28.dip.t-dialin.net) joined #pypy.12:26
kenaan03arigo 10r6596114 07M(pypy/extradoc/talk/ep2009/abstract.txt): Update with a link to the slides. ...12:29
kenaan03arigo 10r6596214 07M(pypy/extradoc/sprintinfo/ep2009/people.txt): Update the list of people to the best of my knowledge. ...12:37
arigato (n=arigo@c-c348e353.022-54-67626719.cust.bredbandsbolaget.se) left irc: "See you"12:39
antocuni (n=antocuni@host71-86-dynamic.7-79-r.retail.telecomitalia.it) joined #pypy.12:39
morphis (n=morphis@hfw-ext-wlan.rz.hs-bremen.de) joined #pypy.12:46
kenaan03antocuni 10r6596314 07M(pypy/extradoc/sprintinfo/ep2009/people.txt): I guess this was the real intention of r65958 :-) ...12:49
gutworth (n=bpeterso@64-131-8-201.usfamily.net) joined #pypy.13:02
ousado_ (n=johnny@p5B3C1E5C.dip0.t-ipconnect.de) joined #pypy.13:23
ousado (n=johnny@p5B3C2080.dip0.t-ipconnect.de) left irc: Read error: 110 (Connection timed out)13:40
morphis (n=morphis@hfw-ext-wlan.rz.hs-bremen.de) left #pypy.13:41
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) left irc: Remote closed the connection13:42
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.13:43
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) left irc: Remote closed the connection13:47
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.13:48
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) left irc: Remote closed the connection13:48
tav_ (n=tav@ joined #pypy.14:20
tav (n=tav@ left irc: Read error: 60 (Operation timed out)14:25
jewel_ (n=jewel@dsl-242-129-65.telkomadsl.co.za) joined #pypy.15:05
witulski (n=stupsi@kleinbonum.cs.uni-duesseldorf.de) joined #pypy.15:13
witulski (n=stupsi@kleinbonum.cs.uni-duesseldorf.de) left #pypy.15:14
arigato (n=arigo@c-c348e353.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.15:23
Action: arigato sprained *again* his foot15:23
verte (n=verte@dsl-220-235-228-125.vic.westnet.com.au) left irc: Read error: 110 (Connection timed out)15:23
arigatonot badly, but it hurts a lot to double-sprain it15:24
arigatoI just want to say "ouch! ouch! ouch!" to someone, now that pedronis left to a conference :-)15:25
zejnI'm glad I can help. :D15:25
lucian (n=lucian@ joined #pypy.15:26
Trundle (n=andy@p5B134B28.dip.t-dialin.net) left irc: "..."15:46
Nick change: tonguetone -> necktone15:49
lsoto (n=lsoto@pc-148-86-104-200.cm.vtr.net) joined #pypy.15:51
lsoto (n=lsoto@pc-148-86-104-200.cm.vtr.net) left #pypy ("Saliendo").15:52
kenaan03arigo 10r6596414 pypy/branch/pyjitpl5/pypy/jit/metainterp 07M(codewriter.py pyjitpl.py test_virtualizable.py): Handle correctly the situation where we access another ...16:02
lucian_ (n=lucian@ joined #pypy.16:04
witulski (n=stupsi@kleinbonum.cs.uni-duesseldorf.de) joined #pypy.16:09
witulski (n=stupsi@kleinbonum.cs.uni-duesseldorf.de) left #pypy.16:10
panni_ (i=hannes@ip-88-152-22-213.unitymediagroup.de) joined #pypy.16:19
lucian (n=lucian@ left irc: Read error: 110 (Connection timed out)16:21
AirCastle (i=AirCastl@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: Success16:29
jacob22_ (n=jacob@c-22c5e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.16:30
fijal (n=fijal@c-75-71-66-41.hsd1.co.comcast.net) got netsplit.16:30
jacob22 (n=jacob@c-22c5e055.1321-1-64736c11.cust.bredbandsbolaget.se) got netsplit.16:30
fijal (n=fijal@c-75-71-66-41.hsd1.co.comcast.net) returned to #pypy.16:31
exarkun_ (n=exarkun@wordeology.com) joined #pypy.16:38
exarkun (n=exarkun@intarweb.us) left irc: "leaving"16:38
Nick change: exarkun_ -> exarkun16:39
jacob22 (n=jacob@c-22c5e055.1321-1-64736c11.cust.bredbandsbolaget.se) got lost in the net-split.16:41
lucian_ (n=lucian@ left irc: Read error: 60 (Operation timed out)16:50
fuzzymanarigato: when do you leave for EuroPython?16:59
arigatofuzzyman: sunday around 5pm17:12
fuzzymanoh right - see you on Sunday then :-)17:16
fuzzymanI'll be there from Saturday evening - I'm giving a tutorial on Sunday17:16
fuzzymanis Carl better yet?17:17
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.17:19
RealNitro (n=jens@nat/sun/x-01589e91bccb08eb) left irc: "Leaving"17:23
arigatoI don't know :-(  no news from him17:24
arigatoit's a bit crazy if you ask me, to come to both EuroPython and ECOOP (Italy) which are one after the other17:24
arigatogiven that Carl never takes the plane17:24
morphis (n=morphis@p5489A02D.dip0.t-ipconnect.de) joined #pypy.17:24
arigatoso I hope that he will at least feel well enough to come to one of these events17:25
fijalarigato: hi17:30
fijalarigato: welcome to the sprained-leg-club17:30
arigatobah :-)17:30
fijalhow did you do it?17:31
arigatoboth times, just misplacing my foot17:32
fijalI also misplaced my foot, but at least while falling from considerable height :)17:32
dialtone_ (n=dialtone@unaffiliated/dialtone) left irc: Read error: 110 (Connection timed out)17:34
fijalarigato: does your fix fixes the crash?17:34
arigatoI did't try so far17:35
Action: fijal tries17:35
fijalthe crash with hybrid gc was stupid at best :-/17:36
arigatoyes, thanks for fixing it17:36
fijalarigato: if you want, spli breaks optimize.py17:36
fijal(or optimize4.py)17:36
fijalin test_jit17:36
fijaljust disable usage of optimize217:37
arigatofor now I'm happy that test_virtualizable.py is actually a real unit test :-)17:37
fijalI know, you're telling me that second time :)17:38
arigatojust a matter of saying that I'm more happy continuing work on virtualizables than fixing bugs in optimize417:39
fijalwhat needs to be done? residual calls?17:39
kenaan03arigo 10r6596514 07M(pypy/branch/pyjitpl5/pypy/jit/metainterp/test/test_virtualizable.py): The next test. ...17:39
fijalyou might want to think what to do with framestack at some point17:40
arigatoyes, but later17:40
fijal^^^^ do you know why this creates 3 loops?17:41
arigatodepends how you count, I suppose17:41
fijal3 times assembler is called to compile a loop17:42
fijalso 1 is a loop and one is a bridge17:42
fijaland I have no clue what the 3rd one is17:42
fijalentry bridge?17:44
fijalbut why there is entry bridge?17:44
kursor (n=kursor@brln-d9bacf81.pool.mediaWays.net) joined #pypy.17:57
dialtone (n=dialtone@adsl-99-136-101-166.dsl.pltn13.sbcglobal.net) joined #pypy.17:57
fijalarigato: it does18:04
fijalarigato: == it fixes the problem18:07
fijalso we can run pystone and richards18:07
fijalcan I steal your attention for a moment?18:07
arigatono, pasta is requiring my attention any minute now :-)18:07
fijalok :)18:07
fijaltell me when you're back18:08
lucian (n=lucian@ joined #pypy.18:10
antocunifijal: what are the performances of pystone and richards?18:15
fijalthey're slower than without jit18:15
antocuniwhich optimize are you using?18:15
antocuniah, ok18:15
fijal(it does not really matter)18:15
fijalthat won't help18:16
fijalthe basic example being a call18:16
fijalwe need to start from example that repeatedly calls simple helper18:16
fijaland I think without inlining optimize won't help18:17
fijalpython-level inlining18:17
antocunishoudln't it be done automatically because of the way tracing works?18:18
fijalmore or less yeah18:19
fijalthere is a couple of issues18:19
fijaland I want to wait till armin eats his whatever he's eating18:19
jewel_ (n=jewel@dsl-242-129-65.telkomadsl.co.za) left irc: Read error: 60 (Operation timed out)18:21
lucian (n=lucian@ left irc: Read error: 60 (Operation timed out)18:23
arigatobut sorry, I'm not completely up to discussing inlining18:35
fijalI can give it a try I guess18:36
fijalI was thinking whether making framestack a linked list would not help18:37
morphis (n=morphis@p5489A02D.dip0.t-ipconnect.de) left irc: "none"18:51
fijalantocuni: do you think you can provide a working virtuals without perfect specialization?18:52
setanta1 (n=setanta@ joined #pypy.18:52
antocunifijal: no, what I'm doing is to rewrite perfect specialization18:52
fijalbut perfect specialization as it is now inherits problems18:53
fijalthat needs to be tackled before doing that18:53
antocunifijal: sorry, I'm in a real hurry, need to go immediately18:53
antocuni (n=antocuni@host71-86-dynamic.7-79-r.retail.telecomitalia.it) left irc: "Leaving"18:53
hpk (n=chatzill@HSI-KBW-085-216-104-027.hsi.kabelbw.de) joined #pypy.18:55
bigdog (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) left irc: Read error: 110 (Connection timed out)19:02
AirCastle (i=AirCastl@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.19:08
pjenvey_ (n=pjenvey@astound-69-42-4-166.ca.astound.net) joined #pypy.19:18
bigdog1 (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) joined #pypy.19:20
fijalbigdog1: hey19:21
bigdog1 (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) left irc: Read error: 60 (Operation timed out)19:36
kenaan03fijal 10r6596614 pypy/branch/pyjitpl5/pypy/jit/metainterp 07M(codewriter.py pyjitpl.py): Change recursival_call to be written as recursive_call in graphs. Does not ...19:47
arigato (n=arigo@c-c348e353.022-54-67626719.cust.bredbandsbolaget.se) left irc: "See you"19:50
fijal (n=fijal@c-75-71-66-41.hsd1.co.comcast.net) left irc: "Leaving"19:52
Nick change: tav_ -> tav20:01
kenaan03benjamin 10r6596714 pypy/branch/pyjitpl5/pypy/jit/tl/spli 05A(execution.py) 07M(interpreter.py objects.py targetspli.py test_interpreter.py test_serializer.py test_translated.py): add an execution context with a framestack ...20:01
bigdog (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) joined #pypy.20:01
AirCastle (i=AirCastl@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: Read error: 110 (Connection timed out)20:07
kenaan03benjamin 10r6596814 07M(pypy/branch/pyjitpl5/pypy/jit/tl/spli/targetspli.py): fix translation ...20:10
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) left irc: Read error: 104 (Connection reset by peer)20:11
fijal (n=fijal@97-118-106-177.hlrn.qwest.net) joined #pypy.20:29
fijal (n=fijal@97-118-106-177.hlrn.qwest.net) left irc: Read error: 60 (Operation timed out)20:36
kursor (n=kursor@brln-d9bacf81.pool.mediaWays.net) left irc: 20:37
jewel_ (n=jewel@dsl-242-129-65.telkomadsl.co.za) joined #pypy.20:37
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.20:45
kenaan03fijal 10r6596914 pypy/branch/pyjitpl5/pypy/jit/tl 07M(tl.py tlc.py): Fix tests. Support virtualizables in tl correct way, disable ...20:49
fijal (n=fijal@97-118-106-177.hlrn.qwest.net) joined #pypy.20:53
lucian (n=lucian@ joined #pypy.20:57
arigato (n=arigo@c-c348e353.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.21:06
cami (n=cami@adsl-84-227-13-42.adslplus.ch) joined #pypy.21:10
cami (n=cami@adsl-84-227-13-42.adslplus.ch) left #pypy.21:11
lameiro (n=lameiro@ip-62-235-198-132.dsl.scarlet.be) joined #pypy.21:14
lucian_ (n=lucian@ joined #pypy.21:37
kenaan03arigo 10r6597014 pypy/branch/pyjitpl5/pypy/rpython 07M(rclass.py rvirtualizable2.py rclass.py rvirtualizable2.py rvirtualizable2.py test_rvirtualizable2.py): Re-enable the production of 'promote_virtualizable'. ...21:46
lucian__ (n=lucian@ joined #pypy.21:48
fijalsvn up21:56
kenaan03fijal 10r6597114 pypy/branch/pyjitpl5/pypy/jit/metainterp 07M(history.py pyjitpl.py test_recursive.py): Basic inlining seems to work without any trouble. This is wrong so far, but ...21:58
fijalarigato: are you fine with my last checkin?21:58
arigatono test21:58
fijaltest_recursive is not a test?21:58
arigatobecause it would pass in a much easier way21:59
fijalso what should be a test?21:59
fijalthere are no asserts about calls21:59
fijalis this what you're talking about?21:59
lizardo (n=lizardo@ left irc: "Leaving"21:59
fijalsorry sorry21:59
fijalI was looking at graphs to write it down :)21:59
fijalbut forgotten to21:59
fijalbut otherwise?22:00
fijalit does not yet have a counter22:00
arigatoI suppose it's fine, as a first step, yes22:00
fijalnext step would be to add a counter (starting with a test)22:00
arigatocounting what?22:01
fijala recursion level22:01
arigatothat's not good enough22:01
fijalso we don't match the merge point with different recursion level22:01
lucian (n=lucian@ left irc: Read error: 110 (Connection timed out)22:01
arigatoah, right -- a recursion flag then22:01
fijalit's more than a flag, it needs to be a counter22:02
fijalbasically the level of recursion should match22:02
arigatono, it needs to be zero22:02
arigatoyou cannot match two random different pieces of code if they happen to have the same counter22:03
fijalbut you would want to catch the inner loop22:04
arigatoyes and no22:04
arigatothe point is that as soon as you introduce inlining,22:04
arigatoyou need to add heuristics to avoid very long traces22:04
arigatonow the case where an inner function loops is supposed to be relatively rare22:05
fijalin this case, you usually hit the inner loop earlier, right?22:05
fijalyou should basically detect if inner loop has an assembler22:05
arigatoyes, as a regular loop, not as an inner loop at all22:05
fijaland if inner loop has an assembler, not inline it at all22:06
fijaldoes it make any sense?22:06
arigatoI don't know22:08
fijalI might start then with kind of a psyco hack22:08
fijalonly inline if there is no JUMP_ABSOLUTE inside22:08
fijalthat sounds easy enough22:08
fijalyou require interpreter to provide a function for bytecode checker22:09
fijallike can_inline(code)22:09
fijalwhich will simply check it22:09
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) left irc: Remote closed the connection22:09
lucian_ (n=lucian@ left irc: Read error: 110 (Connection timed out)22:09
fijala bit, yes22:09
fijalbut easy to change with something else22:09
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) joined #pypy.22:09
arigatoI'm temted to tell you "go into a branch" :-)22:10
fijallet's say that as soon as I introduce something more invasive I will go into the branch22:10
arigatomore invasive than this obscure hack? :-/22:11
fijalit might be a bit obscure22:11
fijalbut it's at least dead-simple22:11
arigatoI bet it will require a bit of tweaking all over the place22:11
arigatoI agree that it's simple22:12
fijalchanging a bit topic, people are amazed about dotviewer22:13
fijalthe assembler level dotviewer should more or less works with windows binary, right? (modulo obtaining assembler)22:13
arigatoyou cannot be talking about pypy/trunk/dotviewer/?22:14
fijalI'm talking about jit/backend/x86/viewcode.py22:14
kenaan03fijal 10r6597214 07M(pypy/branch/pyjitpl5/pypy/jit/metainterp/test/test_recursive.py): check operations ...22:15
arigatoI suppose so, but there are heavy hacks about Windows in Psyco (where part of this code originally came from) which I've ignored22:15
fijalbut those hacks are about running psyco on windows or about interpreting this code by viewer?22:16
arigatoin viewcode.py22:16
fijalI mean, the target is to reuse the viewcode.py on assembler dump coming from somewhere else22:16
arigatoI don't remember which ones, maybe they don't apply or something22:17
arigatobut I certainly will never pretend anything is "just port it to Windows, it's trivial" :-)22:17
fijalI told the guy that he's free to try and come by here and share his experience22:17
fijalok :)22:17
arigatohe can definitely look in Psyco22:17
fijalother, slightly more clever hack, would be to raise in case we see can_enter_jit and restore the state and generate residual call22:19
fijalbut it's more involved22:19
zejn (n=hruske@89-212-104-197.dynamic.dsl.t-2.net) left irc: Read error: 60 (Operation timed out)22:23
kenaan03fijal 10r6597314 07M(pypy/branch/pyjitpl5/pypy/jit/metainterp/test/test_recursive.py): refactor test a bit for the future reuse this interpreter ...22:27
AirCastle (i=AirCastl@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.22:35
amaury_ (n=chatzill@35.85.197-77.rev.gaoland.net) joined #pypy.22:40
lucian (n=lucian@ joined #pypy.22:46
lucian_ (n=lucian@ joined #pypy.22:54
hpk (n=chatzill@HSI-KBW-085-216-104-027.hsi.kabelbw.de) left irc: Read error: 110 (Connection timed out)22:57
setanta1 (n=setanta@ left irc: "Leaving."22:57
kenaan03arigo 10r6597414 pypy/branch/pyjitpl5/pypy 07M(codewriter.py test_virtualizable.py warmspot.py rvirtualizable2.py rvirtualizable2.py rvirtualizable2.py): Still in-progress. ...22:58
gonsor (n=quassel@kel30.kel.stud.uni-goettingen.de) left irc: Remote closed the connection23:02
lucian__ (n=lucian@ left irc: Read error: 110 (Connection timed out)23:09
kenaan03arigo 10r6597514 07M(pypy/branch/pyjitpl5/pypy/jit/metainterp/executor.py): Minor detail. ...23:11
kenaan03benjamin 10r6597614 pypy/branch/pyjitpl5/pypy/jit/tl/spli 07M(serializer.py test_serializer.py): serialize nested code objects ...23:11
kenaan03fijal 10r6597714 pypy/branch/pyjitpl5/pypy 07M(pyjitpl.py test_recursive.py warmspot.py jit.py): Enough of a hack for an interpreter to provide a simple check whether ...23:12
lucian (n=lucian@ left irc: Read error: 110 (Connection timed out)23:13
mwhudson_ (n=michaelh@118-92-161-120.dsl.dyn.ihug.co.nz) joined #pypy.23:13
lucian (n=lucian@ joined #pypy.23:15
kenaan03arigo 10r6597814 pypy/branch/pyjitpl5/pypy/jit/backend/llgraph 07M(llimpl.py runner.py): Rename the 'type' field, which sometimes conflicts with ...23:16
mwhudson (n=mwh@canonical/launchpad/mwhudson) left irc: Read error: 104 (Connection reset by peer)23:17
mwhudson (n=mwh@canonical/launchpad/mwhudson) joined #pypy.23:18
Action: gutworth finds these rtyper warnings a little disturbing23:22
lucian__ (n=lucian@ joined #pypy.23:23
fijalgutworth: heh :023:25
lucian_ (n=lucian@ left irc: Read error: 110 (Connection timed out)23:28
lameiro (n=lameiro@ip-62-235-198-132.dsl.scarlet.be) left irc: Remote closed the connection23:30
gutworth[rtyper:WARNING] prebuilt instance <pypy.jit.tl.spli.objects.SPLINone object at 0x72e8b0> has no attribute 'co_consts_w'23:32
gutworthmaybe because None is in co_const_w with Code?23:33
lucian_ (n=lucian@ joined #pypy.23:33
fijalput an assert that a class that you want is a code class23:34
fijalbecause it gets promoted23:34
kenaan03fijal 10r6597914 pypy/branch/pyjitpl5/pypy/module/pypyjit 05A(test_can_inline.py) 07M(interp_jit.py): Implement inlining heuristics for pypy (without caching). Not that it is ...23:36
lucian_ (n=lucian@ left irc: "Leaving..."23:40
gutworthI will just add as_interpclass()23:41
fijalfor example23:42
fijalwell no23:42
fijalyou really need an assert23:42
gutworthwhy does it have to be an assert?23:43
gutworthknowntypedata works on an isinstance23:43
fijalisinstance is fine23:44
fijalit does not need to be as assert23:44
arigato (n=arigo@c-c348e353.022-54-67626719.cust.bredbandsbolaget.se) left irc: "See you"23:46
gutworthha. inlining must be really agreesive23:46
fijalgutworth: it does inline everything that has no loop23:46
lucian (n=lucian@ left irc: Read error: 110 (Connection timed out)23:46
fijalI fear it'll inline recursion23:46
fijalanyway, we need to sort out other problems like virtuals23:47
fijal (n=fijal@97-118-106-177.hlrn.qwest.net) left irc: "Leaving"23:50
lucian__ (n=lucian@ left irc: Read error: 110 (Connection timed out)23:55
--- Fri Jun 26 200900:00

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