[00:55] <stakkars> hmm. didn't go to the party.Going to bed - tired
----- silence for 44 minutes ----- [01:39] xyz359 (n=xyz359@pool-68-238-168-248.tampfl.fios.verizon.net) left irc: "http://www.purepwnage.com/";
----- silence for 4 hr and 30 minutes ----- [06:09] xyz359 (n=xyz359@pool-68-238-168-248.tampfl.fios.verizon.net) joined #pypy.
----- silence for 36 minutes ----- [06:45] xyz359 (n=xyz359@pool-68-238-168-248.tampfl.fios.verizon.net) left irc: "http://www.purepwnage.com/";
----- silence for 2 hr and 2 minutes ----- [08:47] Shoragan (n=shoragan@d072.apm.etc.tu-bs.de) joined #pypy.
----- silence for 42 minutes ----- [09:29] mfp (n=batsman@186.Red-83-57-201.dynamicIP.rima-tde.net) left irc: Read error: 110 (Connection timed out)
----- silence for 19 minutes ----- [09:48] Action: lac is back
[09:59] mfp (n=batsman@57.Red-83-32-13.dynamicIP.rima-tde.net) joined #pypy.
----- silence for 39 minutes ----- [10:38] <mwh> morning
[10:39] <mwh> how is gtbg today? :)
[10:41] Action: xorAxAx gets "invalid literal for long(): x80000000" while trying to translate pypy on self-compiled py2.5a0
[10:41] <xorAxAx> (on windows)
[10:42] <mwh> er, that's a funny one
[10:42] <mwh> does py.py run properly?
[10:43] Action: xorAxAx tries that
[10:45] <mwh> my guess would be something funny in rarithmetic, but that's very much a guess
[10:45] <mwh> i think armin ran pypy on 2.5a0 on linux
[10:46] <xorAxAx> mwh: that works (after cleaning pycs)
[10:47] <mwh> well, i guess there's always tracebacks :)
[10:47] <xorAxAx> the last line it prints is "TYPE(ParserError)"
[10:47] <mwh> hmm
[10:47] <xorAxAx> typeobject
[10:48] <mwh> where is the failing call?
[10:48] <xorAxAx> flowgraph in module ... pypy.__builtin__...)Module.pick_builtin
[10:48] <mwh> !?
[10:48] <mwh> can you paste the last few hundred lines of the output into a pastebig?
[10:48] <xorAxAx> yeah
[10:51] <xorAxAx> mwh: http://paste.ubuntuusers.de/760
[10:52] <mwh> eh-uh
[10:52] <mwh> it's an app-level crash?
[10:53] <xorAxAx> no idea, does it look like that?
[10:53] <xorAxAx> the "spamblock" doesnt look very applevellike
[10:53] <xorAxAx> hmm, app_array
[10:53] <xorAxAx> geninterplevel
[10:54] <xorAxAx> 'x80000000' is really not valid as a long
[10:54] <xorAxAx> in none of my python installations :)
[10:54] <mwh> it's in _cache
[10:54] <xorAxAx> (see tb, there is it)
[10:54] <xorAxAx> yeah, i cleaned it before
[10:54] <mwh> oh
[10:55] <mwh> hrm
[10:55] arigo (n=arigo@c-178b70d5.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.
[10:55] <mwh> can you delete _cache and then run transtator/test/test_geninterp with 2.5 ?
[10:55] <xorAxAx> ok
[10:55] <mwh> arigo: hello
[10:56] <arigo> hello
[11:00] <xorAxAx> _cache is in pypy/ right? then i deleted it already and the test doesnt fill it
[11:00] <xorAxAx> hmm, all tests fail
[11:00] <xorAxAx> of course i dont expect this to work because its 2.5, but its interesting nevertheless :)
[11:01] <xorAxAx> ah, bad marshall data. deleting pycs ...
[11:01] <xorAxAx> ok, cache is filled again :)
[11:02] <xorAxAx> 28 tests passed
[11:02] <xorAxAx> .oO(yes, i cleaned the pycs before translate_pypy as well)
[11:03] <xorAxAx> where does app_array come from? from the pypy tree or the py2.5 lib folder?
[11:03] <mwh> geninterp weirdness
[11:03] <mwh> "joy"
[11:03] <xorAxAx> yeah, geninterp seems to be a very stable component of pypy :)
[11:04] <mwh> uh
[11:04] <mwh> why do i have to have docutils installed to run py.test ?
[11:05] <xorAxAx> mwh: i asked that myself as well ;-)
[11:05] <xorAxAx> some conftest module imports it
[11:08] <mwh> ../Python/compile.c:3843: failed assertion `d_lineno >= 0'
[11:08] <mwh> Abort trap
[11:08] <mwh> "hmm"
[11:10] <mwh> arigo: you ran pypy on 2.5 right?
[11:10] <stakkars> xorAxAx: yes, it is, unless the rest of PyPy is changed in a way that geninterp's assumptions become wrong.
[11:10] <arigo> mwh: er, at one point
[11:11] <mwh> it's being odd at me now
[11:12] Action: arigo updates python
[11:13] <arigo> mwh: running py.test crashes compile.c in this way? which dir?
[11:13] <mwh> uh, test_geninterp
[11:13] <mwh> but it's flowcontext.py
[11:14] <arigo> grr, no docutils :-)
[11:14] <mwh> yes
[11:14] <mwh> i just filed a bug about that
[11:14] <mwh> try ~/blah/python flowcontext.py
[11:15] <arigo> he
[11:15] <arigo> looks like a new AST bug
[11:15] <arigo> it used to work
[11:16] <arigo> although I'm not sure I ever tried with a debug python
[11:17] <mwh> >>> from pypy.objspace.flow.model import *
[11:17] <mwh> ../Python/compile.c:3843: failed assertion `d_lineno >= 0'
[11:17] <mwh> Abort trap
[11:17] <xorAxAx> iterestingly, i dont see that bug
[11:18] <xorAxAx> ah, i have a release build
[11:18] <xorAxAx> so no assertions active :)
[11:18] <stakkars> arigo: where are you? (physically)
[11:18] <arigo> stakkars: at Linnegatan, where I usually am
[11:18] <arigo> i.e. in Laura and Jacob's previous flat
[11:18] <xorAxAx> mwh: so the fix is to build a releaseversion :)
[11:18] <stakkars> is there some meetingat Laura'splanned to day?
[11:18] <mwh> xorAxAx: i have my doubts :)
[11:19] <xorAxAx> at least at this point, in order to be able to reproduce the bug
[11:20] <mwh> sorry, i'm chasing the cpython bug now :)
[11:20] <arigo> mwh: here is a minimal example:
[11:20] <arigo> assert 1, ([s for s in x] +
[11:20] <arigo> [s for s in x])
[11:20] <arigo> pass
[11:20] <xorAxAx> :-)
[11:21] <mwh> oh, argh
[11:21] <arigo> stakkars: no, but I'm sure you can show up there :-) we saw them yesterday evening
[11:22] <arigo> mwh: do you want to chase more, or should I add it to the SF tracker about AST bugs?
[11:22] <mwh> arigo: i should pack, and, uh, travel to sweden :)
[11:22] <arigo> :-)
----- silence for 16 minutes ----- [11:38] Rhamphoryncus (n=rhamph@unaffiliated/rhamphoryncus) left irc: "My vapourware can beat up your vapourware! | Support ISO 8601! http://www.cl.cam.ac.uk/~mgk25/iso-time.html";
[11:46] <braintone> arigo - r20632 - A helper for Reprs that represent true/false objects with non-null/null pointers.
[11:47] <braintone> arigo - r20633 - Hacked on conftest.py until it again does the Right Thing with interp- and app-level KeyboardInterrupt.
[11:49] xlthlx (n=Miranda@bsd.pme.nthu.edu.tw) joined #pypy.
[12:00] <mwh> arigo: is that why you sometimes have to bounce on ^C to exit py.test ?
[12:00] <arigo> yes
[12:00] <arigo> no longer, hopefully
[12:00] <mwh> cool
[12:01] <arigo> there was already code trying to do that, but half-heartedly
[12:01] <mwh> i see
----- silence for 16 minutes ----- [12:17] xlthlx (n=Miranda@bsd.pme.nthu.edu.tw) left irc: Read error: 104 (Connection reset by peer)
[12:25] mwh (n=mwh@62-31-83-2.cable.ubr01.azte.blueyonder.co.uk) left irc: "bye"
[12:25] -ChanServ (ChanServ@services.)- You do not have channel operator access to [#pypy-sync]
[12:28] cfbolz (n=carlson@c-51c6e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.
[12:28] <cfbolz> hi!
[12:28] <arigo> hi Carl!
[12:28] <cfbolz> hi armin! pity not to have met you yesterday
[12:36] mwh (n=mwh@62-31-83-2.cable.ubr01.azte.blueyonder.co.uk) joined #pypy.
[12:37] <mwh> hey carl
[12:41] <cfbolz> hey michael!
[12:41] <cfbolz> where are you now?
[12:42] <cfbolz> I still don't know the room number yet, will go there at 4pm (they are closed before)
[12:44] <mwh> i'm still at home
[12:44] <mwh> i'll leave in about 1.5 hours
[12:45] <cfbolz> I see. when do you arrive again?
[12:47] <mwh> um, 22:30 or so?
[12:47] <mwh> um, 21:50
[12:47] <mwh> so i guess i'll get to korsvägen ~2230
[12:48] <cfbolz> ok, it is possible that I am still at Laura's at that time still. I will send you a text saying where I am + the room number
[12:48] <mwh> that would be very useful
[12:49] <mwh> i think i can remember how to get to sgs, i'm sure i can find laura's
[12:49] <stakkars> cfbolz: you are at Laura's right now?
[12:49] <cfbolz> mwh: we could also say we meet here and go to sgs then, if you want
[12:49] <cfbolz> stakkars: yes
[12:50] <mwh> well, i'm going to save some 'how to get to sgs' maps now
[12:50] <mwh> don't let me influence your plans too much :)
[12:50] <cfbolz> ok, then I'll just send that text
[12:51] <mwh> sounds like a good plan
[12:52] <cfbolz> yep
[12:52] <cfbolz> stakkars: why?
[12:52] <stakkars> does anybody know when rxe is going to arrive?
[12:53] <cfbolz> is he coming?
[12:53] <cfbolz> if yes, probably monday
[12:53] <cfbolz> or even tuesday
[12:53] <stakkars> cfbolz: I'm just sitting in the hotel, doing some homework, and tried to figure out if I'mexpected to show up,somewhere and when
[12:53] <lac> I don't know if he is coming.
[12:53] <lac> he said he wanted too, but never told me more than that.
[12:53] <stakkars> he said he willcome.
[12:53] <lac> great. There is a bed here for him.
[12:54] <cfbolz> stakkars: you are expected to show up here tomorrow morning the latest
[12:54] <cfbolz> lac: hi Laura!
[12:54] <cfbolz> :-)
[12:54] <stakkars> as I understood it, he's at EWT and flying directly here
[12:55] <cfbolz> mwh: py.test does not require docutils anymore now
[12:55] <stakkars> I need to talk to him
[12:55] <mwh> cfbolz: thanks
[12:56] <stakkars> lac: sorry about yesterday, I was wrapped up in some conversations and then fell asleep :)
[13:03] <cfbolz> mwh: argh! your blog is pointing to draft-dynamic!
[13:03] <cfbolz> that link is dead now!
[13:04] <cfbolz> and to draft-low-level too
[13:04] <mwh> isn't there a redirect?
[13:04] <mwh> it's an old 'this week...'
[13:04] <cfbolz> I know. there is no redirect :-(
[13:06] <mwh> what's this 'redirections' file, then?
[13:06] <cfbolz> it's a relict
[13:07] <mwh> ah
[13:07] <cfbolz> I updated it just as a reminder
[13:07] <cfbolz> maube now is the time then to fix this
[13:07] <cfbolz> although I probably would need hpk,
[13:07] <cfbolz> so fixing the blog still makes sense
[13:08] <hpk> hi
[13:08] <stakkars> hi Holger!
[13:08] <cfbolz> hi holger@
[13:09] <hpk> moin
[13:09] <stakkars> to which WP do I assign last week's pypy-tb meeting?
[13:10] <hpk> to the best fitting WP where tismerysoft is allocated for
[13:10] <mwh> ok, blog updated
[13:10] <hpk> stakkars: i know that's not very precise, but there is no "TB" wp
[13:10] <stakkars> I see. If I don't have WP14, don't use it at all.
[13:11] <stakkars> (we should have assigned a bit to all tb members I guess)
[13:11] <cfbolz> mwh: cool
[13:11] Action: mwh wonders what would happen if you just checked in a .htaccess file
[13:12] <hpk> cfbolz: your docutils fix is a bit shallow
[13:12] <hpk> cfbolz: there already was logic for checking/skipping
[13:13] <cfbolz> hpk: what really matters is the second checkin, I guess I can revert the first
[13:14] <hpk> cfbolz: i guess so, the conftest was supposed to already skip things appropriately and it did at one point
[13:14] <cfbolz> ok, I will have a look
[13:16] <cfbolz> hum, there is a doctest test failing: py/documentation/test_conftest.py
[13:16] <cfbolz> even before my checkin
[13:18] <hpk> doesn't fail for me
[13:18] Action: hpk still has a visitor, is briefly out
[13:18] <cfbolz> hpk: yes, but you have docutils, right?
[13:19] <hpk> yes
[13:20] <cfbolz> I don't
[13:20] <hpk> cfbolz: should be fixed, i am off
[13:21] <cfbolz> I will do it
[13:22] <hpk> cfbolz: i meant, i have just fixed that
[13:23] <cfbolz> yes, I figured by now :-)
[13:23] <cfbolz> thanks
[13:23] <cfbolz> works
----- silence for 19 minutes ----- [13:42] <braintone> arigo - r20639 - Simplifying the L3 graph model...
[13:44] <cfbolz> arigo: cool!
[13:44] <cfbolz> (although on the other hand depressing that you rewrote the whole outcome of the paris sprint in one day :-)
[13:47] <hpk> cfbolz: it wasn't much more than a day or two of discussions and coding anyway :)
[13:47] <stakkars> giving up on the idea of competition is crucial, here :)
[13:47] <cfbolz> it was a bit more :-)
[13:47] <cfbolz> stakkars: sure
[13:48] <stakkars> (I mean not in general, but in particular)
[13:53] <arigo> cfbolz: that look me quite more than a day of thinking
[13:54] <cfbolz> I know. I was not really serious
[13:54] <arigo> :-)
[13:58] Gromit (n=bear@dialin-212-144-183-213.pools.arcor-ip.net) joined #pypy.
[13:58] <Gromit> hi
[13:59] <arigo> hi
[14:00] <braintone> arigo - r20640 - Got rid of the intermediate buffer, by copying data around the stack. Moved the code in a function that gets specialized for the three different stacks.
[14:01] <cfbolz> any idea what the unit of measurement in a ps is?
[14:02] <Gromit> ps == playstation?
[14:02] <cfbolz> postscript
[14:02] <Gromit> AFAIK 1/72 of an inch
[14:03] <cfbolz> aha. does this thing have a name?
[14:03] <stakkars> arigo: why do we have three different stacks?
[14:03] <Gromit> probaly it has a new
[14:03] <stakkars> arigo: just becuase we can't do better ATM, or is it a principle?
[14:03] <mwh> cfbolz: "points" isn't it?
[14:04] <cfbolz> oh, so _that_ is a point
[14:04] <Gromit> mwh: points sounds familiar
[14:04] <stakkars> 1 pt == 1/72 in
[14:04] <cfbolz> stakkars: for the same reason why we have three different sorts of things to store away in stackless
[14:05] <mwh> addresses floats and ints?
[14:05] <cfbolz> yes
[14:05] <stakkars> that's no reason at all.
[14:05] <cfbolz> it sure is
[14:05] <mwh> well, doubles, but...
[14:06] <stakkars> I agree that it makes sense for the stackless approach, since it clarifies everything and does not cost time.
[14:06] <cfbolz> stakkars: it also fits the real processors model (having different sorts of registers)
[14:06] <stakkars> but for an L3 interpreter, maintaining more stacks than physically necessary needs to include other considerations.
[14:06] <Gromit> which real processor?
[14:07] <cfbolz> Gromit: how many processors use the same registers for integers and doubles?
[14:07] <braintone> arigo - r20641 - Make and use a more appropriate specialization tag.
[14:07] <stakkars> cfbolz: how many processors do use three different stacks?
[14:08] Action: Gromit just tries to remeber how sparcs handle floats
[14:08] <cfbolz> stakkars: what does that have to do with it?
[14:08] <stakkars> I have no objection against this, if it is driven by reasonable reasoning. That's why I asked if we want to do it this way, or if we simply cannot express it differently.
[14:09] <Gromit> at least it doesn't make sense on an x-scale, due to lack of floating point hardware
[14:09] <cfbolz> stakkars: it was the result of a few days of thinking in paris
[14:10] <stakkars> if we are trying to model something close enough to real processors, it is stronglyrelated, IMHO.
[14:10] <stakkars> (unless the stacks are going away, being turned into registers after all)
[14:10] <arigo> cfbolz: the real reason why we have three stack is that there is no sane way to be RPython with a single one
[14:10] <cfbolz> yes, that's the plan
[14:11] <arigo> stakkars: the goal of the L3 interp is to be "rather fast", not "as fast as it would be possible in C"
[14:11] <cfbolz> arigo: except boxing
[14:11] <arigo> cfbolz: yes, but that's most probably costing more
[14:11] <stakkars> arigo: that's what I hopedto heat. tx
[14:11] <stakkars> ^h^h^h^h^h
[14:11] <cfbolz> arigo: sure
[14:12] <stakkars> arigo: that's what I expected to hear
[14:12] <stakkars> when we are supporting true processors, the L(n) for which n might that be, then? :-)
[14:15] <stakkars> arigo: about "sane way" and RPython: I thought we have an address notation and operations on that. Can't we express stack access via address ops and type casts? (or is that insane?)
[14:15] <stakkars> unless we want to save type information for further optimization, like register allocation etc.
[14:16] <arigo> stakkars: well, as I said the goal is to keep the L3Interp readable
[14:18] <stakkars> ok, I agree on readability. "debuggable" is something else, not that easy with 3 stacks, butok.
[14:21] <cfbolz> why is debugging so much harder with three stacks?
[14:22] <stakkars> it is less comfortable if you have to use existing debuggers. forget it.
[14:23] Gromit (n=bear@dialin-212-144-183-213.pools.arcor-ip.net) left irc: "lunch time"
[14:25] <mwh> right, time to go
[14:25] <mwh> see you later! :)
[14:25] mwh (n=mwh@62-31-83-2.cable.ubr01.azte.blueyonder.co.uk) left irc: "-> .se"
[14:25] -ChanServ (ChanServ@services.)- You do not have channel operator access to [#pypy-sync]
[14:27] <stakkars> it is probably because I was running in "real mode" for a while
[14:28] <cfbolz> :-)
[14:29] <stakkars> and I'd like to see PyPy "hitting earth" a little more often
[14:31] <cfbolz> yes, but we did not promise "hitting earth" that directly to the EU. and we have to make sure that the EU project succeeds
[14:42] <stakkars> that's true. And the reason why I'm going to use Stackless as a show-case for "Real mode PyPy"
[14:43] pedronis (n=Samuele_@c-178b70d5.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.
[14:43] Gromit (n=bear@dialin-212-144-178-132.pools.arcor-ip.net) joined #pypy.
[14:43] <Gromit> re
[14:44] <stakkars> (actually, all that much optimization allocated to WP07 was meant to be this, at least as I perceived it)
[14:57] <stakkars> Gromit: yes?
[14:58] <Gromit> yes?
[14:59] <stakkars> you said "re" :)
[15:00] <Gromit> I justed wanted to indicate that I "returned"
[15:00] <stakkars> got it
[15:03] <stakkars> arigo: what will happen if we add 64 bit support: grow a fourth stack? :-)
[15:04] <Gromit> :-]
[15:04] <stakkars> and what happens for a machine where this is the same as an integer? (or don't these questions make sense)
[15:04] <pedronis> arigo is having lunch somewhere
[15:04] <Gromit> e.g. on an alpha
[15:04] <pedronis> stakkars: not in the context of what RPython currently allows
[15:05] <stakkars> pedronis: I might have lunch there, too and ask him directly :-)
[15:06] <stakkars> pedronis: not what: supporting 64 bit, or not growing another stack?
[15:06] <pedronis> RPython has ints which are Signed which are C long at moment
[15:07] <stakkars> but we started discussing an extension.
[15:07] <stakkars> which is needed anyway, see os.write
[15:07] <stakkars> not write, but all the file position related stuff
[15:08] <pedronis> the JVM for example sort of uses two 32bits stack positions for 64 bits numbers
[15:09] <stakkars> sure, and there are a bazillion other examples how real and virtual machines do that, but what do you mean?
[15:10] <pedronis> that we need to do things in order: if we extend the extent of allowed int ranges, we need to think that through, starting or focusing on l3interp is not really the right way to think about this
[15:11] <stakkars> I thought the other way round: the way l3interp drives its decisions is ignoring the problems to come.
[15:12] <pedronis> my example was just pointing out that the number of stacks in l3 and the integer ranges you support are not completely related
[15:12] <pedronis> iow we can have a larger int range but still have 3 stacks
[15:17] <arigo> this is all mostly irrelevant at this point
[15:18] <arigo> it's not like adding or removing a few stacks is difficult
[15:18] <stakkars> no. it just doesn'tmake sense
[15:19] <arigo> if you want to rewrite l3interp.py without making it more obscure that it's already is, you're welcome
[15:20] <arigo> but keep in mind that there are as usual many levels involved
[15:20] <stakkars> no, I want to support more datatypes, and I fear new assumptions which make this harder.
[15:21] <arigo> the problem is that we have to progress in the direction of the JIT too
[15:22] <arigo> but well, all this is my fault
[15:22] <arigo> I shouldn't have started hacking around on l3interp
[15:22] <arigo> it would be better not to think about it during this sprint
[15:23] <arigo> it would probably turn out to be like at Paris,
[15:24] <arigo> i.e. mostly hand-wavy guesses about what we really need
[15:25] <Gromit> what must be done, that we know better what we will need?
[15:26] <stakkars> use PyPy in a production environment, and you know very well
[15:28] <Gromit> pro*** en****, the f*** word!
[15:29] <Gromit> sorry could not resist
[15:38] <braintone> arigo - r20642 - Concrete short-term JIT plans.
[15:38] <arigo> Gromit: this :-)
[15:39] <Gromit> will check it out, thanks
[15:41] xyz359 (n=xyz359@pool-68-238-168-248.tampfl.fios.verizon.net) joined #pypy.
[15:42] <stakkars> arigo: well :)
[15:44] Action: arigo is in "research" mode -- he doesn't think a JIT can be invented without it
[15:45] <arigo> stakkars, cfbolz: do you feel like making a plan for dinner?
[15:45] Action: stakkars agrees that nothing can be invented outside of "research" mode (or it is just an application)
[15:50] stakkars (n=tismer@213.15.68.61) left irc:
[15:51] stakkars (n=tismer@213.15.68.61) joined #pypy.
[15:56] Nick change: stakkars -> stakkars|afk
[16:08] <cfbolz> arigo: yes
[16:08] <cfbolz> now
[16:08] <cfbolz> :-)
[16:11] <arigo> you will be at SGS later?
[16:12] <cfbolz> I am at laura's now, but I have to go to SGS at one point, yes
[16:12] <cfbolz> (before 8
[16:12] <cfbolz> (
[16:13] <cfbolz> what are your plans?
[16:13] <arigo> none so far
[16:13] <cfbolz> I could also just come to your place and we then decide what to do
[16:13] <arigo> yes, that's a good plan :-)
[16:14] <arigo> do you need instructions to arrive here?
[16:14] <cfbolz> I can ask laura
[16:14] <arigo> ok
[16:14] <cfbolz> I have a map here, so she can show me
[16:14] <arigo> ok. the street address is Linnegatan 47
[16:14] <cfbolz> I guess I will be there between 5.30 and 6, is that fine?
[16:14] <arigo> sure
[16:14] <cfbolz> great!
[16:14] <arigo> we can go to a nearby resturant then
[16:14] <cfbolz> yes
[16:15] <cfbolz> I actually have some ideas about the mini-interpreter
[16:15] <cfbolz> we can discuss that then
[16:21] Gromit (n=bear@dialin-212-144-178-132.pools.arcor-ip.net) left irc: "reconfiguring lcr"
[16:24] ericvrp (n=eric@ip54503ec6.speed.planet.nl) joined #pypy.
[16:25] ericvrp (n=eric@ip54503ec6.speed.planet.nl) left irc: Client Quit
[16:28] _hannes (i=ore@i577B5081.versanet.de) joined #pypy.
[16:38] <braintone> arigo - r20643 - Adapted convertgraph to the new model. Fun so far :-)
[16:49] Gromit (n=bear@does-d9b91938.pool.mediaWays.net) joined #pypy.
[16:49] <cfbolz> arigo: it is possible that I am a bit late -- desert took longer than I expected
[16:50] <cfbolz> I am leaving to SGS now
[16:50] <arigo> fine :-)
[16:50] <cfbolz> see you then
[16:50] <arigo> ok, see you
[16:50] cfbolz (n=carlson@c-51c6e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: "bye"
----- silence for 37 minutes ----- [17:27] xyz359 (n=xyz359@pool-68-238-168-248.tampfl.fios.verizon.net) left irc: "http://www.purepwnage.com/";
[17:31] sanxiyn (n=tinuviel@218.51.138.12) joined #pypy.
[17:32] <sanxiyn> Seems to be a bug:
[17:32] <sanxiyn> test/a.py: import b
[17:32] <sanxiyn> test/b.py: (empty)
[17:32] <sanxiyn> python a.py # ok
[17:32] <sanxiyn> py.py a.py # ImportError: No module named b
[17:32] <sanxiyn> Eh, I mean, py.py test/a.py
[17:32] <sanxiyn> Any idea?
[17:33] <arigo> which python version?
[17:34] <arigo> 2.3 and 2.4 differ in that respect, if I remember correctly
[17:34] _hannes (i=ore@i577B5081.versanet.de) left irc: Read error: 104 (Connection reset by peer)
[17:35] <sanxiyn> python2.3 test/a.py # ok
[17:35] <sanxiyn> python2.4 test/a.py # ok
[17:35] <sanxiyn> python2.3 /path/to/py.py test/a.py # ImportError
[17:35] <sanxiyn> python2.4 /path/to/py.py test/a.py # ImportError
[17:36] <sanxiyn> I think, when python path/to/script.py is run, path/to is added to sys.path?
[17:36] <arigo> ah
[17:36] <sanxiyn> (CPython adds it, and PyPy doesn't?)
[17:36] <arigo> indeed, that may be the job of site.py, which PyPy doesn't run
[17:37] <sanxiyn> arigo: python -S test/a.py still succeeds.
[17:38] <arigo> ah
[17:39] <sanxiyn> Hm, I modified a.py to print sys.path.
[17:39] <sanxiyn> python test/a.py
[17:39] <arigo> just did so too
[17:40] _hannes (n=yuuhu@i577B5081.versanet.de) joined #pypy.
[17:41] <arigo> where is sys.path initialized in CPython...
[17:42] <sanxiyn> pypy.module.sys.state setinitialpath
[17:42] <sanxiyn> Eh, CPython...
[17:42] <arigo> :-)
[17:45] <sanxiyn> Here: http://pxr.openlook.org/pxr/source/Python/sysmodule.c#1255
[17:45] _hannes (n=yuuhu@i577B5081.versanet.de) left irc: Read error: 104 (Connection reset by peer)
[17:46] _hannes (n=yuuhu@i577B5081.versanet.de) joined #pypy.
[17:46] <sanxiyn> (Well, I *think* it's there.)
[17:46] Nick change: stakkars|afk -> stakkars
[17:47] <arigo> indeed, thanks
[17:47] <arigo> it's where a new sys.path[0] is inserted
[17:48] <sanxiyn> I see CPython does incredible circus to determine sys.argv[0]...
[17:49] <arigo> yes
[17:51] <arigo> hum, in a plain 'python' we get sys.argv == [''] but in a plain 'py.py' we get []
[17:53] <arigo> it's all a bit messy because py.py and pypy-c don't use the same start-up code
[17:53] <arigo> for pypy-c, it's in translator/goal/app_main.py
[17:54] <arigo> py.py uses pypy/interpreter/{interactive.py,main.py}
[17:54] <sanxiyn> arigo: Hm, above code is just "script directory insertion" part, and
[17:54] <sanxiyn> arigo: Real job is on Modules/getpath.c.
[17:55] <sanxiyn> /* Return the initial module search path. */
[17:55] <arigo> ok, I see
[17:55] Gromit_ (n=bear@does-d9b91932.pool.mediaWays.net) joined #pypy.
[17:56] <arigo> we have a nice mess in PyPy as well :-/
[17:56] <sanxiyn> getpath.c is 673 lines long :(
[17:58] Gromit (n=bear@does-d9b91938.pool.mediaWays.net) left irc: Nick collision from services.
[17:58] Nick change: Gromit_ -> Gromit
[17:59] <stakkars> hi again
[18:00] _hannes (n=yuuhu@i577B5081.versanet.de) left irc: Read error: 104 (Connection reset by peer)
[18:00] <sanxiyn> arigo: Same-looking code all different :(
[18:01] <arigo> hi stakkars. would you like to join us for a restaurant?
[18:01] <stakkars> nice offer!
[18:02] <arigo> do you know where our place is? Linnegatan 47 ?
[18:03] <stakkars> but I'm in a bad mood, thinking of PyPy's future and what peopleareexpecting from it
[18:03] <stakkars> guessso.
[18:04] <sanxiyn> Bon appetit. :-)
[18:05] <arigo> :-)
[18:05] <arigo> stakkars: if you hurry, join us here, otherwise... we're trying to choose the restaurant...
[18:06] <lac> stakkars: the #6 and #1 tram run there from Brunsparken
[18:07] <stakkars> I think it's better if I stay here and write up some considerations/discussion proposals instead of attacking people during dinner (switched from real mode to aggressive mode)
[18:07] <arigo> well, as you like -- real-life discussions can be better than IRC ones, though
[18:08] <xorAxAx> and they are better than sending letters :)
[18:08] <lac> plus good food can cheer you up.
[18:08] hpk (n=hpk@merlinux.de) left irc: Read error: 104 (Connection reset by peer)
[18:09] hpk (n=hpk@merlinux.de) joined #pypy.
[18:10] <sanxiyn> OT: I read NY Times (as well as many local newspaper) articles on Dr. Hwang... Sad.
[18:11] <xorAxAx> sanxiyn: who is dr. hwang?
[18:11] <sanxiyn> xorAxAx: Leading researcher on human stem cell field.
[18:11] <sanxiyn> xorAxAx: Read http://www.nytimes.com/2005/12/04/opinion/04sun2.html
[18:12] <xorAxAx> thanks
[18:12] <sanxiyn> (It's on the first page of newspapers here in Korea.)
[18:14] <stakkars> arigo, lac: you are right.
[18:14] <xorAxAx> sanxiyn: and how is it reported about by korean newspapers?
[18:14] <sanxiyn> xorAxAx: Not really different. Lots more details though.
[18:16] <stakkars> ATM I'm writing up my perspective of the project, and I hope we can make upsome time on the next two days to discuss this a little bit (proposing a real meeting)
[18:16] <stakkars> thenI promise to try to get into virtual abstract mode for the rest of the week. :-)
[18:17] <xorAxAx> stakkars: dont forget to activate the A20 gate ;-)
[18:20] <braintone> arigo - r20645 - Insert the dir of the script in sys.path[0]. Thanks Seo. This initialization is rather messy at the moment, it will need a good refactoring at one point...
[18:20] <stakkars> ? I'm talking of a project that's continously pissing off its potential users.
[18:22] __doc__ (i=doc_pyth@84-73-210-113.dclient.hispeed.ch) joined #pypy.
[18:22] <xorAxAx> stakkars: so, what is the virtual abstract mode?
[18:23] Cheery (i=Henri@a81-197-18-99.elisa-laajakaista.fi) joined #pypy.
[18:23] <stakkars> look at the current sprint goals. All fine stuff, and according to what we promised to the EU, it is ok.
[18:23] Nick change: arigo -> arigo_dinner
[18:24] <arigo_dinner> (stakkars: you know Samuele's mobile in case you want to join us, otherwise ask Laura)
[18:26] <Gromit> a rather silly question: when will pypy be ready for production use?
[18:26] <stakkars> arigo_dinner: thanks. Does Laura come with you?
[18:26] <xorAxAx> Gromit: depends on your goals
[18:26] <xorAxAx> Gromit: stakkars uses it in production environments
[18:26] <stakkars> Gromit: I learned today that production use was not in our goals.
[18:27] <Gromit> my goals can be stated in one sentence: A complete cpython replacement with java's performance
[18:28] <Gromit> (and a smalltalk like programming environment)
[18:28] <stakkars> we are going to implement a Stacklessmodule, which is useless, since we know how to do it, and nobody will use PyPy unless we fulfil the basic expectations about speed (which are implicit)
[18:28] <stakkars> we are workingon a JIT, which is also nothing really new, because we know how Psyco works and that this is doable.
[18:29] <xorAxAx> .oO(a smalltalk like environment %-))
[18:30] <stakkars> My understanding of all of these was that we would work on it after the other goals are reached. Now I have the impression that the JIT idea should fill in what we didn't reach so far.
[18:30] <Gromit> xorAxAx: and that's not something liek eclipse
[18:30] <xorAxAx> Gromit: i know smalltalk, at least good enough to understand your claim
[18:30] <xorAxAx> Gromit: and how much it differs from reality and current implementation of pypy ;-)
[18:30] <xorAxAx> Gromit: why do you think that pypy would solve this "smalltalk-goal" per se?
[18:30] <Gromit> ok
[18:31] <xorAxAx> the current interpreter is very cpython like
[18:31] <xorAxAx> with pyc files etc.
[18:31] <xorAxAx> of course you could start implementing your new world-loader
[18:31] <xorAxAx> :-)
[18:31] <Gromit> it's object space's should be well suited to implement smalltak like vm images
[18:31] <xorAxAx> yeah
[18:31] <stakkars> WHat I'd propose is to make an amendment if possible, stating that we postpone advanced technologies, in favor of getting the ground work to a level where we really become a CPython alternative. Doing anything without this makes a fine EU project, but no applications.
[18:32] <xorAxAx> Gromit: but you need to find guys that want to implement that ;-)
[18:32] <xorAxAx> .oO(or use squeak in the meantime :))
[18:32] <Gromit> sigh, squeak is an other sad story of doing things in an ivory tower
[18:33] Action: stakkars should maybe go to the restaurant and bite into a steak?
[18:33] <xorAxAx> stakkars: yes
[18:33] <Gromit> stakkars: yep, you should do that
[18:33] <xorAxAx> Gromit: ok, there is visual age
[18:34] <stakkars> but I might bite somebody else.
[18:34] <Gromit> and Smalltalk/X
[18:34] <xorAxAx> Gromit: you know, ibm, big company, productive .... (discontinued)
[18:34] <Gromit> stakkars: that's why you should go :)
[18:34] <xorAxAx> i guess because >>> not ('java' in 'smalltalk')
[18:34] <Gromit> yep
[18:35] <xorAxAx> ok, i am gone
[18:35] <sanxiyn> Gromit: Cincom?
[18:35] <Gromit> they paid a lot of money to aquire OTI
[18:36] <Gromit> that wa steh company taht implemeted visual age's core
[18:36] <sanxiyn> Gromit: What is OTI?
[18:36] <Gromit> and envy developer
[18:36] <Gromit> which was a OODB based repostitory, whcih IBM ditched in favour of CVS
[18:37] <Gromit> that topic is too sad
[18:37] <sanxiyn> Squeak... what's wrong with Squeak.
[18:38] <Gromit> it's lack of integration with exsting window systems
[18:42] pedronis (n=Samuele_@c-178b70d5.022-54-67626719.cust.bredbandsbolaget.se) left irc: Read error: 110 (Connection timed out)
[18:50] _hannes (n=yuuhu@i577B5081.versanet.de) joined #pypy.
[18:57] _hannes (n=yuuhu@i577B5081.versanet.de) left irc: Read error: 104 (Connection reset by peer)
[19:03] _hannes (n=yuuhu@i577B5081.versanet.de) joined #pypy.
[19:03] sanxiyn (n=tinuviel@218.51.138.12) left irc: "Bye"
[19:08] xyz359 (n=xyz359@pool-68-238-168-248.tampfl.fios.verizon.net) joined #pypy.
[19:16] xyz359 (n=xyz359@pool-68-238-168-248.tampfl.fios.verizon.net) left irc: "http://www.purepwnage.com/";
----- silence for 1 hr and 51 minutes ----- [21:07] _hannes (n=yuuhu@i577B5081.versanet.de) left irc: Read error: 110 (Connection timed out)
[21:15] tic (n=vision@c-8faae255.019-35-67626717.cust.bredbandsbolaget.se) left irc: Read error: 113 (No route to host)
[21:24] tic (n=vision@c-8faae255.019-35-67626717.cust.bredbandsbolaget.se) joined #pypy.
----- silence for 54 minutes ----- [22:18] <braintone> bea - r20646 - my part of the eu-workshop talk,,,,
[22:24] Rhamphoryncus (n=rhamph@unaffiliated/rhamphoryncus) joined #pypy.
[22:37] mwh (n=mwh@c-51c6e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.
[22:38] cfbolz (n=carlson@c-51c6e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.
[22:41] mwh (n=mwh@c-51c6e055.1321-1-64736c11.cust.bredbandsbolaget.se) left irc: Client Quit
[22:41] -ChanServ (ChanServ@services.)- You do not have channel operator access to [#pypy-sync]
[22:42] mwh (n=mwh@c-51c6e055.1321-1-64736c11.cust.bredbandsbolaget.se) joined #pypy.
[22:49] _hannes (n=yuuhu@i577B4133.versanet.de) joined #pypy.
----- silence for 18 minutes ----- [23:07] Cheery (i=Henri@a81-197-18-99.elisa-laajakaista.fi) left irc: "Leaving"
----- silence for 40 minutes ----- [23:47] Gromit (n=bear@does-d9b91932.pool.mediaWays.net) left irc: "Client Exiting"
[23:48] __doc__ (i=doc_pyth@84-73-210-113.dclient.hispeed.ch) left irc:
[00:00] --- Mon Dec 5 2005