#pypy IRC log for Friday, 2012-01-27

zk (~is@whatit.is) left irc: Quit: ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff00:06
zk (~is@whatit.is) joined #pypy.00:07
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.00:22
ericflo (~ericflo@75.103.8.110) left irc: Quit: ericflo00:29
ericflo (~ericflo@75.103.8.110) joined #pypy.00:30
kenaan12alex_gaynor default 118276de5ece4c 15/pypy/module/micronumpy/tool/numready/main.py: better organization, and show data for numpy.generic as well00:36
mikfc (9665bcc5@gateway/web/freenode/ip.150.101.188.197) joined #pypy.00:39
squiddy (~squiddy@f053087106.adsl.alicedsl.de) joined #pypy.00:46
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 240 seconds01:00
bbot24Failure: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/78801:03
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: Ping timeout: 247 seconds01:04
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.01:05
voidspac_ (~voidspace@87-194-212-65.bethere.co.uk) left irc: Quit: Goodbye...01:18
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 272 seconds01:21
nettok (~quassel@190.148.246.190) joined #pypy.01:25
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) left irc: Ping timeout: 240 seconds01:28
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!01:34
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/103401:37
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/4501:38
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection01:49
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.01:49
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/69102:00
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/80202:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/87302:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/121302:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/68802:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/153202:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/34102:00
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/194202:00
bbot2Started: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/35102:00
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.02:03
kenaan12alex_gaynor default 11457e2166648c 15/pypy/module/micronumpy/tool/numready/main.py: use a .html extension02:11
RhyoliteAlex_Gaynor: so the basic problem is the Django ORM syntax evolved instead of being designed from the beginning in a language-complete way?02:12
Alex_GaynorRhyolite: Basically.02:12
RhyoliteFew people remember the adage: "Plan for success"02:13
Alex_GaynorThe flip side is, there's a pretty large set of queries that are expressed very naturally :)02:14
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen02:23
RhyoliteThose two goals do not have to be mutually-exclusive02:25
nettok (~quassel@190.148.246.190) left irc: Remote host closed the connection02:25
Alex_GaynorNot at all, I just meant my critique doesn't necessarily mean we should chuck the bathwater out, instead we may be best just learning the lessons for next time.02:26
wleslienext time what? next time someone is encoding a relational algebra in python?02:27
wleslieor designing software in general?02:27
Alex_GaynorThe latter, although I have every confidence the last ORM hasn't been written yet.02:28
Rhyolitehow are you progressing with your new one?02:30
Alex_GaynorHaven't made time for it in a whiel, I've got some very nice designs in my head :)02:30
Rhyolitecombining simple and powerful is difficult, but that's the sign of a good design02:31
squiddy (~squiddy@f053087106.adsl.alicedsl.de) left irc: Quit: Leaving02:38
nettok (~quassel@190.148.246.190) joined #pypy.02:52
Moku (~John@osbk-4db06c7e.pool.mediaWays.net) joined #pypy.03:03
Shanita (~John@osbk-4db141e3.pool.mediaWays.net) left irc: Ping timeout: 240 seconds03:03
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 248 seconds03:06
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.03:17
mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) left irc: Remote host closed the connection03:19
mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) joined #pypy.03:19
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Read error: Connection reset by peer03:25
bbot23Success: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/21703:25
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.03:26
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/34103:27
DanKluev (~root@82-169-adsl.vntc.ru) left irc: Ping timeout: 248 seconds03:41
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.03:47
DanKluev (~root@109-109-213-254-xdsl.vntc.ru) joined #pypy.03:54
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 276 seconds03:58
DanKluev (~root@109-109-213-254-xdsl.vntc.ru) left irc: Ping timeout: 248 seconds04:10
anish_ (~anish@123.201.34.94) joined #pypy.04:16
Tobu (~Tobu@unaffiliated/tobu) left irc: Read error: Operation timed out04:20
DanKluev (~root@84-247-ppp.vntc.ru) joined #pypy.04:25
ericflo (~ericflo@75.103.8.110) left irc: Quit: ericflo04:27
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/69104:29
durin42 (~durin@adium/durin42) left irc: Ping timeout: 252 seconds04:34
durin42 (~durin@adium/durin42) joined #pypy.04:36
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) left irc: Ping timeout: 240 seconds04:37
anish_ (~anish@123.201.34.94) left irc: Remote host closed the connection04:43
anish_ (~anish@219.91.250.194) joined #pypy.04:49
DasIch (~DasIch@p4FFDFED3.dip.t-dialin.net) left irc: Quit: DasIch04:56
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen05:05
Telugodu (~rambaho@202.3.77.231) left irc: Quit: Ex-Chat05:08
ericflo (~ericflo@c-50-131-58-97.hsd1.ca.comcast.net) joined #pypy.05:10
anish_ (~anish@219.91.250.194) left irc: Ping timeout: 245 seconds05:11
ericflo (~ericflo@c-50-131-58-97.hsd1.ca.comcast.net) left irc: Quit: ericflo05:18
anish_ (~anish@219.91.250.194) joined #pypy.05:25
bbot23Success: 15http://buildbot.pypy.org/builders/own-linux-x86-64/builds/80205:29
mcdonc_ (~mcdonc@ip72-209-213-54.dc.dc.cox.net) joined #pypy.05:39
mcdonc_ (~mcdonc@ip72-209-213-54.dc.dc.cox.net) left irc: Quit: Gotsta go05:51
nettok (~quassel@190.148.246.190) left irc: Remote host closed the connection05:55
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.06:05
DasIch (~DasIch@p3E9901EF.dip.t-dialin.net) joined #pypy.06:07
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen06:18
AlekSi (~AlekSi@85.235.191.82) joined #pypy.06:24
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-32/builds/153206:26
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-Ojit-no-jit-linux-x86-32/builds/87306:28
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-app-level-linux-x86-64/builds/68806:28
espes (~espes@123-243-221-248.tpgi.com.au) left irc: Read error: Connection reset by peer06:34
espes (~espes@123-243-221-248.tpgi.com.au) joined #pypy.06:35
bbot23Success: 15http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/121306:39
mat^2 (~mathias@212.130.113.35) left irc: Ping timeout: 248 seconds06:52
anish_ (~anish@219.91.250.194) left irc: Quit: Leaving07:03
anish_ (~anish@219.91.250.194) joined #pypy.07:03
mikfc (9665bcc5@gateway/web/freenode/ip.150.101.188.197) left irc: Quit: Page closed07:13
bbot24Failure: 15http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/35107:15
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.07:20
vad (~vad@77.72.196.122) joined #pypy.07:21
Tobu (~Tobu@2a01:e35:8a2c:b230:213:d3ff:fe85:5614) joined #pypy.07:27
Tobu (~Tobu@2a01:e35:8a2c:b230:213:d3ff:fe85:5614) left irc: Changing host07:27
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.07:27
DasIch (~DasIch@p3E9901EF.dip.t-dialin.net) left irc: Quit: DasIch07:36
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Read error: Connection reset by peer07:36
JaRoel|4d (~jaroel|4d@sink.jaroel.nl) left irc: Remote host closed the connection07:36
mattip (~mattip@bzq-79-177-109-132.red.bezeqint.net) joined #pypy.07:37
bbot23Success: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/194207:41
bbot2Started: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1943 [py3k]07:41
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.07:46
wleslie (~verte@python/site-packages/verte) left irc: Quit: ~~~ Crash in JIT!07:51
stefanop (~stefanop@77.72.196.122) joined #pypy.07:58
AlekSi (~AlekSi@85.235.191.82) left irc: Remote host closed the connection08:02
_root_ (~Scalable@li252-14.members.linode.com) joined #pypy.08:12
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.08:14
dmitrey (~quassel@77.121.195.200) joined #pypy.08:24
antocuni_ (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) joined #pypy.08:29
Tobu (~Tobu@unaffiliated/tobu) left irc: Read error: Operation timed out08:31
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) joined #pypy.08:44
bbot24Failure: 15http://buildbot.pypy.org/builders/own-linux-x86-32/builds/1943 [py3k]08:45
dmitrey (~quassel@77.121.195.200) left irc: Ping timeout: 245 seconds08:47
kvda (~kvda@124-169-134-221.dyn.iinet.net.au) joined #pypy.08:47
fijalhi08:48
mattip'lo08:48
mattipfijal: I added the compile stuff, using the format !30! now creates a FlatIterator on a arange(30) array.08:51
mattipall tests now pass correctly.08:51
fijalhi08:52
mattipbranch numppy-flatitter08:52
fijalmaybe something like a function would be better08:52
fijalbut I'm not the one who should throw stones about compile syntax08:52
mattipyes, but it is hard.08:52
fijalwhy?08:53
mattipI couldn't find an example of this kind of python syntax:08:53
fijalcan't you just add an extra if to the list of functions already there?08:53
mattiparange(30).flat08:53
fijalno08:53
fijalbut you could do08:53
fijalflat(|30|)08:53
mattipflat is not a numpy method, only an array method.08:54
fijalso?08:55
mattipok, let me stew on it a bit.08:55
fijalas far as compile is concerned it does not metter08:55
arigato (~arigo@adsl-89-217-207-222.adslplus.ch) joined #pypy.08:55
fijalaround like 396 there is a giant if "what to do for function X"08:55
fijalarigato: morning08:55
arigatohi08:57
kenaan12arigo extradoc 116d59f50abf58 15/planning/stm.txt: Update08:57
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) joined #pypy.08:58
cfbolzhello08:58
cfbolzarigato: thanks for the comment, good point08:58
cfbolzarigato: will probably add it to the post itself08:58
arigatook08:58
arigatoand hi08:58
arigatoI'm not sure any of it makes sense for someone that only knows Python :-)08:59
cfbolzarigato: yes, I guess so as well08:59
cfbolzarigato: but we always had kind of varied topics08:59
arigatosure08:59
cfbolzarigato: you managed to compile pypy-c with stm?09:01
arigatoyes, with --gc=none09:01
cfbolzcool09:02
arigatoit goes as far (and as fast) as I imagined this would go, and the next step is basically to start from scratch again (extradoc/planning/stm.txt)09:02
cfbolzheh09:02
arigato:-)09:02
cfbolzarigato: can you run two things in parallel?09:02
arigatoyes, it parallelizes nicely09:02
cfbolzdid you try something that conflicts too?09:03
arigatoI managed to run 4 richards on the 4 tannit09:03
cfbolznot bad09:03
arigatothis gets some conflicts too, probably due to imprecisions in the conflict detection mechanism09:03
cfbolzok09:03
arigatobut it still scales nicely09:03
fijalarigato: this does not yet contain the JIT part right?09:03
arigatofijal: no, but the JIT part should follow relatively easily09:04
arigato(I mean in the txt :-)09:04
kenaan12arigo extradoc 11d2cbf8878e65 15/planning/stm.txt: update09:06
zain (~textual@46.120.28.30) joined #pypy.09:10
AlekSi (~AlekSi@85.235.191.82) joined #pypy.09:17
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.09:20
dmitrey (~quassel@77.121.195.200) joined #pypy.09:24
bivab (~david@ip-62-143-59-5.unitymediagroup.de) joined #pypy.09:26
dmitreyfolk, what about weekly or monthly reports on numpypy progress in pypy blog? Do you like the idea?09:27
fijaldmitrey: yes09:27
fijaldmitrey: I'll post a report tomorrow09:27
dmitreyok09:27
dmitreyfijal: do you intend to make these posts regular, weekly or monthly?09:28
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.09:29
fijalI think weekly-bi-weekly09:33
fijalweekly might be a bit too much09:33
fijalmattip: I can do it if you want, unless you want to know more about compile :)09:35
Nick change: antocuni_ -> antocuni09:36
dmitreydo you mean once per fortnight?09:36
antocuniuhm, is there any python3 expert here? :-)09:36
antocuniconsider this code: http://paste.pocoo.org/show/541462/09:36
mattipfijal: almost done. Remeber my goal for the week?09:36
antocuniin python2 it's a syntaxerror because we cannot use exec() in presence of nested functions with free variables09:36
antocuniin python3, it just works (and prints 5, "incorrectly")09:37
antocuniI suppose that it's because now exec is a function and no longer a statement09:37
antocunido you know if it's explicitly written down somewhere (e.g., in a PEP)?09:38
fijaldmitrey: once ever week or two I would say :)09:38
cfbolzantocuni: probably not09:39
antocunipff09:39
arigatouh, obscure09:40
arigatoantocuni: what if you try, from f(), to return (h, x) ?09:40
antocunipff**209:40
antocunix is also 509:40
arigato:-(09:40
arigatoof course09:41
mattipfijal: sorry I'm going to have to renig. I did something like flat(a) but am now back to where I started, 09:41
mattipgetting errors that suggest the space does not recognize W_FlatIterator09:41
arigatobut if you kill the (uncalled so far) definition of h(), then x will suddenly be 43?09:41
mattipCould you clean it up?09:41
fijalcfbolz: would you come for a numpypy sprint to leipzig btw?09:41
fijalmattip: yeah09:41
fijalmattip: can you commit what you have?09:41
mattipon the way.09:42
antocuniarigato: no, still 509:42
arigatoah09:42
antocunithat's completely obscure09:42
arigatothat makes a tiny bit more sense09:42
antocuni*tiny*, yes09:43
arigatoit means that exec() with only one argument is pointless to assign variables09:43
kenaan12mattip numppy-flatitter 11794461ec9ef3 15/pypy/module/micronumpy/test/test_iter.py: more tests09:43
kenaan12mattip numppy-flatitter 1132b53b45a4c3 15/pypy/module/micronumpy/: failed attempt to redo compile.py09:43
cfbolzfijal: in principle yes, but it really depends on how my writing is going09:43
fijalcfbolz: ok09:43
fijalcfbolz: cool09:43
fijalI think it's very cool for mike to make such an offer btw09:43
antocuniarigato: yes, but it's a completely different behavior than python209:43
cfbolzyes totally09:43
antocuniI'd expect to find it written down somewhere09:44
fijalit's backwards incompatible isn't it?09:44
Action: mattip off to run errands, back later09:44
mattip (mattip@bzq-79-177-109-132.red.bezeqint.net) left #pypy ("bye").09:44
arigatoantocuni: it is documented actually09:45
arigatohttp://docs.python.org/release/3.2/library/functions.html#exec09:45
arigatosee the last Note;:"09:45
antocuniah, I see09:45
antocuniI wonder how it's implemented09:46
antocunibecause in the bytecode there is just a CALL_FUNCTION to exec09:47
antocuniso it needs to play with frames to get the locals(), I suppose09:47
Action: antocuni looks at the C source09:47
florian2 (~work@dslb-094-221-181-079.pools.arcor-ip.net) joined #pypy.09:47
arigatoI expect it's implemented by: just call globals() and locals() to fill in missing arguments09:48
antocuniah no09:48
antocuniif you don't pass anything, it uses globals() for both09:48
fijalhttp://morepypy.blogspot.com/2012/01/py3k-and-numpy-first-stage-thanks-to.html09:48
antocuniso I expect the global "x" to be 43 after09:49
arigatoeeeeh?09:49
antocuniuhm, no09:49
antocuniah indeed09:49
antocunilocals=globals only if you explicitly pass a globals09:50
antocuniyes indeed, if you don't pass anything it just calls PyEval_GetGlobals() and PyEval_GetLocals()09:51
antocuni"good"09:51
arigato":-)"09:54
antocunifijal: we should update the donation progress bars on pypy.org09:54
lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) joined #pypy.09:55
fijalantocuni: I'm doing it, but I can't do 17 different things at the same time09:55
lesshastehi09:55
fijalincluding replying to you on IRC09:55
antocunifijal: sorry, it was just meant as a reminder09:56
G2P (~G2P@fw-asn1.ornis.com) joined #pypy.09:56
christophler (~web49_an_@84.45.87.204) joined #pypy.09:58
kenaan12fijal pypy.org[extradoc] 11347134143935 15/: update progress bars09:59
kenaan12fijal pypy.org[extradoc] 110721767ab411 15/: merge09:59
fijalmeh10:00
fijalI updated but it does not refresh10:00
antocunifijal: it has refreshed for me10:00
antocuniuhm, but only numpy10:00
kenaan12fijal pypy.org[extradoc] 11f663f346f8dc 15/js/script2.js: make it not cache10:01
fijalwtf10:01
vad__ (~vad@77.72.196.122) joined #pypy.10:01
antocunifijal: ok, both bars look good now :-)10:02
fijalI'm not sure10:02
fijalthey're not for me10:02
antocunithey are for me after I clicked "refresh"10:02
vad (~vad@77.72.196.122) left irc: Ping timeout: 260 seconds10:03
dmitreyfijal: I have some free time right now, are there any work suitable for me? 10:04
lesshastedmitrey, all the parts of numpy needed to get openopt working :)10:09
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Read error: Connection reset by peer10:09
dmitreylesshaste: those missing funcs are mostly have to be done in interpreter level, while I guess I have  no enough time to learn it, I write in applevel10:10
lesshastedmitrey, it might be worth learning about the interpreter level now so you can work on it in the future10:11
lesshastedmitrey, a little cost now for a lot of gain later :)10:11
dmitreyI guess in future most of my efforts will be toward further OpenOpt development10:11
lesshastedmitrey, sure. I was just thinking that to get openopt and pypy working together in the near future, that might be the most efficient way to help10:13
lesshasteI could be wrong of course10:13
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) joined #pypy.10:13
lesshastefijal, any views?10:14
fijallesshaste: on openopt?10:15
stefanophi everybody! so after reading fijal's blogpost of october 27 I started working on the json decoder (just to start working on pypy). I was able to improve performance slightly, should I commit somewhere?10:15
lesshastefijal, on what would be most helpful at this point for dmitrey to do10:15
Action: lesshaste suspects the answer is not to increase the number of things to 18 that have to be done at the same time :)10:16
fijalstefanop: did you learn how to use jitviewer?10:17
apoirier (~apr@210.168.3.109.rev.sfr.net) joined #pypy.10:17
fijalstefanop: a branch on bitbucket + a pull request is usually a way to go10:17
kvdahi all10:17
stefanopfijal: yup! even if it took me a bit =)10:17
fijaldmitrey: I think it would be really useful if you learn our workflow10:17
fijaldmitrey: your patches are usually not directly applicable10:18
kvdafijal, is the whole of speed.pypy.org stored in bitbucket.org/pypy/benchmarks ?10:18
fijalbesides, tests should usually also test cases like views, multidim views, compund arrays10:18
fijalkvda: no, just the runner10:18
fijalkvda: the web site is on github10:18
fijalstefanop: and why only slightly? ;-)10:18
fijalstefanop: maybe we can look at it together then10:18
kvdai see, so it wasn't you that added me to the benchmarks access list? I got an email the other day and I though it might be the src to speed.pypy.org10:19
stefanopfijal: hey sure! thanks =) it's around 20% faster decoding, I wasn't able to speed up string concatenation, though10:19
fijalstefanop: did you try StringBuilder?10:20
kvdasrc access to speed.pypy.org would be nice so that I can see if I can implement some json serving for the main site10:20
kenaan12fijal numppy-flatitter 1193ed7361747f 15/pypy/module/micronumpy/: an absolutely crucial test10:20
fijalkvda: it's just that I added all pypy contributors to benchmarks10:20
fijalbecause it made sense10:20
fijalso everyone got a mail10:20
fijaldon't feel special ;-)10:20
kvdaah right10:20
fijalregarding speed - mail miquel tobami at gmail10:20
stefanopfijal: yup! it's slowing down by a lot the scanstring method10:20
fijalstefanop: scanstring?10:21
nephics (~jacob@78-72-71-194-no33.tbcn.telia.com) joined #pypy.10:21
stefanopfijal: in json/decoder.py10:21
fijalstefanop: it's impossible to write it using regex right?10:22
mattip (~mattip@bzq-79-177-109-132.red.bezeqint.net) joined #pypy.10:22
fijalstefanop: ok, let me look at your branch after I'm done with flatitter10:22
stefanopfijal: didn't actually try, I just focused on cleaning the current code and adding some speedup trick. I may try in a few minutes, though10:22
stefanop(forking)10:23
fijalok10:23
fijalmattip: ok, I commited a test that clearly shows the problem10:23
fijalmattip: I know RPython error messages are scary, but this one was actually readable10:23
fijal"no method X"10:23
Nick change: vad__ -> vad10:24
kenaan12fijal numppy-flatitter 115a27b974638a 15/pypy/module/micronumpy/compile.py: fix the wrong name10:25
fijalmattip: done10:27
mattipfijal: what is left till I can convince you it is ready for merge?10:27
fijallet me have a look10:28
fijalmight be nothing10:28
fijaldo you understand the problem btw?10:28
Action: mattip looking now...10:28
mattipahh. I never got an actual flatiterator compiled via the way I wrote it, right?10:29
fijalwell yes, because you had a typo10:31
fijalit said so very clearly IMO10:31
mattipIt was one of five "Blocked block" errors, and I was used to seeing the other four so the signal got lost in the noise.10:36
mattip;(10:36
fijalok10:36
fijalmattip: we have to decide if setitem needs jitdriver10:37
mattipok, now that we have a working zjit, it should be simple.10:37
fijalyes10:38
fijalfeel like doing it?10:38
mattipok.10:38
fijalbtw, please use a while loop10:38
fijalI don't remember exactly why10:38
mattipOK, it's easy to recast for.. as while... with a counter10:39
fijalyes10:39
mattipI have a few mour hours before I shut down for the weekend, I'll get back in an hour or two with progress report10:40
mattips/mour/more/10:40
fijalcool10:40
fijalmattip: I can merge it and finish few rough edges10:40
fijalI have to add another jitdriver on my branch though ;)10:40
mattipgive me a chance to stare at the  --viewloops, that's always fun :)10:41
fijalok10:42
fijaldid you notice btw that the big loop is not relevant?10:42
fijalit took me a while10:42
fijaler not loop10:42
fijalthe first big block of stuff10:42
squiddy (~squiddy@g224193025.adsl.alicedsl.de) joined #pypy.10:42
mattipit's all optimized out, right?10:43
fijalyou have two blocks10:44
fijalthe first one is "preambler"10:44
fijalthe second one is the correct one10:44
fijalpreamble10:44
fijalpreamble runs only once when starting to go into assembler10:44
fijaland then you loop into the second one10:44
mattipahh, in pygame view. Yes, you explained this before10:44
fijalit's to for example put loop-invariant code10:45
fijalwhich does not have to be in the loop, but has to be somewehre10:45
mattipfwiw, it's really cool. One of the best debugging tools I've seen. Seriously.10:45
cfbolzmattip: :-)10:47
fijal:-)10:47
fijalmattip: have you run jitviewer?10:47
fijalit's different and serves a different purpose but it's also cool10:48
mattipno, that's still way beyond me.10:48
fijalcommon10:48
mattipthe loop viewer cuts away all the cruft and lets you see whats important, where you spend time10:48
cfbolzfijal: jitviewer doesn't help with numpypy though, does it10:48
fijalit totally does10:49
fijalwhy wouldn't it?10:49
fijalmattip: we're big on tooling10:49
mattipon most of my real work I have 1:20 ratio loc code to tools 10:50
mattipand that's counting all the input testing as part of the code10:50
kenaan12RonnyPfannschmidt jitviewer 1134292f6ef484 15/: shuffle around app code, so we have templates+static in the package10:50
kenaan12RonnyPfannschmidt jitviewer 115b778b5ea015 15/.hgignore: some additions to .hgignore10:50
mattipas long as we're chatting, congrats on the fund raising successes.10:52
fijalI think we're more like 1:3 if you count tests :)10:52
kvda (~kvda@124-169-134-221.dyn.iinet.net.au) left irc: Quit: x___x10:53
stefanop (~stefanop@77.72.196.122) left irc: Read error: Connection reset by peer10:56
stefanop (~stefanop@77.72.196.122) joined #pypy.10:57
kenaan12fijal numppy-flatitter 11e0d5bd060533 15/pypy/module/micronumpy/test/test_zjit.py: finish the test11:03
kenaan12fijal numpy-back-to-applevel 112ba6d0106f54 15/pypy/module/micronumpy/: implement take for test_compile, also be more explicit about wrapping11:03
cfbolzfijal: because mapping back the trace you get from numpypy to python code is not so easy11:03
cfbolze.g. you don't have equivalent bytecodes11:03
dmitreyfijal: For my (openopt) users I would propose to compare PyPy vs CPython speed on DerApproximator (http://openopt.org/DerApproximator) (it requires minimal efforts in comparrison with other suite modules) but I see these funcs are missing for this : atleast_1d, atleast_2d, isfinite, hstack, asscalar, isscalar, asfarray, asarray, isnan. Is it posible to implement missing funcs on applevel for now, and rewrite those that should be present in interplevel later?11:04
dmitrey I could go for some of them right now.11:04
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 248 seconds11:04
fijalcfbolz: ok11:04
fijalcfbolz: but you get a description of operations (roughly)11:04
fijalit's useful for me, let's put it that way11:04
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.11:05
PiotrSikora (~none@nginx/adept/piotrsikora) left irc: Excess Flood11:05
fijaldmitrey: look11:06
fijaldmitrey: implementing those functions don't take that long11:06
fijaland we're not *that* worried about having to run benchmark today vs a week from now11:06
fijalso I would say we can wait a bit11:06
PiotrSikora (~none@nginx/adept/piotrsikora) joined #pypy.11:07
fijalobviously nothing stops you from implementing them and testing, but I think it's overall more effort11:07
fijaldmitrey: did you check how many of those are actually applevel?11:08
fijaland isnan is implemented on a branch11:08
fijalas is asarray I think11:09
dmitreywhy isnan and other funcs are not in main trunc?11:09
fijalbecause branch awaits a review11:09
zain (~textual@46.120.28.30) left irc: Quit: Computer has gone to sleep.11:09
fijaland also I'm adding stuff11:11
fijaldmitrey: why are you in so much rush?11:11
fijaldoes it really matter if the function is on trunk today or in a week?11:11
dmitreybecause people come to OpenOpt, try it, see how CPython is slow vs competitors ( C/Fortran soft or commercial soft) and never return to the project.11:13
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen11:13
dmitreyon unvectorized code11:13
fijalwell11:16
fijalfull numpy support is a few months away11:16
fijaldo you think it matters?11:17
fijalI'm serious here11:21
dmitreyWhat do you mean by "full numpy support"? dir(numpy) has >500 entries, and some of them are modules (like linalg, fft), while dir(numpypy) about 130.11:21
fijalwe won't get openopt supported before full numpy is working11:21
fijalyeah, I mean those 500 entries and those modules11:21
dmitreyBut really 70% of users use 30% of capabilities, and vise wersa11:21
fijalthis is not really true in numpy I think11:22
fijalwell11:22
dmitreyso porting 30 % of numpy capabilities would be enough to get OO working11:22
fijaldo you think we're doing some stupid functionality that noone cares about?11:22
dmitreyit doesn't really matter +/- 5%11:22
dmitreyor even 10%11:22
dmitreyno, there always be people interested in any involved routine11:23
fijalI think you're wrong but that does not matter11:23
fijalare we doing something in your opinion silly?11:23
dmitreybut isn't it better to focus on mainstream funcs like atleast_1d etc? And implement them at leas on applevel for now? 11:24
lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) left irc: Remote host closed the connection11:24
dmitreyhonestly, my impression that you are more interested in process than result :(11:24
mattipeach user has a different 30%, and interoperability with blas, PIL, opencv or other c code is part of the intersection of almost all.11:24
mattipyou must be very lucky if OpenOpt calls no ctypes stuff.11:25
dmitreyOpenOpt 100% pure Python-numpy-scipy11:25
fijaldmitrey: we do care about quality but am I working on something silly?11:25
lucianalthough implementing it app-level might not be a bad idea. the numpy folks should do that :)11:25
dmitreyI think you too much care of quality11:26
dmitreyBugs revealed later could be fixed on-demand11:26
mattiplucian: numpy has no convolution. scipy does. should it be done at app level?11:27
fijalthat's not how you make a working interpreter with a JIT11:27
dmitreyas well as some rare special things to be implemented11:27
fijaldmitrey: for example I would consider shapes of zero-sized arrays to be completely obscure corner case11:27
fijal(for example)11:27
fijalapparently you got hit by it11:27
lucianmattip: i don't really know. if i were part of the numpy community, i'd work on app-level versions of everything anyway11:27
dmitreyI got it while writing tests for hstack/vstack11:28
lucians/app-level/pure python/11:28
dmitreyI mean things like array subclassing11:28
dmitreysmall % of people deal with it11:28
dmitreyso it can be dereferred for futur11:29
dmitreye11:29
Da_Blitzhmm, there is a linux.conf.au talk that does somthing simmilar to the taint space11:29
fijaldmitrey: yes, subclassing got deferred11:30
fijaldmitrey: I'm fine with that11:31
fijalhowever empty_like that has obscure performance cannot be commited11:31
fijalespecially that we'll have to implement it again11:31
fijalthen why bother?11:31
fijaldo seriously few days make such a difference?11:32
kenaan12arigo extradoc 11f2a69fdb7e80 15/planning/stm.txt: Add more thinking.11:32
azanella (~azanella@189.6.80.131) joined #pypy.11:32
fijaldmitrey: I mean I appreciate your help and I already fixed some bugs, but "hastily hack something together" is not really what I believe in11:33
fijaland timelines are not "a year"11:33
fijalit's seriously around few months11:33
zain (~textual@46.120.28.30) joined #pypy.11:34
mikefc dmitrey: perhaps do a fork on github?  write the applevel code you need, and merge in pypy's changes as they happen and replace your code?11:34
mikefcum "fock on bitbucket"11:34
mikefcfork11:34
ronan (~ronan@host86-172-60-8.range86-172.btcentralplus.com) left irc: Quit: Ex-Chat11:35
arigatocfbolz: done writing the stm plan.  now I need to find an army of students, and go in holidays :-)11:35
amaury_ (~amaury_@74.125.57.34) joined #pypy.11:35
arigato...ah wait, I'm already in holidays.11:35
mikefcthen pypy could steal some of your testcases :)11:35
dmitreyI don't know how much time will pass till my applevel funcs will be overwritten by interplevel, maybe the time will be insufficient and thus lots of my efforts will be mere waste of time11:36
Tobu (~Tobu@unaffiliated/tobu) left irc: Read error: Operation timed out11:38
fijaldmitrey: that's what I'm trying to suggest11:38
fijaldmitrey: I would suggest if you want to help to learn our workflow and how to implement functions on interp-level if necessary11:42
kenaan12fijal numpy-back-to-applevel 11e656f86d3017 15/pypy/module/micronumpy/: write a jitdriver for take11:42
mattipfijal: I rewrote the zjit test to call [] on the flatiter http://paste.pocoo.org/show/541528 but11:43
dmitreyso where's the instructions? Yet I'm not sure they are not so difficult that I can't allocate enough time to do it11:43
fijaldmitrey: there are no real instructions11:43
fijalyou have to look at code and at tests11:43
mattipI'm still getting an assert in check_trace_count => loop is 011:44
fijalmattip: didn't I write that btw?11:44
fijalI would kill check_trace_count is not that interesting11:44
fijalmattip: ah11:44
fijalmattip: you have to hg up :)11:44
fijalmattip: I wrote those and fixed that11:44
mattipdid that.11:44
fijalmattip: you didn't merge conflicts or something11:44
mattipok. I'll start over.11:45
fijalyour head should be e0d5bd06053311:45
fijalno it's done11:45
fijalthe original test did not call any loop11:45
Tobu (~Tobu@2a01:e35:8a2c:b230:213:d3ff:fe85:5614) joined #pypy.11:47
Tobu (~Tobu@2a01:e35:8a2c:b230:213:d3ff:fe85:5614) left irc: Changing host11:47
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.11:47
anish_ (~anish@219.91.250.194) left irc: Ping timeout: 255 seconds11:49
fijalmattip: I'm off, will be back tonight maybe11:49
fijalI'll look into merging your branch it seems good IMO11:50
mattipmy internet too slow, missed the change11:50
fijalprovided test coverage is right :)11:50
mattiphave fun.11:50
mattipi will only be back sat eve your time.11:50
vad (~vad@77.72.196.122) left irc: Ping timeout: 255 seconds11:55
zain (~textual@46.120.28.30) left irc: Quit: Computer has gone to sleep.11:57
kenaan12arigo default 114451c9eca510 15/pypy/: Bah, Windows fails on another place that raises the same message.12:00
fijalronny: ping12:03
fijalronny: py.test coverage stopped working12:03
fijalhttp://paste.pocoo.org/show/541538/12:04
lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) joined #pypy.12:04
ronnyfijal: reinstall cov_core12:04
ronny(unfortunately cov_core simply cant be shipped easyly)12:05
fijalworked thx12:06
cfbolzarigato: the plan looks good to me12:08
cfbolzhas some subtle points and lots of tuning though12:08
cfbolzarigato: also, the interactions with the jit aren't clear to me12:12
nirbheek (~nirbheek@gentoo/developer/flyingspaghettimonster/nirbheek) joined #pypy.12:16
Tobu (~Tobu@unaffiliated/tobu) left irc: Read error: Operation timed out12:19
dmitrey (~quassel@77.121.195.200) left irc: Remote host closed the connection12:20
kenaan12fijal numpy-back-to-applevel 11b8cb7e93e732 15/pypy/module/micronumpy/test/test_numarray.py: increase test coverage12:21
mattipfijal: ping12:23
fijalpong12:23
mattipI would try to claim that the test you added does not call getitem12:23
mattipbut does iterate over the flatiterator12:23
fijalya12:23
mattipI was trying to add a jit_merge_point to getitem12:24
fijalthat's enough to make test_zjit work at all12:24
fijalyes and you said to leave it to you :)12:24
mattipI can do view_loops on regular tests? without zjit?12:24
fijalno12:24
fijalyou have to write test_zjit12:24
fijalthe thing is in order to loop12:24
fijalyou have to do a slice getitem or so12:24
mattipso in order to see how good my merge point is, I need a test in zjit12:25
fijalit won't get compiled until you have something like 3 iterations over the loop or 412:25
fijalyes12:25
mattipahh, so this http://paste.pocoo.org/show/541528/ is just too short?12:25
fijalI think 2 too short yes12:26
mattipok, thnks.12:26
nedbat (~nedbat@python/psf/nedbat) joined #pypy.12:32
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.12:39
lucianAlex_Gaynor: nice talk http://blip.tv/pycon-us-videos-2009-2010-2011/pycon-2011-the-data-structures-of-python-489816012:41
stakkars_ (~tismer@i59F72CE7.versanet.de) joined #pypy.12:46
Nick change: ojii -> ojii|afk12:52
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy.12:59
arigatoantocuni: a bug probably of module/_ffi12:59
arigatohttps://bugs.pypy.org/issue102013:00
arigatois there no way to run a _ffi-less pypy nowadays?13:01
arigatoI think that ctypes really need it13:01
cfbolzarigato: do you feel like reviewing the next prolog blog post?13:04
arigatocfbolz: yes, sure13:04
cfbolzarigato: it lives here: https://bitbucket.org/cfbolz/misc/src/8ce4283175bd/blog/blogpe2.rst13:05
cfbolzthank you13:06
arigatoit's a private repo13:06
cfbolzgrr, it shouldn't be13:07
cfbolzsec13:07
cfbolzarigato: now?13:07
arigatook13:08
cfbolzarigato: you should be able to push, or just send me a diff13:08
derdon (~derdon@p5DE89AF5.dip.t-dialin.net) joined #pypy.13:14
mattip (mattip@bzq-79-177-109-132.red.bezeqint.net) left #pypy ("bye").13:15
kenaan12mattip numppy-flatitter 11ef26cbf01db4 15/pypy/module/micronumpy/: add jit_merge_point s, add tests for them, tests fail13:20
antocuniarigato: no, _ffi is required for ctypes nowadays13:22
antocuniyou could try to fish an older version of ctypes which is based on _rawffi, though13:23
antocuniwhy do you think that the problem is _ffi, btw?13:23
vad (~vad@77.72.196.122) joined #pypy.13:32
Tobu (~Tobu@unaffiliated/tobu) left irc: Read error: Operation timed out13:44
khs (~khs@151.43.202.84.customer.cdi.no) joined #pypy.13:49
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection13:51
DasIch (~DasIch@p3E9901EF.dip.t-dialin.net) joined #pypy.13:53
zain (~textual@bzq-79-182-196-50.red.bezeqint.net) joined #pypy.14:00
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.14:00
Arach (~arach@128-75-162-28.broadband.corbina.ru) left irc: Read error: Operation timed out14:06
mattip (~mattip@bzq-79-177-109-132.red.bezeqint.net) joined #pypy.14:07
mattipfijal: I am out of time. If you could make the tests pass that would be great.14:08
mattipfijal: and if you could finish the branch and merge it that would be great too.14:09
mattipttfn14:09
mattip (~mattip@bzq-79-177-109-132.red.bezeqint.net) left irc: Client Quit14:09
Rhy0lite (dje@nat/ibm/x-dujkygizglasgmyn) joined #pypy.14:10
Nick change: ojii|afk -> ojii14:12
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection14:17
EnCuKou (~encukou@ip-94-113-220-25.net.upcbroadband.cz) left irc: Remote host closed the connection14:17
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.14:19
Arach (~arach@FTTB-dynamic-79.104.203.17.ranetka.ru) joined #pypy.14:21
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 245 seconds14:21
azanella (~azanella@189.6.80.131) left irc: Quit: Ex-Chat14:22
azanella (~azanella@189.6.80.131) joined #pypy.14:22
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: Remote host closed the connection14:28
amaury_ (~amaury_@74.125.57.34) left irc: Ping timeout: 240 seconds14:29
ronan (~ronan@host86-172-60-8.range86-172.btcentralplus.com) joined #pypy.14:31
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.14:31
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: Remote host closed the connection14:34
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.14:36
vad_ (~vad@77.72.196.122) joined #pypy.14:37
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.14:39
vad (~vad@77.72.196.122) left irc: Ping timeout: 240 seconds14:39
lesshaste (~lesshaste@87-194-206-189.bethere.co.uk) left irc: Remote host closed the connection14:43
amaury_ (amaury_@nat/google/x-zoqqfqpatoxhmcqy) joined #pypy.14:45
whitelynx (~whitelynx@63.241.75.144) joined #pypy.14:52
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) joined #pypy.14:54
arigatoantocuni: unclear so far, actually, because I'm getting different segfaults when restarting14:54
cocoatomo (~cocoatomo@p2118-ipbf901souka.saitama.ocn.ne.jp) left irc: Remote host closed the connection14:54
arigatothe first one made me think that something was missing around the external call done by libffi.py14:54
arigatobut it doesn't really seem to be the case after all14:55
antocuni"good"14:56
SukhE (~SukhE@unaffiliated/sukhe) joined #pypy.15:00
arigatocfbolz: pushed two comments15:12
cfbolzarigato: thanks15:13
cfbolzarigato: what did you think, overall?15:13
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy.15:13
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: Read error: Connection reset by peer15:13
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.15:14
arigatoit's a nice introduction, but also Prolog appears better suited to partial evaluation than tracing15:14
cfbolzarigato: yes15:14
cfbolzarigato: I tried writing it in Python, but it got a bit too verbose15:14
arigatoI imagine15:14
cfbolzpattern matching is really concise, but some of the "let's bind a variable here" can be obscure15:15
arigatolisp, maybe? :-)15:15
cfbolzarigato: or haskell? :-)15:15
Action: arigato runs away15:15
arigatobut yes, maybe15:15
lucianwhat is this for, if you don't mind my asking?15:15
cfbolzarigato: the tracing would be easier, yes15:16
cfbolzarigato: but the pe is annoying, because of the state15:16
arigatoyes, I see15:16
cfbolzlucian: it's the second part of the recent blog post: http://morepypy.blogspot.com/2012/01/comparing-partial-evaluation-and.html15:16
luciani see15:16
cfbolzarigato: I guess you just write your own monad there15:16
arigatoI don't know, maybe it really makes sense to use a different language for the blog post too?15:17
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: Remote host closed the connection15:17
cfbolzor just use a state monad or something15:17
cfbolzarigato: you mean rewrite it to, say, lisp?15:17
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.15:17
arigatoI don't know if lisp is the definite answer15:17
arigatoand I also don't know if it's worth trying hard to find the best answer :-)15:18
cfbolzarigato: indeed :-)15:18
cfbolzarigato: the original goal was to use a mathematical formalism15:18
cfbolzbut I wanted to test it somehow15:18
arigatoyes, I see15:19
arigatomaybe in the blog post you can replace some arguments with "..."?15:19
arigatobut then it becomes a bit hand-wavy15:19
cfbolzyes15:20
cfbolzarigato: I guess I'll add an explanation what the FullTrace is, and live with the fact that it's mostly understandable for people who know prolog15:20
arigatoas I said, I think it can be easier to group some of the extra arguments into a single name15:21
cfbolzarigato: it's a bit of a tradeoff, though15:22
cfbolzarigato: I'll try to see what it looks like15:22
luciancfbolz: your PhD thesis is quite interesting15:23
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) left irc: Remote host closed the connection15:23
cfbolzlucian: it's not written yet15:23
luciancfbolz: the post is :)15:23
inhahe (~inhahe@adsl-074-186-083-219.sip.mia.bellsouth.net) joined #pypy.15:24
Action: arigato debugging greenlets in pypy: how is it possible to end up with the frame's vable_token pointing to some prebuilt object...15:24
arigatoit's supposed to be a pointer in the C stack15:24
jterrace (~jterrace@aegis.CS.Princeton.EDU) joined #pypy.15:25
cfbolzlucian: thanks15:26
luciani'm itching to write a scheme implementation now :)15:27
cfbolzlucian: heh. please go ahead, would love to look at that15:27
cfbolzlucian: I can mail you the full code, if you want15:28
Rhy0litecfbolz: does your analysis suggest that tracing is different than partial evaluation because more variables are held constant / specialized and guards added?15:28
cfbolzRhy0lite: tracing is different from partial evaluation because there is really nothing "partial" about tracing. tracing is just evaluation15:29
Rhy0litehehe.  that's one way to put it :-)15:29
cfbolzRhy0lite: but then, the optimizers that are used in a tracer are again just like partial evaluators15:29
jterracefijal: fyi, cumsum is marked as check mark on numpy status page, but it just throws a NotImplementedError15:29
luciancfbolz: i'm not entirely sure i understand that language. I may have an easier time with just pypy. but sure, that'd be nice if i find the time15:29
Rhy0liteyes, that's what I mean by everything held constant15:29
cfbolzso in a sense tracing is taking those parts of partial evaluation that work (the "just evaluate those things that you can, and leave the others") and replacing the parts that don't (controlling unfolding) by a much more pragmatic mechanism15:30
Rhy0litecould it be considered an extreme case of PE?15:30
cfbolzRhy0lite: it's PE with an oracle that tells which paths to look at15:30
cfbolzthe oracle is driven by actual executions15:30
Rhy0litethat's a good description15:31
cfbolzsamuele came up with it15:31
cfbolzRhy0lite: the blog posts are the nucleus of an eventual paper that will try to explain this to PE researchers15:32
Rhy0liteyes.  it's an excellent start.15:33
cfbolzthanks15:35
cfbolzarigato: yes, putting the two strange arguments into a structure looks nicer15:35
cfbolzarigato: thanks for the idea15:35
arigato:-)15:35
arigatocfbolz: nice description (a few lines above) of what tracing :-)15:37
arigato...is :-)15:37
cfbolzshould copy it somewhere15:37
Action: lucian should maybe learn prolog15:37
cfbolzlucian: if you like puzzles, it's a wonderful language15:38
arigato...test_greenlet segfaults even on a "shadowstack --jit off" pypy...15:39
cfbolz:-((15:40
arigatoit looks like the vable_token it's reading is just uninitialized value, because there is no write to vable_token that writes a different value than 0 or -1, apart in generated assembler15:41
arigatohum no, uninitialized values should be zero... :-/15:41
Action: arigato wants undodb-gdb again15:42
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Ping timeout: 252 seconds15:43
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.15:48
SukhE (~SukhE@unaffiliated/sukhe) left irc: Quit: leaving15:51
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) left irc: Remote host closed the connection15:52
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) joined #pypy.15:53
Arach (~arach@FTTB-dynamic-79.104.203.17.ranetka.ru) left irc: Ping timeout: 240 seconds15:56
arigato(did I mention it recently: "snow, snow!")15:59
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.16:00
jukeks (~Jukkis@gateway/shell/tkk.fi/x-lvotyrczjiahldop) joined #pypy.16:01
cfbolzarigato: :-)16:03
jukeksi just noticed  how PIL's putdata() on an 1920x1080 png image takes 9 seconds using pypy 1.7 and 0.2 seconds using python 2.716:04
jukekshi!16:04
arigatoppppfpfpfppffff recompiled with gcc -O0 instead of -O1, the bug disappears16:06
florian2 (~work@dslb-094-221-181-079.pools.arcor-ip.net) left irc: Ping timeout: 252 seconds16:07
Arach (~arach@128-75-240-5.broadband.corbina.ru) joined #pypy.16:08
arigatoah no, it's just hitting the maximum stack usage16:09
Action: arigato increased it and it's again hitting the original problem16:09
AlekSi (~AlekSi@85.235.191.82) left irc: Remote host closed the connection16:09
derdon_ (~derdon@pD9E1D731.dip.t-dialin.net) joined #pypy.16:12
dmalcolm (~david@c-24-61-12-82.hsd1.ma.comcast.net) joined #pypy.16:15
derdon (~derdon@p5DE89AF5.dip.t-dialin.net) left irc: Ping timeout: 245 seconds16:15
azanella (~azanella@189.6.80.131) left irc: Quit: Ex-Chat16:16
vad_ (~vad@77.72.196.122) left irc: Remote host closed the connection16:17
Arach (~arach@128-75-240-5.broadband.corbina.ru) left irc: Ping timeout: 240 seconds16:20
arigatoargh, across a call, root_stack_top is not preserved16:21
azanella (~azanella@189.6.80.131) joined #pypy.16:21
Arach (~arach@128-75-163-251.broadband.corbina.ru) joined #pypy.16:21
Moku (~John@osbk-4db06c7e.pool.mediaWays.net) left irc: Read error: Connection reset by peer16:25
stefanop (~stefanop@77.72.196.122) left irc: Remote host closed the connection16:25
vad (~vad@77.72.196.122) joined #pypy.16:26
Moku (~John@osbk-4db06c7e.pool.mediaWays.net) joined #pypy.16:26
sunoano (~sa@unaffiliated/sunoano) left irc: Quit: Leaving.16:34
Moku (~John@osbk-4db06c7e.pool.mediaWays.net) left irc: Ping timeout: 276 seconds16:41
dmitrey (~quassel@77.121.195.200) joined #pypy.16:42
lucian (~lucian@93-97-174-114.zone5.bethere.co.uk) left irc: Read error: Operation timed out16:42
vad (~vad@77.72.196.122) left irc: Remote host closed the connection16:43
arigato (~arigo@adsl-89-217-207-222.adslplus.ch) left irc: Quit: See you16:44
jukeks (Jukkis@gateway/shell/tkk.fi/x-lvotyrczjiahldop) left #pypy.16:44
amaury_ (amaury_@nat/google/x-zoqqfqpatoxhmcqy) left irc: Ping timeout: 252 seconds16:50
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen16:51
Telugodu (~rambaho@202.3.77.231) joined #pypy.16:53
Tobu (~Tobu@unaffiliated/tobu) left irc: Ping timeout: 272 seconds16:55
Moku (~John@osbk-4db06c7e.pool.mediaWays.net) joined #pypy.16:56
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection16:58
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.17:05
cfbolz (~cfbolz@fwstups.cs.uni-duesseldorf.de) left irc: Quit: Leaving17:05
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.17:13
florian2 (~work@dslb-094-221-181-079.pools.arcor-ip.net) joined #pypy.17:14
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Read error: Operation timed out17:19
mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) left irc: Remote host closed the connection17:20
fijaljterrace: well I guess this is "too bad"17:21
fijaljterrace: it's not intended to be a scientific measure just "rough measure of progress"17:21
fijalmaybe we should remove those functions or at least not have them in __all__17:22
Telugodu (~rambaho@202.3.77.231) left irc: Quit: Ex-Chat17:23
G2P (~G2P@fw-asn1.ornis.com) left irc: Quit: Leaving.17:29
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) joined #pypy.17:32
stakkars_ (~tismer@i59F72CE7.versanet.de) left irc: Quit: schnarch17:49
mcdonc (~mcdonc@ip72-209-213-54.dc.dc.cox.net) joined #pypy.17:49
nedbat (~nedbat@python/psf/nedbat) joined #pypy.17:50
jterracefijal: yeah i figured you did it in an automated way18:00
jterracejust wanted to let you know18:00
fijalI am aware and I don't care :)18:02
Alex_Gaynorjterrace: much as we're motivated, reviewing all 500 numpy.* names by hand seems a bit much :)18:02
jterracehah i know18:02
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.18:02
fijaljterrace: if you use this list to make serious assumptions, go and write a test suite for your project instead :)18:03
jterraceoh i have a test suite18:09
jterraceit just fails with NotImplementedErrors :)18:09
fijalwell you can help it not fail ;-)18:13
fijalbut improving the quality of blue viewer is not the way18:13
jterraceagreed18:14
jterracewhat's the status of the axis iterator?18:14
fijalit works18:14
fijalexcept for stuff like all and any18:14
jterraceoh i also noticed this morning that numpypy.nan doesn't exist18:14
jterracebut nan did seem to work when i tested it18:14
fijalthat's a oneliner18:15
jterraceyeah18:15
jterraceand then maybe implementing numpy.isnan would be easy then too18:15
fijalI think it's already done on a branch18:15
fijalisnan is also done on the same branch18:15
jterraceoh ok18:15
fijalAlex_Gaynor: sudo review18:15
jterracewhenever i get some free time, ill help out where i can18:16
fijaljterrace: cool18:16
fijaleh it seems pypy has killed my machine :/18:19
fijalmwhudson: feel like helping a bit with sensor stuff?18:19
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Operation timed out18:20
nedbat (~nedbat@python/psf/nedbat) left irc: Ping timeout: 255 seconds18:22
dimazest (~dimazest@balticom-202-59.balticom.lv) left irc: Ping timeout: 276 seconds18:23
kenaan12fijal numppy-flatitter 112adecee1eb73 15/pypy/module/micronumpy/test/test_zjit.py: fix tests18:23
apollo13 (apollo13@unaffiliated/apollo13) left #pypy ("Leaving").18:24
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection18:25
Alex_Gaynorfijal: back-to-applevle branch?  I'll do it after I finish eating18:26
fijalomg eating?18:26
fijalyou have *code to review*18:26
fijalhow can you eat?18:26
fijalI think I confused mattip by committing a failing test18:27
Alex_Gaynorwell, I take the food out of the foil, and then I put it in my mouth18:27
Action: Alex_Gaynor can probably multitask18:27
florian2 (~work@dslb-094-221-181-079.pools.arcor-ip.net) got netsplit.18:27
Arach (~arach@128-75-163-251.broadband.corbina.ru) got netsplit.18:27
ronan (~ronan@host86-172-60-8.range86-172.btcentralplus.com) got netsplit.18:27
nephics (~jacob@78-72-71-194-no33.tbcn.telia.com) got netsplit.18:27
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) got netsplit.18:27
zk (~is@whatit.is) got netsplit.18:27
mwhudson (~mwh@linaro/mwhudson) got netsplit.18:27
jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) got netsplit.18:27
mikefc (~mike@60-242-240-196.static.tpgi.com.au) got netsplit.18:27
JaredW (~JaredW@118.142.17.238) got netsplit.18:27
guibou (~guibou@kraken.insecable.net) got netsplit.18:27
mvt (~mvt@53530442.cm-6-4a.dynamic.ziggo.nl) got netsplit.18:27
legogris (~legogris@c80-217-207-157.bredband.comhem.se) got netsplit.18:27
nopf (~froschif@147.172.224.4) got netsplit.18:27
Ormod_ (~valtha@ohmu.fi) got netsplit.18:27
alexsuraci (~alexsurac@li175-235.members.linode.com) got netsplit.18:27
idnar (~quassel@unaffiliated/idnar) got netsplit.18:27
exarkun (~exarkun@wordeology.com) got netsplit.18:27
jterrace (jterrace@aegis.CS.Princeton.EDU) left #pypy.18:28
guibou (~guibou@kraken.insecable.net) returned to #pypy.18:28
idnar (~quassel@unaffiliated/idnar) returned to #pypy.18:28
exarkun (~exarkun@wordeology.com) returned to #pypy.18:28
#pypy: mode change '+o exarkun' by ChanServ!ChanServ@services.18:28
nopf (~froschif@147.172.224.4) returned to #pypy.18:28
JaredW (~JaredW@118.142.17.238) returned to #pypy.18:28
legogris (~legogris@c80-217-207-157.bredband.comhem.se) returned to #pypy.18:28
ronan (~ronan@host86-172-60-8.range86-172.btcentralplus.com) returned to #pypy.18:28
nephics (~jacob@78-72-71-194-no33.tbcn.telia.com) returned to #pypy.18:28
Arach (~arach@128-75-163-251.broadband.corbina.ru) returned to #pypy.18:28
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) returned to #pypy.18:28
mwhudson (~mwh@linaro/mwhudson) returned to #pypy.18:28
florian1 (~work@dslb-094-221-181-079.pools.arcor-ip.net) joined #pypy.18:28
jacob22 (~jacob@c-c4c4e055.1321-1-64736c11.cust.bredbandsbolaget.se) returned to #pypy.18:28
zk (~is@whatit.is) returned to #pypy.18:29
Alex_Gaynorfijal: WTF, who wirtes "reduce(product, a.shape) == 0" instead of "any(s == 0 for s in a.shape)"18:31
fijalpeople who use numpy compulsively?18:32
Ormod (~valtha@ohmu.fi) joined #pypy.18:33
Alex_Gaynorfijal: apparently :/18:33
fijalwell18:33
Alex_Gaynorfijal: we really need someplace where we list all of the changes we made to numpy code, like we have -modified18:33
fijalI found out people who use numpy does not necessarily know python18:33
fijalnote that18:33
fijalAlex_Gaynor: not yet18:33
fijalI would go for "files we have now are to be removed and replaced by originals"18:34
Alex_Gaynorfijal: you `if 0`'d something out an `arra.flags.maskna` any reason not to add flags even if they're all false?18:34
fijalother than more work to add if 0, no18:34
fijalwe should at least correctly report current flags18:35
Alex_GaynorATM Contigious and writable will be true18:35
Alex_Gaynoraligned might be as well18:35
fijalcontigious - depends18:36
Alex_Gaynorah yes, contigious depends on strides18:36
fijalaligned - I would like to add a flag to raw malloc that says "aligned"18:36
fijaland allocate our arrays aligned18:36
fijaleven more18:36
fijalI would allocate them aligned and rounded to the nearest 128bits18:37
fijalor whatever is the size of XMM register18:37
fijalso we don't have to deal with crap when doing SSE18:37
Alex_Gaynoryes, cheaper to always align18:37
Alex_Gaynoralthough you can still have issues with slices18:37
Alex_Gaynoron smaller datatypes18:37
fijalyou mean they might start unaligned?18:37
mvt (~mvt@53530442.cm-6-4a.dynamic.ziggo.nl) got lost in the net-split.18:38
Ormod_ (~valtha@ohmu.fi) got lost in the net-split.18:38
alexsuraci (~alexsurac@li175-235.members.linode.com) got lost in the net-split.18:38
mikefc (~mike@60-242-240-196.static.tpgi.com.au) got lost in the net-split.18:38
florian2 (~work@dslb-094-221-181-079.pools.arcor-ip.net) got lost in the net-split.18:38
Alex_Gaynoryes, you have e.g. numpy.ones(10000, dtype="int8")[3:]18:38
arigato (~arigo@xdsl-188-154-145-222.adslplus.ch) joined #pypy.18:38
fijalI would say for views it's "too bad"18:38
fijalwe can think about it later, but most views are not amenable to vectorization very well anyway18:38
Alex_Gaynorwell, it's not like it's super difficult to have a small loop so it's aligned18:38
fijalwell that's not true18:39
sunoano (~sa@cpc3-nott15-2-0-cust249.12-2.cable.virginmedia.com) joined #pypy.18:39
sunoano (~sa@cpc3-nott15-2-0-cust249.12-2.cable.virginmedia.com) left irc: Changing host18:39
sunoano (~sa@unaffiliated/sunoano) joined #pypy.18:39
Alex_Gaynor"whatever" though18:39
fijalbut vectorizing small datatypes is probably not working anyway :)18:39
Alex_Gaynoryes, indeed, because neither JIT nor RPython know how to do math on them18:39
Alex_Gaynorwhy don't they anyways18:39
Alex_Gaynorjust easier?18:39
fijalask arigato why there is no math on SHORT :)18:39
Alex_Gaynorarigato: Why is there no math on SHORT?18:40
Alex_Gaynor:)18:40
dimazest (~dimazest@balticom-202-59.balticom.lv) joined #pypy.18:40
apoirier (~apr@210.168.3.109.rev.sfr.net) left irc: Quit: Leaving...18:42
fijalAlex_Gaynor: axis needs an overhaul in order to make it fast18:46
fijalbut I suppose "we don't worry now"18:46
Alex_Gaynorfijal: to fix iteration order?18:46
Alex_Gaynorfijal: is it slower than NumPy or just suboptimal?18:46
fijalit's probably very slow18:47
fijalas in stuff does not get virtualized18:48
Alex_Gaynoroh, too bad18:48
kenaan12fijal numppy-flatitter 11d4ba294094ac 15/pypy/module/micronumpy/: fix tests18:49
fijalno sorry18:49
fijalit'18:49
fijals a[bool_array] = stuff18:49
Alex_Gaynorfijal: ok we should add a product function to numpy, there is size = 1 for x in s: size *= x like 50 million times18:50
fijalyeah general indexing by bool arrays can be done better18:50
fijalyes :)18:50
fijalfeel like doing it?18:50
fijalsize = support.product(shape)18:50
Alex_Gaynorfijal: I'll do that, if you make `flatten()` call ravel and then do a copy if needed18:51
fijalon that branch?18:51
fijalok18:51
Alex_Gaynoryes, since you added ravel18:51
fijalsure18:52
fijalone sec18:52
Alex_Gaynorfijal: make sure it doesn't doubel copy on weird things18:52
fijalI'm reviewing numppy-flatitter18:52
nephics (~jacob@78-72-71-194-no33.tbcn.telia.com) left irc: Quit: Leaving18:54
fijalwow18:56
Alex_Gaynor?18:56
fijalmattip wrote a one-page-comment about how iterators and strides work18:56
fijaland then put it in the test file :)18:56
fijalI'm moving it to interp_iter18:57
Alex_Gaynorfijal: why did you change bool method?18:59
Rhy0liteWhy doesn't PyPy source include a top-level Makefile to build targetpypystandalone?19:00
aboudreault (~alanb@osgeo/member/aboudreault) joined #pypy.19:00
Alex_Gaynorfijal: there's a comment # xxx check rpythonization, remove it I guess?19:00
Alex_Gaynoralso,why is invert defined twice19:00
fijalyes remove it19:00
fijalwhat did I do with bool method?19:00
kenaan12fijal numppy-flatitter 1113420231bfda 15/pypy/module/micronumpy/test/test_numarray.py: a few missing tests19:01
fijalno idea19:01
kenaan12fijal numppy-flatitter 1135aaba0342c7 15/pypy/module/micronumpy/: move around a very useful comment19:01
Alex_Gaynorfijal: you changed it from bool(x) to x != 019:01
vad (~vad@host221-110-dynamic.25-79-r.retail.telecomitalia.it) joined #pypy.19:01
kenaan12fijal numppy-flatitter 1181af4f4c5c21 15/: close merged branch19:01
fijalRhy0lite: what do you mean?19:01
kenaan12fijal default 11caa2d47782ba 15/pypy/module/micronumpy/: (mattip, fijal reviewing) Merge numppy-flatitter branch (yes, with all the typos). This improves flat iterator and ...19:01
Alex_Gaynorfijal: also, you have a test_ravel with no tests19:01
fijalAlex_Gaynor: ah, I hoped it works on stuff other than float/int19:01
fijalbut it does not19:01
fijalso I call for_computations19:01
fijalmove it back to bool please?19:01
Alex_Gaynorcan you do it?19:02
Alex_Gaynoreverything else looks ok btw19:02
fijalyes19:02
fijalso what do I have to do?19:02
Alex_GaynorI'll do support.product now19:02
Alex_Gaynorfijal: remove that XXX, fix some of the changes to types.py19:02
fijalchange bool, improve flatten to call ravel19:02
Alex_Gaynoryes19:02
fijalimprve test19:02
fijalanything else?19:02
Alex_Gaynornope19:02
Rhy0litehttp://paste.pocoo.org/show/541764/19:03
fijalRhy0lite: well, because we're not from GNU church I believe ;-)19:04
fijalI personally have nothing against it19:04
Rhy0liteMake existed before GNU :-)19:05
fijalI would add echo "this will take some time, about an hour on a modern machine, enjoy!"19:05
mvt (~mvt@53530442.cm-6-4a.dynamic.ziggo.nl) joined #pypy.19:05
Rhy0liteI would warn people in the output from translate.py, not the Makefile19:05
kenaan12fijal numpy-back-to-applevel 11a113c99b0ec2 15/pypy/module/micronumpy/types.py: remove cruft from type.py19:05
Rhy0liteIt just seems easier than expecting people to remember the magic incantation19:06
fijalno, I don't want to see it19:06
Rhy0liteand one can place Makefile at the top-level that runs a sub-Make in the goal directory19:06
fijalit seems only useful for the kind of people that would call the Makefile :)19:06
fijalbut I won't insist19:06
fijalyeah, I think a make like this is fine19:06
fijalAlex_Gaynor: uh?19:07
fijalAlex_Gaynor: my test_ravel has 3 asserts19:08
Alex_Gaynorfijal: not the one in test_numeric.py19:08
kenaan12fijal numpy-back-to-applevel 11b6f8f63b46ec 15/pypy/module/micronumpy/test/test_numarray.py: remove print19:08
fijalah that one19:09
fijalremove that one simply19:09
fijalravel is interp-level19:09
Alex_Gaynoryou remove it19:09
zain (~textual@bzq-79-182-196-50.red.bezeqint.net) left irc: Quit: Computer has gone to sleep.19:10
Alex_Gaynorfijal: FWIW IMO, flat get and set didn't need a jit driven19:13
fijalyeah?19:13
fijalhow do you do it otherwise?19:13
fijala.flat[:] = stuff19:14
fijal?19:14
Alex_GaynorIMO it should call setslice on base somehow19:14
Alex_Gaynorso you automatically get memcpy if it's flat and others19:14
fijalyou cannot19:14
fijalnot with step at least19:15
fijalflatiter's view with step cannot be expressed using strides19:15
Alex_Gaynorstupid strides19:16
fijalwell no, you seriously can't do much better19:16
fijalI thought about it :)19:16
kenaan12alex_gaynor numpy-back-to-applevel 114ecd878caff2 15/pypy/: merged default19:16
kenaan12alex_gaynor numpy-back-to-applevel 116deca798be13 15/pypy/module/micronumpy/appbridge.py: Don't use __import__ with lots of arguments, it can results in double imports and other silly ...19:16
kenaan12alex_gaynor numpy-back-to-applevel 11dc4cacb0a280 15/pypy/module/micronumpy/: cleanup19:16
kenaan12alex_gaynor numpy-back-to-applevel 116ab531465c38 15/pypy/module/micronumpy/: merged upstream19:17
dmitreyfijal: why wouldn't commit tests for unimplemented yet funcs? it could be automatically traced, e.g. def test_diag: if 'diag' not in numpypy.__dict__ : return (i.e. if unimplemented yet)19:17
alexsuraci (~alexsurac@li175-235.members.linode.com) joined #pypy.19:17
florian1 (~work@dslb-094-221-181-079.pools.arcor-ip.net) left irc: Quit: Leaving.19:17
dmitreythose tests could be checked in CPython numpy19:18
khs (~khs@151.43.202.84.customer.cdi.no) left irc: Quit: Leaving19:18
fijalAlex_Gaynor: hey you broke stuff19:18
fijalit no longer imports tests19:18
Alex_Gaynorfijal: ?19:19
fijalImportError: cannot import name support19:19
Alex_Gaynoreh, forgot to add19:19
fijalwhen I try to run test_numarray.py19:19
Alex_Gaynorfijal: pushed19:20
kenaan12alex_gaynor numpy-back-to-applevel 11706c23fe89c5 15/pypy/module/micronumpy/support.py: forgotten file19:20
fijaldmitrey: it's a mess to track19:20
fijaldmitrey: why would you test against cpython if we have no functionality?19:20
dmitreyit could be for future purposes, someone will have to go for it in either way19:21
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.19:21
dmitreyand as soon as he will have an implementation, it will be automatically tested19:21
ericflo (~ericflo@75.103.8.110) joined #pypy.19:22
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.19:24
fijaldmitrey: there is a bit no point in writing a test when you don't have an implementation19:24
vad (~vad@host221-110-dynamic.25-79-r.retail.telecomitalia.it) left irc: Remote host closed the connection19:24
fijaleven if you can test it on numpy19:24
kenaan12fijal numpy-back-to-applevel 11938d4dcbaa69 15/pypy/module/micronumpy/: make flatten use ravel, remove previous version, improve tests19:24
kenaan12fijal numpy-back-to-applevel 119c98420734a5 15/pypy/module/test_lib_pypy/numpypy/core/test_fromnumeric.py: remove an empty test19:24
fijalfor example corner cases of implementation might be different19:24
fijaland those will be untested19:24
fijalanyway, it does not make anything any easier, so why bother?19:24
fijalAlex_Gaynor: ok, happy?19:24
dmitrey (~quassel@77.121.195.200) left irc: Remote host closed the connection19:25
Alex_Gaynorfijal: using copy() for flatten() on scalars is wrong19:26
Alex_Gaynorand why are you calling get_concrete19:26
fijalreshape calls get_concrete anyway19:26
Alex_Gaynoreh, that's silly, but ok19:26
fijalyeah a bit19:27
Alex_Gaynorcan you fix flatten on scalar?19:27
fijalshould that not copy?19:27
fijalok19:27
Alex_Gaynorflaten() on scalar returns an array length 1, not a scalar19:27
Alex_Gaynorsee my previous implementation of it19:27
fijaloh19:27
fijalok19:27
fijalthen these commits make a bit no sense19:27
fijalit does not make things any simpler IMO19:27
fijaland has bugs19:28
fijalwell, you did not test it19:28
kenaan12antocuni py3k 11742faaa192e9 15/pypy/interpreter/astcompiler/symtable.py: bah, fix the signature of note_import_star, which I broke in d882e23cc8ae19:29
kenaan12antocuni py3k 11116575861a5d 15/pypy/interpreter/test/test_syntax.py: kill a lot of snippets from the invalid cases: in python3 it is now possible to call exec() inside a function which...19:29
kenaan12antocuni py3k 11b421a4de9302 15/pypy/interpreter/test/test_syntax.py: port the syntax of some AppLevel tests to python319:29
kenaan12antocuni py3k 117586f1183218 15/pypy/interpreter/test/test_syntax.py: we no longer need to import print_function from __future__. Also, s/__builtin__/builtins19:29
kenaan12antocuni py3k 11c5b8818fe586 15/pypy/interpreter/test/test_syntax.py: kill import print_function from __future__ also here, and adjust the import of StringIO19:29
kenaan12antocuni py3k 1158915e03655e 15/pypy/interpreter/test/test_syntax.py: don't import unicode_literals from __future__, and adapt unicode-->str and str-->bytes19:29
kenaan12antocuni py3k 1159514c43c5c0 15/pypy/interpreter/test/test_syntax.py: fix the syntax of exception clauses. Now all applevel tests pass with --appdirect=/usr/bin/python319:29
kenaan12antocuni py3k 11ba117ac345c3 15/pypy/interpreter/test/test_syntax.py: exec can no longer modify the local scope, so use an explicit scope instead19:29
kenaan12antocuni py3k 111b76cb58345b 15/pypy/interpreter/test/test_syntax.py: kill this test about warnings and conditional expressions. We stopped issuing the warning long time ago19:29
kenaan12antocuni py3k 115127c7fa289c 15/pypy/interpreter/test/test_syntax.py: exec() can no longer modify the local scope, so we need to explicitly pass a namespace and fish the interesting thi...19:29
kenaan12antocuni py3k 11f3e4991575d5 15/pypy/: kill the --appdirect option and leave only --runappdirect, which by default tries to run the tests with the python3...19:29
kenaan12antocuni py3k 117aeb630cb68e 15/pypy/interpreter/test/test_syntax.py: fish the name of the python3 executable from the conftest19:29
kenaan12antocuni py3k 11a47a0d7d1ec5 15/pypy/interpreter/test/test_compiler.py: py3k-ify the syntax of this test19:29
kenaan12antocuni py3k 111089c2a88694 15/pypy/interpreter/test/test_compiler.py: py3k-ify the syntax of this test as well19:29
kenaan12antocuni py3k 11cf23b6ea0ec0 15/pypy/interpreter/test/test_compiler.py: py3k-ify the syntax of these three tests19:29
kenaan12antocuni py3k 113014f5e44e49 15/pypy/interpreter/test/test_compiler.py: we can no longer index the keys() of a dict, use next(iter(...)) instead19:29
kenaan12antocuni py3k 11131fb90722ed 15/pypy/interpreter/test/test_compiler.py: sanitize the test class hierarchy: now we have only one compiler, so there is no need to have a base class and two ...19:29
fijalspammer19:29
antocunisorry :-)19:30
fijalAlex_Gaynor: ok, let me clean it up19:30
fijalAlex_Gaynor: ah, I can explain why .ravel is hard to implement on a virtual array19:31
fijalit's even in planning19:31
kenaan12edelsohn ppc-jit-backend 112743ff5a00f1 15/pypy/jit/backend/ppc/runner.py: Move definition of self.asm from __init__() to setup(). Define get_latest_force_token() and get_on_leave...19:31
Rhy0litenow translation with the JIT gets a little farther19:32
Alex_Gaynorfijal: ok19:33
bivab (~david@ip-62-143-59-5.unitymediagroup.de) left irc: Quit: bivab19:34
kenaan12alex_gaynor py3k 11d41f30d6c1ca 15/pypy/interpreter/test/test_compiler.py: tiny beautification19:39
antocuni (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds19:39
nedbat (~nedbat@python/psf/nedbat) joined #pypy.19:41
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) left irc: Ping timeout: 245 seconds19:43
pjenveyspam sponsored by Google19:43
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) joined #pypy.19:54
dmitrey (~quassel@77.121.195.200) joined #pypy.19:54
rguilleberthi19:55
zain (~textual@46.120.28.30) joined #pypy.20:02
Moku (~John@osbk-4db06c7e.pool.mediaWays.net) left irc: Read error: Connection reset by peer20:03
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) joined #pypy.20:04
amaury (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Client Quit20:05
Kaskuka (~John@osbk-4db06c7e.pool.mediaWays.net) joined #pypy.20:06
Tobu (~Tobu@unaffiliated/tobu) left irc: Read error: Operation timed out20:06
fijalAlex_Gaynor: now, review20:07
Alex_Gaynorfijal: what do I need to review?20:07
fijalpushing....20:08
fijalI think a wifi router restart would help20:08
kenaan12fijal numpy-back-to-applevel 11891a2ea64919 15/pypy/module/micronumpy/: clean up scalar reshape and ravel20:08
fijalpushed20:08
fijalRhy0lite: so what's the problem now?20:09
Alex_Gaynorfijal: does Scalar.shape = [x] fail correctly?20:09
Alex_Gaynor(it should raise a NotImplementedError or so)20:10
fijalsilently fauls?20:10
fijalfails20:10
Alex_Gaynor?20:11
Alex_Gaynorthat doesn't sound like a good behavior, IMO20:11
fijalwell it'll still raise ValueError if the shape's product is not 120:12
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) left irc: Ping timeout: 255 seconds20:12
fijalanyway20:12
fijalcan we make this not branch-relevant?20:12
Alex_Gaynorsure20:12
Alex_GaynorI guess it's ready20:12
fijalI can fix it once the branch lands20:12
fijalcool20:13
fijalI merge it then20:13
fijalAlex_Gaynor: you're a dick for generating conflicts20:14
fijalchanging \ to () for imports20:14
Alex_Gaynorfijal: () is much nicer!20:15
fijaland mercurial is so lame20:15
pjenveygranted fijal, it seems like everything causes a conflict for you20:15
pjenveylike newlines20:15
pedronis (~pedronis@73-53.195-178.cust.bluewin.ch) joined #pypy.20:17
fijalpjenvey: adding tests is the usual cause20:17
fijalso right now mercurial took two tests20:18
fijalmerged the identical assert20:18
fijaland put the rest as a conflict20:18
Alex_Gaynorfijal: it does that so much in jit tests20:18
pjenveyyea that one you pasted seemed kind of lame20:18
fijalpjenvey: ya20:18
fijalpjenvey: I get on average 3 of those per day20:18
fijalI'm considering how nice were the days of svn20:19
fijalI'll kill the next person who claim merging with mercurial is easier than svn20:19
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.20:19
pjenveywas that when you merged from default to a feature brnach?20:19
fijalAlex_Gaynor: I'm tempted to reply to dude on hackernews about "how compatible pypy ever be" with "it's going to be "how compatible pygtk is soon""20:19
Alex_Gaynorfijal: heh, true20:20
fijaleh quote error20:20
fijalpjenvey: ya20:20
fijalpjenvey: but now I'm merging back and it's as lame20:20
zain (~textual@46.120.28.30) left irc: Quit: Computer has gone to sleep.20:20
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.20:20
Rhy0litefijal: I'm not sure how far back to begin the paste, but the current failure is: http://paste.pocoo.org/show/541792/20:21
fijalRhy0lite: do you pass test_ztranslation by chance?20:22
fijalit seems like the answer is "no"20:22
fijalRhy0lite: do you need help with test_ztranslation?20:22
fijalIf i can get an account somewhere I can probably help20:22
Rhy0liteno, it doesn't20:23
fijalwell20:23
fijalthis is the first thing to fix20:23
fijalit won't work until you fix it even20:23
Rhy0literight, but it's the same problem there20:23
fijalgood, right?20:24
Rhy0liteyes20:24
fijaldo you need help?20:24
kenaan12fijal numpy-back-to-applevel 11a49c9222bf0e 15/: close merged branch20:24
fijalbrb, router reboot20:24
kenaan12fijal default 1175784011b73f 15/: (fijal, alex) Merge numpy-back-to-applevel branch. This branch removes a bunch of functionality from micronumpy to ...20:24
fijal_ (~fijal@196-210-150-33.dynamic.isadsl.co.za) joined #pypy.20:26
Rhy0litefijal: If you want to help, you should get an account on the GCC Farm20:28
fijal_I'm not really that thrilled to help, but I can ;-)20:29
fijal_as in if you get stuck I can unstuck you20:29
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) left irc: Ping timeout: 272 seconds20:29
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer20:29
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.20:30
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer20:32
zain (~textual@46.120.28.30) joined #pypy.20:33
CIA-603ita 07roundup * 10#934/subprocess hangs in 1.7 (but not in 1.6 or 1.5): 20:33
CIA-6[chatting] I have tried the Linux 64 version from today both jit and nojit and i am still20:33
CIA-6getting the build hanging either in the middle or near ... * 14https://bugs.pypy.org/issue93420:33
Tobu (~Tobu@unaffiliated/tobu) joined #pypy.20:35
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.20:36
Nick change: fijal_ -> fijal20:39
fijal_ (~fijal@196-210-150-33.dynamic.isadsl.co.za) joined #pypy.20:48
fijal (~fijal@196-210-150-33.dynamic.isadsl.co.za) left irc: Disconnected by services20:48
Nick change: fijal_ -> fijal20:48
dmitrey (~quassel@77.121.195.200) left irc: Remote host closed the connection20:49
zain (~textual@46.120.28.30) left irc: Quit: Computer has gone to sleep.20:53
kenaan12edelsohn ppc-jit-backend 111a4cf2e87fd7 15/pypy/jit/backend/detect_cpu.py: Add ppc64_64 backend name.20:55
kenaan12edelsohn ppc-jit-backend 11d52428409628 15/pypy/jit/backend/ppc/test/test_ztranslation.py: PPC version of test_ztranslation20:55
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.20:56
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer20:56
antocuni (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) joined #pypy.20:58
amaury_antocuni: good work on py3k21:03
antocunithank you21:03
antocuniI'm trying to fix all the tests in interpreter/ before adding new stuff21:04
antocuniuhm, I'm confused21:04
amaury_yes, this is the right place to start :/21:04
antocuniin pypy 1/2 already evaluates to 0.521:04
antocuniin the py3k branch21:04
amaury_from __future__ import true_division21:04
amaury_by default21:04
antocuniyes, but in pycompiler.py we still use futureFlags_2_721:05
amaury_I may have removed the test21:05
amaury_I mean, I've hardcoded the value of the flag21:05
antocuniah, I see21:06
antocunimakes sense as well21:06
Telugodu (~rambaho@202.3.77.231) joined #pypy.21:08
antocunibut yes, in pypy the default co_flag is 0x24321:08
antocuniwhile in cpython is 0x4321:08
fijaleh21:08
fijalmwhudson: ping?21:08
Action: fijal brings out the green dust, some skulls and tarot cards21:09
fijalpedronis: did you take part in the ppc backend by chance?21:10
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen21:10
amaury_antocuni: see changeset 4c46f703ae2821:10
pedronisfijal: you mean the old one? no21:10
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.21:11
antocuniamaury_: ok, thanks21:11
fijalpedronis: only mwhudson?21:11
antocunibut I suppose we still need to fix the co_flag21:11
pedronisyes21:11
antocuniI don't understand where the 0x200 comes from21:11
fijalpedronis: I'm trying to look at code clearly revived from those days21:12
fijalbut without some talking-to-death I can't tell how it could ever possibly be RPython21:12
fijalwas it ever RPython btw?21:12
amaury_antocuni: maybe PyCF_DONT_IMPLY_DEDENT = 0x20021:12
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer21:12
antocuniah yes21:13
Guest__ (~textual@216.190.27.14) joined #pypy.21:13
antocuniindeed I tried it from the interactive prompt21:13
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.21:13
antocuninow the question is: why it's not there on cpython?21:13
amaury_IIRC pypy interactive prompt is a bit different from CPython21:14
amaury_in CPython you are obliged to finish a block with an empty line21:14
amaury_But I don't know if this is relevant21:14
antocuniright21:15
antocunifrom a quick grep, that flag seems to be never set in cpython21:15
antocunianyway, it's surely not important21:15
antocuniI have to go now21:15
antocunibye21:15
Nick change: Guest__ -> rekamso21:16
Telugodu (~rambaho@202.3.77.231) left irc: Quit: Ex-Chat21:16
kenaan12fijal ppc-jit-backend 11086f6c58e2e9 15/pypy/jit/backend/: kill some already supposed to be dead people21:20
kenaan12fijal ppc-jit-backend 1120a7afe2c42d 15/pypy/jit/backend/: merge default21:20
antocuni (~antocuni@host62-122-dynamic.6-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds21:20
arigato (arigo@xdsl-188-154-145-222.adslplus.ch) left #pypy ("See you").21:21
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer21:23
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.21:23
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection21:24
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen21:25
berdario (~Tiibiidii@host142-83-dynamic.2-87-r.retail.telecomitalia.it) joined #pypy.21:31
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.21:41
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.21:42
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.21:44
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer21:44
amaury_does it make sense to have a split() function in pypy.rlib.rstring?21:44
fijalamaury_: rpython strings have split no?21:50
fox_ (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer21:50
amaury_yes21:50
fijalso what's wrong with that?21:51
amaury_rstring.split() is a RPython implementation21:51
amaury_str.split() has lltype and ootype implementations21:51
tilgovi (~randall@couchdb/developer/tilgovi) joined #pypy.21:53
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen21:53
amaury_btw, I think I have a good solution to forbid \x00 in file names21:53
amaury_I use the annotator to "prove" that a string passed to os.open &co has no NUL byte21:54
fijalsounds like a receipe for a disaster IMO ;-)21:55
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.21:55
amaury_well, it's not too difficult21:56
asmeurer (~asmeurer@dhcp-altamirano-166.resnet.nmt.edu) joined #pypy.21:56
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.21:56
fijalno21:56
fijalit's difficult to fix all the places it can't prove stuff21:56
fijalamaury_: but does sound like a reasonable idea21:57
amaury_A new space operation, space.str0_w()21:57
amaury_returns a SomeString with no_NUL=True21:58
amaury_I had to modify a few places21:58
amaury_mostly in interp_posix.py and imp/importing.py21:58
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Read error: Connection reset by peer21:59
amaury_I had a hard time finding the annotation of some "rpython" functions22:00
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.22:00
amaury_like os.path.join22:00
mwhudson_ (~mwh@121-73-77-183.cable.telstraclear.net) joined #pypy.22:00
mwhudson_fijal: hello22:00
fijalmwhudson_: hey22:00
amaury_this one is annotated in rpython/extregistry.py22:00
amaury_a mess22:00
fijalmwhudson_: can you check out pypy on ppc-jit-backend?22:00
mwhudson_if you tell me how :-)22:01
mwhudson_don't appear to have any hg checkout of pypy on this machine22:01
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection22:02
fijalhg up ppc-jit-backend?22:03
fijalhg pull -u probably22:03
fijalmwhudson_: got it?22:04
mwhudson_no22:04
mwhudson_i'm cloning the repo now22:04
mwhudson_so that'll take a while i guess22:04
fijalya22:05
fijalin the meantime22:05
fijalhttps://bitbucket.org/pypy/pypy/src/ppc-jit-backend/pypy/jit/backend/ppc/ppcgen/codebuilder.py22:05
fijalhas instances of descs from https://bitbucket.org/pypy/pypy/src/ppc-jit-backend/pypy/jit/backend/ppc/ppcgen/form.py22:06
fijalthe question is - was this ever RPython?22:06
mwhudson_probably not?22:07
mwhudson_i think this is the usual pypy metaprogramming type thing22:08
fijalyes22:09
fijalbut those instances land on something that you call22:09
fijalwith assembler22:09
fijalIMO22:09
fijallike this22:10
fijalself.mc.load(r.r30.value, r.SP.value, WORD)22:10
fijalin runtime22:10
mwhudson_mm22:11
mwhudson_well we definitely did have ppc code generation in a runtime binary :)22:11
mwhudson_s/runtime/translated/22:11
mwhudson_can you dig out what was svn trunk in about feb 2007?:)22:12
fijalyou made me choke22:13
Action: fijal pulls again some skulls and green dust22:13
mwhudson_hah22:13
fijalI guess so22:13
mwhudson_although i don't understand how it can of worked, __call__ has never been rpython has it?22:13
fijalI never understood how the timeshifter worked though :)22:13
fijalno22:13
fijalnever22:13
fijalyes, the same here22:13
fijalit looks precisely like the code that's never RPython22:14
fijalor there was some specialization22:14
fijalbut also this kind of *args and **kwds22:14
mwhudson_it's possible that the version of my ancient ppc assembler that we used was hacked somehow22:14
mwhudson_more than the one that's now in the tree22:14
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer22:14
fijalfeb 2007 you say?22:14
mwhudson_yeah, around then22:15
fijalyou know hg invocation by chance?22:15
mwhudson_no22:15
fijalbefore EU period ended?22:15
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.22:15
fijalwas it ever speeding up anything?22:15
mwhudson_yeah, shortly before22:15
fijalor timeshifter never did22:15
mwhudson_fijal: some things, but we never got register allocation to work sensibly22:15
fijalok22:15
mwhudson_so it used heaps of stack22:15
azanella (~azanella@189.6.80.131) left irc: Ping timeout: 240 seconds22:16
mwhudson_+ timeshifter problems, heaps of code generated etc22:16
mwhudson_fijal: i need to go out soon22:16
mwhudson_fijal: but can look into this later22:16
fijalmwhudson_: cool, I'll try to dig it22:16
mwhudson_(i'll leave hg cloning away...)22:16
fijalI suppose the correct neuron didn't fire22:16
fijal"ah this was this weird specialization" was what I was hoping for ;-)22:16
mwhudson_heh22:16
mwhudson_hg doesn't seem to do sensible defaults very well22:17
mwhudson_i don't have a progress bar on this clone22:18
fijalno no no22:18
fijalhg is not a vcs22:18
fijalis a vcs construction toolkit22:18
mwhudson_ah right22:18
fijaljust like roundup22:18
mwhudson_so i'm using rcs in a cvs world22:18
mwhudson_feels like progress!22:18
Action: mwhudson_ afk anyway22:18
Action: fijal never used rcs22:18
fijalmwhudson_: thanks!22:18
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen22:24
amaury_ (~amaury_@46-127-23-192.dynamic.hispeed.ch) left irc: Ping timeout: 244 seconds22:29
kenaan12fijal ppc-jit-backend 1177891e3dee7b 15/pypy/jit/backend/ppc/ppcgen/form.py: this was definitely not supposed to be commited22:30
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.22:35
kenaan12fijal ppc-jit-backend 1171425577dae9 15/pypy/jit/backend/ppc/ppcgen/codebuilder.py: rpythonize the assembler22:45
fijalRhy0lite: I also broke stuff :/22:46
Rhy0liteum, why are you using rassembler?22:46
Rhy0litethat's not suppose to work22:46
squiddy (~squiddy@g224193025.adsl.alicedsl.de) left irc: Quit: Leaving22:47
fijalwell how else are you supposed to do that?22:47
fijalI think you can kill rassembler22:47
Rhy0liteyes, there is a lot of dead files22:48
fijalbut you have to make sure that there is something similar done with the assembler22:48
fijalright now the assembler is just plain not-rpython22:48
Rhy0liteokay22:48
fijalit was rpythonized before using rassembler22:48
Rhy0litebut rassembler.py is not up to date22:48
fijalnow if rassembler does not work at all you need to do something else22:49
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.22:49
fijaleither bring rassembler up to date or somehow make it RPython22:49
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection22:49
fijalRhy0lite: as far as I'm concerned this backend if by far not done22:49
Rhy0litecan you please revert the change that breaks stuff22:49
fijalsure22:50
Rhy0liteI didn't claim it was done22:50
kenaan12fijal ppc-jit-backend 118c2aa655427d 15/pypy/jit/backend/ppc/ppcgen/codebuilder.py: Backed out changeset 71425577dae922:50
fijalwell it's not done in a sense there is absolutely no point in trying translation22:51
Rhy0liteokay22:51
fijalfirst dead stuff should be removed22:51
fijalespecially dead test files22:51
fijalbecause they are harmful22:51
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Read error: Connection reset by peer22:51
Rhy0litedo you see my disagreeing?22:51
fijalyes?22:51
fijalno22:52
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) joined #pypy.22:52
fijalthen you should import tests from metainterp, like x86 backend does22:52
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) joined #pypy.22:52
fijalrassembler need has to be addressed22:52
Rhy0liteyes22:52
fijalI can write the necessary bit if needed22:52
JaRoel|4d (~jaroel|4d@office.fourdigits.nl) left irc: Remote host closed the connection22:53
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 245 seconds22:54
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) joined #pypy.22:56
mwhudson_fijal: hey22:59
fijalmwhudson_: hey, got it figured22:59
mwhudson_fijal: looks like the old ppc jit used a low level interface22:59
mwhudson_ah cool22:59
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/103523:00
bbot2Started: 15http://buildbot.pypy.org/builders/own-macosx-x86-32/builds/78923:00
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/21823:00
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/4623:00
fijalit had a wrapper that created an RPython version23:00
gtaylor (~gtaylor@108-196-160-156.lightspeed.tukrga.sbcglobal.net) left irc: Quit: Konversation terminated!23:00
mwhudson_ok23:00
`fox` (~fox@host65-111-dynamic.30-79-r.retail.telecomitalia.it) left irc: Ping timeout: 252 seconds23:02
mikefc (~mike@60-242-240-196.static.tpgi.com.au) joined #pypy.23:03
Rhy0litehow do I escape from "push creates new remote head" hell for a file that I tried to add?23:03
mwhudson_insn.Insn_None__GPR_GPR_IMM(_PPC.stw,23:04
mwhudson_                                        [gv_size, gv_result, IntConst(lengthoffset)]))23:04
mwhudson_isn't as cute as code.stw(rA=..) but it works i guess23:04
bbot24Failure: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/103523:04
fijalRhy0lite: hg pull -u23:05
fijalhg merge && hg ci -m 'merge' && hg push23:05
Rhy0liteI did pull and update and merge23:05
bbot24Failure: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/4623:05
fijaldid you commit the merge?23:05
bbot24Failure: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/21823:06
kenaan12edelsohn ppc-jit-backend 110696f19f1bba 15/pypy/jit/backend/ppc/test/test_zll_random.py: Add test_zll_random for PPC.23:08
kenaan12edelsohn ppc-jit-backend 118e5ed1cad830 15/pypy/jit/backend/ppc/test/test_zll_random.py: merge23:08
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/1036 [12fijal]23:09
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/219 [12fijal]23:09
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/47 [12fijal]23:09
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) joined #pypy.23:09
kenaan12fijal default 11c9343ef21049 15/pypy/: improve fake objspace and fix translation23:09
bbot213Exception: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/1036 [12fijal]23:10
bbot213Exception: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/219 [12fijal]23:10
bbot213Exception: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/47 [12fijal]23:10
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/1037 [12fijal]23:10
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/220 [12fijal]23:10
bbot2Started: 15http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64-2/builds/48 [12fijal]23:10
Rhy0litefijal: I seem to forget that hg lies23:10
Rhy0litein the sense that I cannot specify files when committing a merge23:10
Rhy0liteeven if the files are the ones being merged23:11
nedbat (~nedbat@python/psf/nedbat) left irc: Disconnected by services23:11
__main__ (~main@adsl-99-173-15-158.dsl.pltn13.sbcglobal.net) joined #pypy.23:11
nedbat_ (~nedbat@python/psf/nedbat) joined #pypy.23:11
fijalno23:12
fijalyou cannot specify files while merging23:12
fijalRhy0lite: generally you have to shift your workflow23:12
fijalhg *does not* tolerate local changes23:12
fijalif you want local changes, you have to either commit them to some obscure branch23:12
fijalor shelve them23:12
Rhy0liteyes, very annoying23:12
fijalthere is an extension for that23:12
Rhy0liteor create lots of clones23:12
fijalI'm fine by now23:13
fijalI have two clones and work with them23:13
fijalusing mostly shelve23:13
fijalbut also commiting stuff23:13
fijalyou can also commit and not push but this is dangerous23:13
Rhy0liteyes23:13
Rhy0liteI'm much more used to having multiple local changes and only committing the ones that are ready23:13
Rhy0liteby specifying files23:14
fijalI know23:14
fijalbut you have to change that habit :)23:14
fijalunfortunately if you ask me23:15
Rhy0liteyes, yes23:15
Rhy0liteand yes23:15
Rhy0liteI want a VCS that adapts to me!23:15
fijalI usually have tons of small feature branches23:15
fijaland commit stuff there23:15
fijalactually you get what you ask for23:15
fijalhg is very easy to extend23:15
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) joined #pypy.23:15
fijalwhich also means that defaults are insanely crap23:15
fijalbe careful what you wish for :)23:15
fijalok I should sleep23:15
Rhy0liteokay23:15
fijalAlex_Gaynor: thanks for the review!23:15
Rhy0litethanks for your help!23:15
Rhy0litegood night23:16
Rhy0lite (dje@nat/ibm/x-dujkygizglasgmyn) left irc: Quit: Leaving23:16
lucian (~lucian@cpc1-newc15-2-0-cust84.gate.cable.virginmedia.com) left irc: Ping timeout: 276 seconds23:21
mwhudson_ (~mwh@121-73-77-183.cable.telstraclear.net) left irc: Ping timeout: 245 seconds23:23
aboudreault (~alanb@osgeo/member/aboudreault) left irc: Quit: Leaving23:23
whitelynx (~whitelynx@63.241.75.144) left irc: Quit: Ex-Chat23:27
tilgovi (~randall@couchdb/developer/tilgovi) left irc: Remote host closed the connection23:35
nedbat_ (~nedbat@python/psf/nedbat) left irc: Ping timeout: 247 seconds23:37
whyking (~quassel@ip68-14-13-208.ri.ri.cox.net) left irc: Ping timeout: 244 seconds23:39
derdon_ (~derdon@pD9E1D731.dip.t-dialin.net) left irc: Remote host closed the connection23:42
fprimex (~brent@fl-67-233-144-239.dhcp.embarqhsd.net) left irc: Quit: quit23:54
pnguyen (~pnguyen@cpe-66-68-12-3.austin.res.rr.com) left irc: Quit: pnguyen23:55
mwhudson_ (~mwh@121-73-77-183.cable.telstraclear.net) joined #pypy.23:56
--- Sat Jan 28 201200:00

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