#pypy IRC log for Thursday, 2008-06-26

antocuni (n=antocuni@host207-123-dynamic.16-79-r.retail.telecomitalia.it) left irc: "Leaving"00:00
joe_mp_ (n=joe@p549C74F9.dip.t-dialin.net) left irc: Read error: 110 (Connection timed out)00:02
hpk (n=chatzill@p549F7068.dip.t-dialin.net) left irc: Read error: 110 (Connection timed out)00:08
joe_mp__ (n=joe@p549C6388.dip.t-dialin.net) left irc: Read error: 110 (Connection timed out)00:09
arianepaola_ (n=ariane@unaffiliated/arianepaola) left irc: Read error: 60 (Operation timed out)00:11
arianepaola_ (n=ariane@unaffiliated/arianepaola) joined #pypy.00:13
mgod (n=mgod@adsl-68-73-206-193.dsl.sfldmi.ameritech.net) joined #pypy.00:34
oierw (n=mathew@ joined #pypy.00:38
Hamled (n=Charles@ joined #pypy.01:07
cfbolz (n=cfbolz@p4FDAC229.dip0.t-ipconnect.de) left irc: "Leaving"01:09
mgod (n=mgod@adsl-68-73-206-193.dsl.sfldmi.ameritech.net) left irc: 01:51
mgod (n=mgod@adsl-68-73-206-193.dsl.sfldmi.ameritech.net) joined #pypy.01:51
mgod (n=mgod@adsl-68-73-206-193.dsl.sfldmi.ameritech.net) left irc: Remote closed the connection01:52
joe_mp__ (n=joe@p549C6CE2.dip.t-dialin.net) joined #pypy.01:53
fzzzy (n=donovan@office.mochimedia.net) left irc: 02:09
dialtone (n=dialtone@adsl-99-136-101-166.dsl.pltn13.sbcglobal.net) left irc: Read error: 110 (Connection timed out)03:09
fijalmwhudson: why you've got so distinct way of writing comments that I can usually guess?03:18
mwhudsonfijal: well, i am a native english speaker03:18
mwhudsonand i'm probably a bit more sarcastic than most people in comments03:19
fijalcan you recognize native/non-native from a 2-line comment?03:19
fijalI think it depends on language usage03:20
fijalindeed, you use words I would never think of03:20
fijal        # XXX because we call inputconst already in replace_malloc, we can't03:20
fijal        # modify the instance, we have to modify the 'rtyped instance'03:20
fijal        # instead.  horrors.  is there a better way?03:20
fijalthe distinctive part here is "horrors"03:20
headius (n=headius@ left irc: 03:48
arianepaola_ (n=ariane@unaffiliated/arianepaola) left irc: Read error: 54 (Connection reset by peer)04:09
Hamled (n=Charles@ left irc: Read error: 60 (Operation timed out)04:11
smokey__ (n=smokey@p549A169E.dip0.t-ipconnect.de) joined #pypy.04:16
smokey_ (n=smokey@p549A17F4.dip0.t-ipconnect.de) left irc: Read error: 110 (Connection timed out)04:33
carrus85 (n=carrus85@ joined #pypy.05:22
joe_mp_ (n=joe@p549C47E2.dip.t-dialin.net) joined #pypy.05:43
Nick change: panni_ -> panni05:47
Nick change: panni -> panni_05:47
headius (n=headius@ joined #pypy.05:55
arianepaola (n=ariane@unaffiliated/arianepaola) joined #pypy.05:57
Nick change: arianepaola -> arianepaola_05:59
joe_mp__ (n=joe@p549C6CE2.dip.t-dialin.net) left irc: Read error: 110 (Connection timed out)06:04
mwhudson_ (n=mwh@118-93-161-210.dsl.dyn.ihug.co.nz) joined #pypy.06:11
mwhudson (n=mwh@118-93-172-167.dsl.dyn.ihug.co.nz) left irc: Read error: 110 (Connection timed out)06:22
mwhudson_ (n=mwh@118-93-161-210.dsl.dyn.ihug.co.nz) left irc: Read error: 104 (Connection reset by peer)06:22
mwhudson_ (n=mwh@118-93-161-210.dsl.dyn.ihug.co.nz) joined #pypy.06:23
mwhudson_ (n=mwh@118-93-161-210.dsl.dyn.ihug.co.nz) left irc: Read error: 104 (Connection reset by peer)06:31
headius (n=headius@ left irc: 06:46
mwhudson_ (n=mwh@118-93-108-100.dsl.dyn.ihug.co.nz) joined #pypy.06:46
headius (n=headius@ joined #pypy.06:48
headius (n=headius@ left irc: Remote closed the connection06:49
fijal (n=fijal@c-76-25-206-12.hsd1.co.comcast.net) left irc: Remote closed the connection06:51
mwhudson_ (n=mwh@118-93-108-100.dsl.dyn.ihug.co.nz) left irc: Read error: 104 (Connection reset by peer)06:52
mwhudson_ (n=mwh@118-93-108-100.dsl.dyn.ihug.co.nz) joined #pypy.06:53
panni_ (i=hannes@ip-88-152-22-213.hsi.ish.de) left irc: "( www.nnscript.de :: NoNameScript 3.81 :: www.XLhost.de )"06:53
Rhamphoryncus (n=rhamph@unaffiliated/rhamphoryncus) left irc: 07:39
dialtone (n=dialtone@adsl-75-6-226-104.dsl.pltn13.sbcglobal.net) joined #pypy.07:58
headius (n=headius@ joined #pypy.08:09
jubo1 (n=cspenc@cpe-76-87-81-219.socal.res.rr.com) left irc: Read error: 104 (Connection reset by peer)08:26
jubo1 (n=cspenc@cpe-76-87-81-219.socal.res.rr.com) joined #pypy.08:29
hpk (n=chatzill@p549F608C.dip.t-dialin.net) joined #pypy.08:30
mwhudson_ (n=mwh@118-93-108-100.dsl.dyn.ihug.co.nz) left irc: Read error: 110 (Connection timed out)08:33
antocuni (n=antocuni@host101-126-dynamic.14-87-r.retail.telecomitalia.it) joined #pypy.08:45
carrus85 (n=carrus85@ left irc: "Leaving"08:54
headius (n=headius@ left irc: 08:56
ghaering (n=gerhard@ joined #pypy.09:15
CIA-303arigo 07async-del * r56083 10pypy/: 09:47
CIA-3issue371 in-progress09:47
CIA-3A branch in which to avoid having app-level __del__() methods09:47
CIA-3called from random places.09:47
jubo1 (n=cspenc@cpe-76-87-81-219.socal.res.rr.com) left irc: "Lost terminal"09:48
CIA-303arigo 07roundup * 10#371/__del__ called from random places: [chatting] Added: pypy/branch/async-del/ - copied from r56082, pypy/dist/ Log: issue371 in-progress A branch in which to avoid having ap ...09:48
harobed (n=stephane@pda57-1-82-231-115-1.fbx.proxad.net) joined #pypy.09:56
jewel_ (n=jewel@dsl-242-163-111.telkomadsl.co.za) joined #pypy.10:12
CIA-303arigo 07async-del * r56084 10pypy/pypy/objspace/std/test/test_listobject.py: A test that shows what all this is about.10:26
CIA-303arigo 07async-del * r56085 10pypy/pypy/ (5 files in 3 dirs): Attempting to fix the issue by delaying calls to app-level dels.10:27
nshepperd (n=neil@CPE-124-180-205-31.vic.bigpond.net.au) joined #pypy.10:38
joe_mp_ (n=joe@p549C47E2.dip.t-dialin.net) left irc: Read error: 110 (Connection timed out)10:46
cfbolz (n=cfbolz@p4FDAAB66.dip0.t-ipconnect.de) joined #pypy.11:04
antocunihi carl!11:09
arianepaola_ (n=ariane@unaffiliated/arianepaola) left irc: Read error: 104 (Connection reset by peer)11:12
pjenvey_ (n=pjenvey@astound-69-42-4-166.ca.astound.net) left irc: 11:13
hpk (n=chatzill@p549F608C.dip.t-dialin.net) left irc: Read error: 104 (Connection reset by peer)11:36
pedronis (i=pedronis@vitaly.openend.se) joined #pypy.11:38
xorAxAx   java emu at ~20% of the real time speed: http://www-jpc.physics.ox.ac.uk/11:38
xorAxAxs/java/& x86/11:38
cfbolzpedronis: hi!11:39
pedroniscfbolz: hi11:39
hpk (n=chatzill@p549F608C.dip.t-dialin.net) joined #pypy.12:14
jewel_xorAxAx, some people are just crazy12:15
CIA-303arigo 07async-del * r56086 10pypy/pypy/interpreter/executioncontext.py: Duh.12:24
cfbolzjewel_: it's actually useful, no?12:25
jewel_more than a vmware-type product?12:26
jewel_and to implement it in java?12:26
cfbolzjewel_: easier to understand than obscure C, usually12:28
jewel_perhaps, but being able to produce native code directly would be a big win12:31
cfbolzjewel_: hotspot does that for you :-)12:32
jewel_i willing to bet that there is some pretty obscure java code in there12:32
jewel_yes, but hotspot has a poor interface as a native code generator12:33
cfbolzjewel_: I don't think they are using that12:33
jewel_i mean, it has no interface12:33
cfbolzprobably there is obscure java, but there performance isn't too shabby12:33
nshepperd (n=neil@CPE-124-180-205-31.vic.bigpond.net.au) left #pypy.12:35
jewel_i still think it's crazy, think about all the hardware you have to emulate12:39
jewel_and what it takes to run an OS like windows or a graphical linux12:39
ghaering (n=gerhard@ left irc: "Leaving"13:10
cfbolzjewel_: more crazy than a gameboy interpreter in python? :-)13:10
antocuni_ (n=antocuni@host252-123-dynamic.6-79-r.retail.telecomitalia.it) joined #pypy.13:18
antocuni (n=antocuni@host101-126-dynamic.14-87-r.retail.telecomitalia.it) left irc: Read error: 110 (Connection timed out)13:26
Nick change: antocuni_ -> antocuni13:26
mgod (n=mgod@adsl-68-73-206-193.dsl.sfldmi.ameritech.net) joined #pypy.13:35
mgod (n=mgod@adsl-68-73-206-193.dsl.sfldmi.ameritech.net) left irc: Client Quit13:39
bgola_ (n=Bruno@ left irc: "leaving"13:46
mgod (n=mgod@adsl-68-73-206-193.dsl.sfldmi.ameritech.net) joined #pypy.13:51
antocunicfbolz: what's the best way to make sure that the rainbow interp has the opcode "red_same_as" built in?13:57
cfbolzno clue :-)13:59
antocuniok :-)13:59
mgod (n=mgod@adsl-68-73-206-193.dsl.sfldmi.ameritech.net) left irc: 14:15
djcwow, good news about the funding14:29
mgod (n=mgod@fw.arb.zattoo.com) joined #pypy.14:38
cfbolz (n=cfbolz@p4FDAAB66.dip0.t-ipconnect.de) left irc: Read error: 113 (No route to host)14:39
joe_mp (n=joe@p549C47E2.dip.t-dialin.net) joined #pypy.14:43
radixhooray google money14:48
radixIt's interesting that you guys put your funding money into the travel for sprints14:48
radixthe Twisted project has decided to avoid doing that because it seems we can get a lot more value out of funding one person to work more consistently14:49
radixlike, depending on where you're flying from, a couple of flights for a few days can easily add up to a month of work for one person14:49
radixthe cost of, rather14:50
radixalthough I guess if most of the funding is intra-european it's not so expensive14:50
xorAxAxyeah, this is not like ibm in the 80's :)14:51
DanielHolth (n=chatzill@ joined #pypy.14:55
headius (n=headius@ joined #pypy.15:10
elmom (n=elmom@hoasnet-fe10dd00-97.dhcp.inet.fi) left irc: Read error: 54 (Connection reset by peer)15:11
headius (n=headius@ left irc: 15:33
bigdo1congratulations on the funding,  both the money and support from google are a very for pypy.  15:35
bigdo1the google spotlight may add some other oppertunites for pypy15:35
headius (n=headius@ joined #pypy.15:46
jacob22radix: We are not looking at the exact lines of code that a piece of funding gives us, we are more concerned with what keeps the spirit of the project alive and what brings us forward in the long term.15:56
jacob22radix: PyPy still doesn't have users which means that keeping the vision alive and keeping pople working towards a specific goal is a major concern. For this, it is important for the members of the project to actually meet, discuss PyPy and go skiing and drinking beer together.15:58
exarkunyaaay skiing beer15:59
exarkunI mean vision15:59
jacob22The origins of PyPy are actually from a beer session after the first EuroPython.16:00
DanielHolthThe effects of beer on vision are well documented.16:00
djcFor the effects of vision on beer, much less so.16:00
djcAlthough PyPy seems to do well in that department.16:00
jacob22In any case, the Google funding mostly goes to a very few individuals, and we want to support the work of more people. Sprint funding is a good way to do so.16:02
jacob22And travelling by budget airlines in Europe is remarkably cheap. You have to put up with being treated as cattle by some airlines, but the ticket often costs less than going to the cinema.16:05
fzzzy (n=donovan@c-71-198-81-84.hsd1.ca.comcast.net) joined #pypy.16:07
fdr- (n=fdr@76-191-216-242.dsl.dynamic.sonic.net) left irc: Remote closed the connection16:13
fzzzy (n=donovan@c-71-198-81-84.hsd1.ca.comcast.net) left irc: 16:25
DanielHolthjacob22, How does the cinema treat you?16:25
quirim (n=potato@dsl-242-35-80.telkomadsl.co.za) joined #pypy.16:30
panni_ (i=hannes@ip-88-152-22-213.hsi.ish.de) joined #pypy.16:35
mgod_ (n=mgod@adsl-71-148-47-118.dsl.sfldmi.sbcglobal.net) joined #pypy.16:40
mgod__ (n=mgod@fw.arb.zattoo.com) joined #pypy.16:41
mgod (n=mgod@fw.arb.zattoo.com) left irc: Read error: 110 (Connection timed out)16:55
mgod_ (n=mgod@adsl-71-148-47-118.dsl.sfldmi.sbcglobal.net) left irc: Read error: 110 (Connection timed out)16:57
joe_mp_ (n=joe@p549C7D89.dip.t-dialin.net) joined #pypy.17:07
cfbolz (n=cfbolz@p4FDA9225.dip0.t-ipconnect.de) joined #pypy.17:09
bea (n=bea@c83-248-87-204.bredband.comhem.se) joined #pypy.17:09
dialtone (n=dialtone@adsl-75-6-226-104.dsl.pltn13.sbcglobal.net) left irc: Read error: 110 (Connection timed out)17:10
cfbolz_ (n=cfbolz@p4FDA9255.dip0.t-ipconnect.de) joined #pypy.17:11
CIA-303arigo 07async-del * r56088 10pypy/pypy/ (6 files in 4 dirs): (log message trimmed)17:13
CIA-3Intermediate check-in. Refactoring the "periodic actions" approach.17:13
CIA-3This should allow signals and app-level destructors to be handled17:13
CIA-3promptly, i.e. before the next opcode is run (so far they have to wait17:13
CIA-3for up to sys.checkinterval opcodes).17:13
CIA-3The 2nd goal is to decrease the overhead in the fast path of17:13
CIA-3executioncontext.bytecode_trace() by making the checks here simpler. It17:13
bigdo1 (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) left irc: Read error: 110 (Connection timed out)17:18
cfbolz__ (n=cfbolz@p4FDA92A9.dip0.t-ipconnect.de) joined #pypy.17:22
fzzzy (n=donovan@office.mochimedia.net) joined #pypy.17:22
bea (n=bea@c83-248-87-204.bredband.comhem.se) left irc: Remote closed the connection17:23
joe_mp__ (n=joe@p549C5E14.dip.t-dialin.net) joined #pypy.17:24
joe_mp (n=joe@p549C47E2.dip.t-dialin.net) left irc: Read error: 110 (Connection timed out)17:27
mgod (n=mgod@fw.arb.zattoo.com) joined #pypy.17:28
cfbolz (n=cfbolz@p4FDA9225.dip0.t-ipconnect.de) left irc: Nick collision from services.17:28
Nick change: cfbolz__ -> cfbolz17:28
mgod__ (n=mgod@fw.arb.zattoo.com) left irc: Read error: 110 (Connection timed out)17:34
cfbolz_ (n=cfbolz@p4FDA9255.dip0.t-ipconnect.de) left irc: Read error: 110 (Connection timed out)17:35
joe_mp_ (n=joe@p549C7D89.dip.t-dialin.net) left irc: Read error: 110 (Connection timed out)17:41
fijal (n=fijal@c-76-25-206-12.hsd1.co.comcast.net) joined #pypy.17:41
Action: pedronis -> home17:45
pedronissee you17:45
dialtone (n=dialtone@adsl-99-136-101-166.dsl.pltn13.sbcglobal.net) joined #pypy.17:45
pedronis (i=pedronis@vitaly.openend.se) left irc: Remote closed the connection17:45
jewel_ (n=jewel@dsl-242-163-111.telkomadsl.co.za) left irc: Connection timed out17:57
devinus (n=devin@ joined #pypy.17:58
jewel_ (n=jewel@dsl-242-165-12.telkomadsl.co.za) joined #pypy.17:58
Rhamphoryncus (n=rhamph@unaffiliated/rhamphoryncus) joined #pypy.17:59
CIA-303fijal * r56089 10extradoc/sprintinfo/post-ep2008/people.txt: my dates18:04
AntonK (n=asl@asl.dorms.spbu.ru) joined #pypy.18:04
cfbolzAntonK: hi!18:04
AntonKevening all18:05
bigdo1 (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) joined #pypy.18:08
pjenvey_ (n=pjenvey@astound-69-42-4-166.ca.astound.net) joined #pypy.18:13
fijalour select implementation on top of poll is nontrivially broken18:15
xorAxAxwasnt there an url to syndicate also the comments?18:24
harobed (n=stephane@pda57-1-82-231-115-1.fbx.proxad.net) left irc: "Ex-Chat"18:24
cfbolzfijal: also here: http://www.reddit.com/info/6p2kz/comments/18:25
Action: fijal has cheaper cinemas than airline tickets in poland18:28
steinrs (i=50cab54c@gateway/web/ajax/mibbit.com/x-839e3e1abf266538) joined #pypy.18:29
steinrshi there... just a quick question: are you guys happy with LLVM? what kinda of issues have you run into? any glaring bugs?18:29
fijalsteinrs: yes, almost all releases had bugs18:29
fijaland we need to stay with HEAD18:29
fijalwe don't use it actively these days though18:30
steinrsthought about using it for a compiler project18:30
fijalnote that our use case is fairly extreme18:31
fijalanother issue was dependency of llvm-gcc which is pain to build18:31
fijalit'll hopefully change with clang18:31
steinrsis that far along?18:31
fijalno idea18:31
fijalnever checked18:31
fijalAntonK: do you know details18:31
fijalsteinrs: we're still able to compile to llvm18:32
fijaljust our main version is C18:32
steinrsi see18:32
AntonKcurrent status of clang is something like this:18:32
AntonKC parsing (incl. prepocessor) - almosttly done18:33
AntonKsema checks - maybe 60-70%18:33
AntonKcodegen - maybe 40%, but some big applications were reported to work18:33
AntonK(afair, even cpython itself)18:33
fijalAntonK: you generate llvm out of it? right?18:34
steinrsok, i was thinking about just targeting llvm byte code directly18:34
cfbolzfijal: which might change quickly if their backend becomes faster than GCCs :-)18:34
steinrsshouldn't need any gcc/clang dependency18:34
AntonKfijal: yes18:34
fijalcfbolz: right18:34
fijalsteinrs: yes, we're not heavily C-based. C has known issues and if llvm becomes faster we can start using mainly this18:35
grgi (n=grgi@f054162143.adsl.alicedsl.de) joined #pypy.18:36
grgi (n=grgi@f054162143.adsl.alicedsl.de) left #pypy ("Ex-Chat").18:37
steinrsi'm curious - what sets pypy appart from the stock python distribution i'm using?18:38
steinrsis the focus on improving speed?18:38
cfbolzthe goals are speed, features, availability18:39
fijalwhat was the reasoning behind implementing select using poll?18:40
cfbolzfijal: that's a common choice, no?18:40
fijalcfbolz: no18:41
fijalcpython doesn't do that18:41
fijaland that makes no sense since kernel does this anyway18:41
fijalthe main problem is that it's buggy18:41
cfbolzfijal: seems extremely old code18:42
fijalis there any reason why it's still like that, or just legacy?18:43
cfbolzprobably because nobody fixed it18:43
fijalcause I fear it would be actually faster to reimplement it using select18:43
fijalthan to fix it18:43
cfbolzfine, I would guess18:44
fijalit would be faster as a side effect18:46
fijalsecond thing18:46
fijalI'm about to implement gc.get_referrers18:46
fijaland the question is18:47
fijalwhen I walk all objects in gc18:47
fijalsorry addresses18:47
fijalhow can I cast it to a pointer to instance?18:47
fijalso I can pass it to app-level?18:47
antocunifijal: cast_adr_to_ptr?18:48
antocunior something like this18:48
fijalbut I need ptr type18:50
josch (n=josch@host140.natpool.mwn.de) joined #pypy.18:50
joschOMG YOU GUYS RULE!! :D18:51
joschi just heard the chaos radio express episode about pypy18:51
joschand I'm deeply impressed18:51
fijalantocuni: what ptr type would be for app-level objects?18:51
antocuniuhm... no clue18:52
jacob22fijal: I think the implementation of select was done as a stopgap measure, to have any select functionality at all.18:52
fijaljacob22: I simply don't understand why it's not using select18:54
fijalprobably because having any external call was a pain these days18:54
jacob22fijal: Yes, I have a vague memory of someone not knowing how to access select.18:55
mgod (n=mgod@fw.arb.zattoo.com) left irc: 18:56
mgod (n=mgod@fw.arb.zattoo.com) joined #pypy.18:56
antocunipypy code swarm: http://it.youtube.com/watch?v=b0cSm1DgCbc18:57
xorAxAxAntonK: clang can do c++?18:59
xorAxAxfijal: i dont think that llvm-gcc is seriously needed anymore18:59
xorAxAxgiven the current infrastructure18:59
xorAxAxsomebody has just to finish the .h cleanup :)18:59
pedronis (n=user@c-91d570d5.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.19:02
josch (n=josch@host140.natpool.mwn.de) left #pypy ("Leaving").19:02
fijalxorAxAx: no, it is19:03
fijalhow else do you compile all accessors?19:03
hpkantocuni: what is this? 19:03
fijalantocuni: you can see that we're sprint based :)19:03
fijalhpk: it's a meme you avoid because you don't read news 6 hours a day19:03
xorAxAxfijal: gcc19:03
xorAxAxfijal: like its done currently19:03
antocunihpk: a visualization of all commits to pypy repo from the beginning until today19:04
xorAxAxi implemented that code :-P19:04
antocunifijal: it's also nice to see the explosion when the EU funding starts19:04
xorAxAxantocuni: cool, does it have nice music? :)19:04
antocunixorAxAx: sure, I added it :-)19:04
xorAxAxantocuni: its not sharp :-(19:04
AntonKxorAxAx: not yet19:05
xorAxAxantocuni: probably vimeo et al. allow for higher resolutions19:05
AntonKc++ is really hard and ugly19:05
xorAxAxAntonK: yeah19:05
AntonKit will require 2-3 years more19:05
antocunixorAxAx: vimeo makes firefox crashing every time I try to upload it19:05
xorAxAxAntonK: because it sounded like somebody asked you whether clang can be used to compile llvm19:05
xorAxAxantocuni: :-(19:05
fijalantocuni: firefox sucks19:05
antocunihere is the original high quality movie: http://wyvern.cs.uni-duesseldorf.de/~antocuni/pypy.avi19:05
cfbolzantocuni: very cool!19:08
xorAxAxis visible19:09
hpkwesdorp though, i guess 19:10
antocunihpk: I don't think wesdorp have a nike like "gvanrossum" :-)19:10
hpkguido v. rossum is gvanrossum on codespeak19:10
hpkoh, ok - i thought xorAxAx meant this literally 19:10
xorAxAxwow, this non-sprint-phase-inactivity is astonishing :)19:11
antocuniunfortunately after a bit things get too confused due to high activity19:11
bigdo1 (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) left irc: Connection timed out19:11
antocuniit's possible to have a higher granularity, but then the video become too long19:12
xorAxAxyeah, some cross artefacts19:12
Nick change: wristtone -> eartone19:12
xorAxAxi guess those are fixeols :)19:12
antocunior merging of branches19:12
xorAxAxhmm, too many codec artefacts IMHO19:14
fijalantocuni: is this trunk only or branches?19:14
antocunifijal: it's dist/ and branch/19:14
fijalcpython's select is #def messs19:14
antocunicould someone try to upload it on vimeo please?19:15
DanielHolthwould it make any sense to compile CPython and run it with lli?19:15
fijalDanielHolth: what is lli?19:15
DanielHolthllvm's interpreter/jit19:15
xorAxAxantocuni: is it probably possible to regenerate it with a nicer codec?19:15
xorAxAxantocuni: that has less artefacts19:15
xorAxAx(and probably more data)19:15
fijalDanielHolth: try19:16
antocunixorAxAx: artefacts?19:16
cfbolzxorAxAx: yes, especially at the end it seems to get quite bad19:16
DanielHolthworth a shot :-)19:16
xorAxAxantocuni: well, its hard to read the nick names19:16
xorAxAxantocuni: starting from 50% on19:16
xorAxAxvery blurred/blocked19:16
antocunixorAxAx: are you talking about the video on wyvern?19:16
antocuniah, right19:17
exarkunis wyvern still a super fast somethingsomething?19:17
xorAxAxyes, that one19:17
cfbolzDanielHolth: what would be use case be?19:17
antocuniok, I've also a higher quality version which is 40 MB19:17
Action: antocuni uploads19:17
fijalnow the cool question is why select module tests hang on my machine19:17
fijalexarkun: superfast?19:18
DanielHolthPerhaps llvm would be able to apply different optimizations.19:18
exarkunfijal: yeasuperfast!19:18
cfbolzexarkun: fast enough19:18
exarkunhave you guys seen those videos19:19
exarkunabout commit history19:19
fijalexarkun: use bigdog's machine, it's super fast19:19
cfbolzexarkun: isn't that what anto pasted above?19:19
exarkunfijal: oh, good point.19:19
antocuniyes, it is19:19
exarkunOh.  I missed that, just got back from lunch. :)19:20
fijalexarkun: I'll reimplement select using select, I have enough19:20
exarkunfijal: have enough what?19:20
antocuniexarkun: very timely though :-)19:20
fijaltrying to track what's broken19:20
pedronisfijal: I think it was implemented using poll, because of the helper macros that are needed to use select19:21
pedronisFD_SET etc are typically macros19:21
fijalpedronis: that doesn't bother us any more19:22
pedroniswell you asked why, that's the answer19:23
fijalgood to know19:23
fijalmy question is "are there any still valid reasons"19:23
mgod_ (n=mgod@fw.arb.zattoo.com) joined #pypy.19:29
jewel_ (n=jewel@dsl-242-165-12.telkomadsl.co.za) left irc: Read error: 113 (No route to host)19:33
mgod (n=mgod@fw.arb.zattoo.com) left irc: Read error: 110 (Connection timed out)19:35
antocunihigher quality than before19:38
DanielHolthWhat generates those statistics?19:39
DanielHolthWhat program makes the animation from the statistics?19:39
antocuniDanielHolth: http://vis.cs.ucdavis.edu/~ogawa/codeswarm/19:40
xorAxAxumm, village people?19:42
xorAxAxantocuni: does it fit to pypy's commit, do you think? :)19:43
xorAxAxits playing village people - YMCA at the end19:44
xorAxAxafter the dire straits19:44
antocuniyes, I know :-). But why do you think fits?19:45
xorAxAxi want to know why it would fit :)19:46
xorAxAx"Taking the song at face value, its lyrics extol the virtues of the Young Men's Christian Association."19:46
antocuniwell, I choose the songs mostly randomly19:46
Action: antocuni afk19:46
bigdo1 (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) joined #pypy.19:58
carrus85 (n=carrus85@ joined #pypy.20:02
CIA-303arigo 07async-del * r56090 10pypy/pypy/ (6 files in 3 dirs): Still in-progress. Added some tests.20:23
CIA-303arigo 07async-del * r56091 10pypy/pypy/interpreter/executioncontext.py: Minimal fix for test_pyframe.20:28
CIA-303arigo 07async-del * r56092 10pypy/pypy/interpreter/pyframe.py: Obscure cases, untested :-/20:30
CIA-303arigo 07async-del * r56093 10pypy/pypy/interpreter/test/test_executioncontext.py: Fix test.20:30
CIA-303arigo 07async-del * r56094 10pypy/pypy/module/gc/interp_gc.py: This line is no longer necessary.20:31
s4msung (n=s4msung@p4FDDAFC2.dip0.t-ipconnect.de) joined #pypy.20:34
radixantocuni: your pdb isn't compatible with the 'stdout' parameter to Pdb.__init__20:41
Action: antocuni looks20:42
CIA-303arigo 07async-del * r56095 10pypy/pypy/interpreter/executioncontext.py: Another potential fix.20:42
antocuniradix: my stdlib's pdb doesn't have a stdout parameter20:43
antocunihas it been added in py2.5?20:43
radixclass Pdb(bdb.Bdb, cmd.Cmd):20:44
radix    def __init__(self, completekey='tab', stdin=None, stdout=None):20:44
CIA-303arigo 07async-del * r56096 10pypy/pypy/interpreter/executioncontext.py: Small optimizations for the case where we don't use tracing.20:45
fijal>               AttributeError: 'ClassDef' object has no attribute 'minid'20:46
fijalwhat that might be???20:46
xorAxAxfijal: class seen too late?20:47
fijalthat's just obscure way of masking AttributeError20:47
fijalyes, it is20:47
fijalClassDef is exceptions.AttributeError20:48
antocuniradix: try now20:49
radixantocuni: thanks20:50
radixfwiw, the reason I needed that is that doctest tries to use it, apparently20:52
antocuniradix: good luck, the interactions between pdb and doctests are far from clear to me20:53
radixantocuni: well, it works now :)20:53
antocunigood :-)20:53
Action: antocuni afk20:53
radixI didn't actually use the pdb support, but doctest instantiates a Pdb object Just In Case I guess20:53
CarlosEDP (n=Carlos@ joined #pypy.21:07
CarlosEDPcongratulations guys for the Google funding... pypy really deserves it...21:07
CarlosEDP_ (n=Carlos@ joined #pypy.21:13
fijalhow can I get 'RuntimeError during flow graph construction'?21:22
fzzzyfijal: what nevow test was the one that failed?21:23
fijalfzzzy: the one that runs gc.get_referrers21:23
fijalit's remove from nevow now21:23
fijal(it was not testing anything)21:23
fzzzyI'm just wondering if I wrote it and why :)21:23
fijalI don't know21:24
fijalit contained pdb or sth (commented out)21:24
fijaland print21:24
fijalbut no clue what it should be21:24
exarkunfzzzy: I think mg wrote it21:24
fzzzyah, whew :)21:24
fzzzythanks for helping get that working exarkun, that's totally awesome21:25
fijalexarkun: yeah, thx from me as well21:25
CIA-303fijal * r56098 10pypy/pypy/rlib/ (_rsocket_rffi.py rpoll.py test/test_rpoll.py): Rpython level implementation of select with a single test21:27
grubber (n=grubber@ joined #pypy.21:31
CarlosEDP (n=Carlos@ left irc: Success21:33
fiddlerwoaroof (i=80f960fd@gateway/web/ajax/mibbit.com/x-6d6d0281433e72c1) joined #pypy.21:34
fiddlerwoarooffijal: how can pypy help me?21:35
AntonK (n=asl@asl.dorms.spbu.ru) left irc: Read error: 110 (Connection timed out)21:36
exarkunit has transparent proxies and tasklets21:36
fijalfiddlerwoaroof: I think it'll be more trouble for you than usage21:38
fijalbut check this out21:38
grubbercfbolz: you around?21:45
cfbolzgrubber: sort of21:45
cfbolzwhat's up?21:45
grubbernot much...wollte mal hallo sagen :)21:46
grubberbeen busy with work, and ahvent been able to do much on the parsing stuff21:46
cfbolzthat's fine, of course :-)21:46
jewel_ (n=jewel@dsl-242-165-12.telkomadsl.co.za) joined #pypy.21:46
cfbolzgrubber: note that bruno, our soc student, might do some parser stuff21:47
grubbercfbolz: good to know... i have some free time in the next couple days... anything i can help out on?21:47
tav (n=tav@ left irc: Connection reset by peer21:48
steinrs (i=50cab54c@gateway/web/ajax/mibbit.com/x-839e3e1abf266538) left #pypy.21:50
CIA-303fijal * r56099 10pypy/pypy/rpython/lltypesystem/rffi.py: Cache results and don't invoke gcc over and over again21:51
cfbolzgrubber: no clue, he is ill right now21:52
CIA-303fijal * r56100 10pypy/pypy/module/select/ (__init__.py app_select.py interp_select.py): 21:53
CIA-3Implement select using select and not poll. This is somewhat faster and21:53
CIA-3doesn't hang tests, which is a win already.21:53
fijalexarkun: done, let's wait an hour for compilation and see21:54
fijalexarkun: does twisted work on 64bit?21:54
fijalI can try compilation on bigdog's machine21:54
grubbercfbolz: any other tasks that wouldnt be too hard to get me going on? i dont want to interfere with what he's been doing21:54
fijalgrubber: are you dead-set on parsing?21:54
grubberfijal: no... just sounded like fun, and it was fun to play with... im willing to help out any way i can (when i can that is)21:55
fijalgrubber: running software on top of pypy is good21:55
fijalmight be tedious in place though21:55
grubberfijal: what do ya mean?21:56
fijallike pick your favorite pypy program/web framework/whatever21:56
fijaland try to run it on top of pypy21:56
fijalother task which I was thinking about lately - write memory profiler, would be fairly simple21:57
cfbolzfijal: weeeell21:57
CIA-303fijal * r56101 10pypy/pypy/module/select/interp_select.py: typo21:58
fijaldepends on the level of complication you want :)21:58
grubberfijal: a mem profiler to profile pypy or a program running on pypy? 21:58
cfbolzgrubber: I don't know, maybe you could look into integrating the regex-based tokenizer into the parser without having to rewrite the parser21:58
cfbolzI still think that looking into parsing is very worthwhile21:58
fijalcfbolz: that's true21:59
fijalgrubber: program running on pypy21:59
fijalbigdo1: what is your twistd doing consuming all CPU? (just curious)22:00
grubbercfbolz: well, the rlib.parsing is the part of pypy i know best22:01
grubberis pypy/interpreter/pyparser the location of the current parser for python?22:01
fijalgrubber: yop. this is ugly mess22:02
cfbolzgrubber: yes22:03
cfbolzgrubber: don't worry. nobody knows this bit really :-)22:03
grubbercfbolz: wow... it is interesting!22:04
grubberso you'd like to replace the tokenizer/lexer with a regex version?22:04
cfbolzgrubber: I wrote a regex-based version of the lexer22:05
cfbolzbut it is not integrated with the rest22:05
grubbercfbolz: the one we played with a while back, right?22:05
pjenvey_ (n=pjenvey@astound-69-42-4-166.ca.astound.net) left irc: 22:05
fiddlerwoaroofwhat do the colors mean in the pypy to c translation?22:06
cfbolzfiddlerwoaroof: mandelbrot fractal22:06
fiddlerwoaroofI see22:07
fiddlerwoaroofIt looked familiar22:07
cfbolzexarkun: ?22:07
exarkuncfbolz: translate.py should also find the coordinates of the resulting pypy-c in the mandelbrot set and report them22:08
grubbercfbolz: i dont know if you looked at the last stuff i had done22:08
exarkunor interpret the resulting pypy-c as coordinates and render that part of the mandelbrot set22:08
grubbermy goal was to create a class that could compare the output of that lexer with the output of the CPython tokenize modules22:08
exarkunthat might actually be possible :)22:08
grubberuse it to help test correctness of the output22:09
cfbolzexarkun: most of them might be boring22:09
cfbolzgrubber: ah, I remember vaguely22:09
grubbercfbolz: do you think that would be useful? or would you rather i try something else?22:09
cfbolzgrubber: no, sounds good22:10
cfbolzgrubber: and then fix the bugs, of course :-)22:10
grubbercfbolz: haha. of course. was heisst "bug" auf de?22:11
cfbolzor "fehler"22:11
grubberhaha. boring :)22:11
c0dewarrior (n=c0dewarr@dsl-242-234-161.telkomadsl.co.za) left irc: "...."22:14
tav (n=tav@ joined #pypy.22:15
fijalfiddlerwoaroof: what options did you use for translation?22:19
grubberfijal or cfbolz: when i was trying to get newline endings to work, i found inconsistency in cpython vs the stdlib tokenize module22:19
grubberi posted a bug, but GvR denied it (it is a bug, IMO)22:20
grubberbut he suggested that we could open files using new-style line endings22:20
grubberthen we dont have to handle \n vs \r vs \r\n22:20
CIA-303fijal * r56102 10pypy/pypy/module/select/interp_select.py: Fix translation22:20
grubberis that doable?22:20
xorAxAxgrubber: i guess you need to add U at the right spot. good luck. :)22:21
cfbolzdidn't armin recently fix that on trunk?22:21
fijalfiddlerwoaroof: if you want to use it, you need to pass few options to the translation22:21
fijal(like --allworkingmoduleS)22:21
fijalcfbolz: no, he said it might be trivial like ...22:21
cfbolzgrubber: yes, so you might have to hunt where the file is actually opened :-(22:22
fijalwe need to update default options...22:23
grubbercfbolz: i can do that... didnt know if that would be "off limits" because of how pypy worked22:23
cfbolzgrubber: it might be on interp-level. in which case we need to do it a bit differently than just adding a U22:24
fiddlerwoaroof (i=80f960fd@gateway/web/ajax/mibbit.com/x-6d6d0281433e72c1) left irc: "http://www.mibbit.com ajax IRC Client"22:26
grubberfijal: has anyone tried it?22:26
cfbolzdon't think so22:27
cfbolzgrubber: write a test, fix the problem and be done :-)22:27
grubberhaha, ok, ill try it out :)22:27
cfbolzgrubber: cool that you work on stuff22:29
fiddlerwoaroof (i=80f960fd@gateway/web/ajax/mibbit.com/x-224a317d7065e428) joined #pypy.22:29
grubbercfbolz: im happy to help out :)22:30
tav_ (n=tav@ joined #pypy.22:30
fzzzy (n=donovan@office.mochimedia.net) left irc: Read error: 113 (No route to host)22:30
radixok guys22:31
radixtime for pypy to win, now that guido says that quality control is a bad idea22:31
cfbolzhe does?22:33
grubberradix: where did you read taht?22:33
radixit's hearsay, of course :)22:33
cfbolzyay, flamewar22:33
exarkunit's on python-dev22:33
fzzzy (n=donovan@office.mochimedia.net) joined #pypy.22:34
exarkun"Heck, we don't even roll back if one of our own buildbots22:34
cfbolzradix: we are working on it22:34
exarkungoes red."22:34
exarkunis one particularly nice quote22:34
xorAxAxred is just another shade of green anyway22:35
radixxorAxAx: exarkun is red, you're green22:35
radix(in my IRC client)22:35
cfbolzin mine too22:37
cfbolzradix: you are blue-greenish22:37
xorAxAxyou are all grey, how boring :)22:37
antocuniradix, cfbolz: seems that we are all using the same IRC client :-)22:38
xorAxAxthere is an irssi script to adapt the xchat hash algo22:38
cfbolzdo we know for sure that GvR isn't color-blind?22:39
xorAxAxthe buildbot community hasnt solved that accessiblity issue, has it?22:39
carrus85exarkun: holy crap, I've had that thing sitting in my inbox and didn't even notice.  (regarding the flamewar).  Thanks for the heads up!22:40
exarkunxorAxAx: pssh.22:40
exarkunxorAxAx: That's what css is for.  Go read the CSS level 3 specification.22:41
xorAxAxexarkun: zsshh pssh tss22:41
exarkunIf there's not actually a browser that supports any of those features, that's probably Mozilla's fault, not buildbot's.22:41
exarkuncarrus85: enjoy :p22:41
xorAxAxexarkun: css level 3 can remap colours based on the features of the user? :)22:41
exarkunxorAxAx: no, way better than that22:42
xorAxAxbut you got a point. we are in irc where social problems are always solved technically :)22:42
exarkunxorAxAx: it can let you provide information in a way which allows it to show up as red/green for people with a "normal" configuration and in a different way for other people, such as those who are R/G colorblind22:42
exarkunxorAxAx: or give hints to a screenreader22:42
exarkunxorAxAx: and so on22:42
exarkunwell, all problems have technical solutions, of course.22:43
cfbolzof course22:43
exarkunin the absolute worst case, you just make the sun go nova.22:43
xorAxAxexarkun: yeah. but probably one wants a default config that works on most browsers for nearly everybody? :)22:43
tav (n=tav@ left irc: Connection timed out22:44
xorAxAxghosts III - 3422:44
Action: xorAxAx recomposed the swarm movie with NIN - ghosts III - 34. it will be available soon on vimeo22:45
exarkunxorAxAx: nah22:46
xorAxAxapproximately 25 min. left until the video will have been processed on their rendering machines22:47
exarkunxorAxAx: hm, they are offering a rendering service now?22:48
xorAxAxlets call it post processing :)22:49
exarkunxorAxAx: url?22:50
xorAxAxexarkun: will publish it when its ready :)22:51
CarlosEDP__ (n=Carlos@ joined #pypy.22:51
Nick change: CarlosEDP__ -> CarlosEDP22:51
exarkunyou confused me22:51
exarkunwhat did you mean "will have been processed on their rendering machines"?22:52
xorAxAxthey are transcoding it currently22:52
xorAxAxwasnt the future perfect habitual correct?22:53
fijalexarkun: nice flamewar btw22:54
exarkunxorAxAx: I wanted the URL to their transcoding service :)22:55
xorAxAxexarkun: ah. thats internal22:56
xorAxAxyou need to upload a movie22:56
exarkunokay I'm just gonna give up22:57
mwhudson (n=mwh@203-114-189-104.cafenet.dyn.inspire.net.nz) joined #pypy.22:58
cfbolzexarkun: you are not exactly it very clear what your question is :-)22:58
cfbolzmwhudson: hi!22:58
CarlosEDP_ (n=Carlos@ left irc: Read error: 54 (Connection reset by peer)22:59
Action: mwhudson is having a day full of the frustrations of modern life23:00
fijalmwhudson: like?23:01
devinus (n=devin@ left irc: 23:01
jacob22mwhudson: Welcome to the century of the fruitbat.23:01
xorAxAxfilling out 27b-6 forms?23:01
mwhudsonwell, mainly having internet disconnected at the old place but not yet connected at the new place23:01
hpk (n=chatzill@p549F608C.dip.t-dialin.net) left irc: Read error: 60 (Operation timed out)23:01
mwhudsonxorAxAx: oh yes, and having both my isp and the tax department not being very keen on answering the phone23:02
Action: pedronis goes to bed23:02
pedronissee you, gn23:02
xorAxAxprobably you need to contact their information retrieval dept. :-)23:02
pjenvey_ (n=pjenvey@ joined #pypy.23:03
mwhudsonnah, it's not really like that, they're actually pretty friendly when you get through23:04
pedronis (n=user@c-91d570d5.022-54-67626719.cust.bredbandsbolaget.se) left irc: "ERC Version 5.2 (IRC client for Emacs)"23:04
cfbolzmwhudson: seen the blog post?23:07
cfbolzhttp://google-opensource.blogspot.com/2008/06/pypy-progress-python-fast-and-flexible.html (also on the pypy blog)23:12
CarlosEDP (n=Carlos@ left irc: "ChatZilla 0.9.83 [Firefox 3.0/2008052906]"23:12
mgod_ (n=mgod@fw.arb.zattoo.com) left irc: 23:13
grubber (n=grubber@ left irc: 23:17
fiddlerwoaroofdoes "[translation] start debugger..." mean something failed?23:20
xorAxAxfiddlerwoaroof: no23:20
xorAxAxjust for convenience23:20
fiddlerwoaroofdoes it mean I'm done?23:20
xorAxAxyes, if there was no error before that line23:20
tav_ (n=tav@ left irc: Connection reset by peer23:23
tav (n=tav@ joined #pypy.23:23
fiddlerwoaroofpypy can't parse mdir1 here: http://pastebin.com/m7c32092423:23
cfbolzfiddlerwoaroof: pypy doesn't support 2.5 features yet23:24
cfbolz(like the ternary operator)23:24
fiddlerwoaroofI see23:24
fiddlerwoaroofthat was such a cool one-liner23:24
cfbolzfiddlerwoaroof: we have a soc project for 2.5 compat23:27
fiddlerwoaroofhow is it progressing?23:27
xorAxAxthey tuned the language a lot by writing a better gc23:29
xorAxAxdifferently said, they tuned the gc for ast interpretation23:29
xorAxAxah, they just made gc cow friendly23:31
xorAxAxit seems like ruby systems need a lot of forks23:31
s4msung (n=s4msung@p4FDDAFC2.dip0.t-ipconnect.de) left irc: "Verlassend"23:33
fijalxorAxAx: just-another-ruby-fork?23:39
fijalany link?23:40
fijalthere was a lot of talk about ruby not addressing problems23:40
xorAxAxsee 5 lines above23:40
fijalsecurity issues23:41
xorAxAxyeah, ruby is the new php23:42
xorAxAxthere it is23:42
fijal"It is observed that a lot of memory in Ruby on Rails applications is taken by AST nodes"23:43
fijalonly as long as you have mostly code23:43
fijal33% memory reduction and faster? Is this for real?23:43
fijalsounds like american supermarket to me23:43
xorAxAx... the codeswarm video23:44
jewel_ (n=jewel@dsl-242-165-12.telkomadsl.co.za) left irc: Read error: 113 (No route to host)23:48
fiddlerwoaroofhow mature is pyp?23:48
cfbolzfiddlerwoaroof: depends on your definiton23:49
cfbolznobody is using it in production yet23:50
fiddlerwoaroofis it usable in a production environment?23:50
cfbolzprobably. hard to know without trying23:51
fiddlerwoaroofwill it support most python2.4 code?23:52
cfbolzit should support everything except CPython extension-modules23:53
cfbolzsome of the more important one have been re-implemented though23:53
fiddlerwoaroofcfbolz: twisted is the most important for me23:53
fiddlerwoarooftwisted.web to be most specific23:54
fijaltwsited.web works as far as I remember23:54
fijalthere are some glitches with select reactor23:55
fijalwhich I'm working on right now23:55
fijalbut poll reactor should work23:55
fijalno pyopenssl though23:55
fiddlerwoaroofdoes pypy support anything like LISP macros?23:55
--- Fri Jun 27 200800:00

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