==== Channel ##pypy: 05/18/05 ====

[00:30] senra (~rodrigo@200165251184.user.veloxzone.com.br) joined #pypy.

[00:33] <senra> hi, could anybody here possibly tell me how to contact Laura Creighton via chat ?

[00:41] cfbolz (~cfbolz@hdlb-d9b946af.pool.mediaWays.net) joined #pypy.

[00:42] <cfbolz> senra: She's in #pypy (here) quite regularly

[00:42] stakkars (upvfodu@i3ED6B51A.versanet.de) joined #pypy.

[00:43] <cfbolz> hi christian!

[00:43] <senra> cfbolz: Thanks I'll try tomorrow.

[00:43] <stakkars> hi Carl Friedrich!

[00:43] senra (~rodrigo@200165251184.user.veloxzone.com.br) left irc: "Leaving"

[00:44] <stakkars> I've been in "closure" whole day. Now we can use print and friends from all of flowspace!

[00:44] <cfbolz> oh, cool!

[00:45] <stakkars> actually, I think we can do this with everything that pyopcode currently implements by "applevel".

[00:46] <stakkars> I also found a way to let sys behave as expected, *always*

[00:46] <cfbolz> sorry, don't know enough about pyopcode

[00:47] <stakkars> the problem is that all printXXX stuff is implemented at app-level.

[00:47] <cfbolz> ok, got that

[00:47] <cfbolz> via some hooks

[00:47] <cfbolz> ?

[00:48] <stakkars> when we are flowing print-alike stuff, we need to give a native implementation.

[00:49] <stakkars> I'm doing that -- by plumbing in simple_call's to exactly the applevel functions which do it.

[00:49] <cfbolz> ah.

[00:49] <cfbolz> but those could be not RPython, right?

[00:49] <stakkars> They happen to be completrely RPython.

[00:50] <cfbolz> If they weren't this wouldn't work because it would confuse the annotator, right?

[00:50] <stakkars> yessir

[00:50] <stakkars> it struck be like an explosion.

[00:51] <cfbolz> :-)

[00:51] <stakkars> I took the implementation of pyopcode and copied it into specialcase.

[00:51] <cfbolz> and worked the whole day on it?

[00:51] <stakkars> Then I redirected the special calls for printing to that, and made them call back into simple_call.

[00:52] <stakkars> yes, almost.

[00:52] <cfbolz> well, you missed all the fun issue discussions

[00:52] <stakkars> whoops?

[00:52] <stakkars> I needed to be alone, as it is very often!

[00:52] <cfbolz> no problem. I wasn't accusing you of anything :-)

[00:53] <stakkars> what did I miss? (well, I can read it)

[00:53] <cfbolz> there were some issues we didn't discuss in depth because of you missing

[00:53] <stakkars> what I wanted to tell you -- I'm very very happy with this solution

[00:54] <cfbolz> yes, it sounds really clean (as far as I can judge that)

[00:54] <stakkars> some ranting about my coding style I fear

[00:54] <stakkars> :-))

[00:55] Action: stakkars reading blogs

[01:00] <stakkars> hello, guys!

[01:00] <cfbolz> I'm still there :-)

[01:00] <stakkars> yes,I was away, sitting in a cafe' allday!

[01:01] <stakkars> please assign to me whatever you see fit.

[01:01] <cfbolz> I think you allready got everything we saw fit

[01:02] <stakkars> the issue of printing has been solved, and I have learned *A LOT*

[01:02] <cfbolz> this being the "windows compatibility" issue?

[01:02] <stakkars> I also will integrate exceptions interpleveling.

[01:03] <stakkars> cfbolz: yes, that's completely resolved.

[01:03] <stakkars> I special-cased sys imports.

[01:03] <cfbolz> Ok. I'll mark it as such?

[01:03] <stakkars> mark what?

[01:03] <cfbolz> the issue as resolved

[01:04] <stakkars> yes

[01:04] <stakkars> the point is: I'm working on a bigger solution.

[01:04] <stakkars> It is not just the sys imports locally in a function.

[01:05] <stakkars> What I'm doing is to set up special cases for pseudo-Constant()s, which are not really constant.

[01:06] <stakkars> I'm cheating flowspace to produce a special call for these, which makes them into SomeObject()

[01:06] <stakkars> thant ensures that these objects are evaluated as late as possible, and it should work with

[01:06] <stakkars> a global "import sys". !!!

[01:07] <cfbolz> cool. So you can just use sys in the regular way?

[01:07] <stakkars> yes!

[01:07] <stakkars> the trick is to examine getattr and setattr for Constant(sys)

[01:08] <stakkars> and if it is so, then I add an instruction of "simple_call, Constant(unspecialize), arg)"

[01:09] <stakkars> this call is built in a way that the specializer is unable to take any type info,

[01:09] <cfbolz> and just leaves the calls to the CPython API?

[01:09] <stakkars> and so, all attribute access will be done at run-time.Phew

[01:09] <stakkars> exactly

[01:09] <cfbolz> ok. I think I got it.

[01:10] <cfbolz> I'm going to bed now. See you tommorrow!

[01:10] <stakkars> unspecialize is currently a simplefunction that tries to do the impossible

[01:10] <stakkars> if id(0) is id(None):

[01:10] <stakkars> return obj

[01:10] <stakkars> that makes the annotator give up.

[01:11] <stakkars> good night

[01:11] <cfbolz> bye

[01:11] cfbolz (~cfbolz@hdlb-d9b946af.pool.mediaWays.net) left irc: "Verlassend"

----- silence for 3 hr and 28 minutes -----

[04:39] thingie24 (~rmt38@valhalla.ccp.cc) left irc: Connection reset by peer

[04:39] thingie24 (~rmt38@valhalla.ccp.cc) joined #pypy.

----- silence for 31 minutes -----

[05:10] stakkars_ (wwhtlxt@i3ED6B51A.versanet.de) joined #pypy.

[05:10] stakkars (upvfodu@i3ED6B51A.versanet.de) left irc: Read error: 131 (Connection reset by peer)

[05:15] yuuh (pjaivjqn@i3ED6B51A.versanet.de) left irc: "utz utz utz"

----- silence for 1 hr -----

[06:15] sanxiyn (~tinuviel@211.104.100.115) joined #pypy.

----- silence for 2 hr -----

[08:15] adim (~adim@logilab.net2.nerim.net) joined #pypy.

----- silence for 48 minutes -----

[09:03] sanxiyn (tinuviel@211.104.100.115) left #pypy ("Bye").

[09:04] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

----- silence for 1 hr and 1 minute -----

[10:05] Gromit (~bear@B74a2.b.pppool.de) joined #pypy.

----- silence for 32 minutes -----

[10:37] arre (ac@ratthing-b3fa.strakt.com) joined #pypy.

[10:38] tea (~tea@cap049-090.kcn.ne.jp) joined #pypy.

[10:41] Gromit (~bear@B74a2.b.pppool.de) left irc: Read error: 104 (Connection reset by peer)

[10:52] Gromit (~bear@B74e0.b.pppool.de) joined #pypy.

----- silence for 1 hr and 24 minutes -----

[12:16] Gromit (~bear@B74e0.b.pppool.de) left irc: Read error: 145 (Connection timed out)

----- silence for 1 hr and 24 minutes -----

[13:40] yuuh (totoud@i528C13CD.versanet.de) joined #pypy.

[13:43] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "Eject! Eject! Eject!"

[13:47] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

[13:58] stakkars_ (wwhtlxt@i3ED6B51A.versanet.de) left irc: Read error: 110 (Connection timed out)

[14:13] arre (ac@ratthing-b3fa.strakt.com) left irc: Remote closed the connection

[14:13] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "Eject! Eject! Eject!"

[14:19] yuuh (totoud@i528C13CD.versanet.de) left irc: Read error: 54 (Connection reset by peer)

----- silence for 40 minutes -----

[14:59] tea (~tea@cap049-090.kcn.ne.jp) left irc: Remote closed the connection

----- silence for 19 minutes -----

[15:18] cfbolz (~cfbolz@hdlb-d9b95ee3.pool.mediaWays.net) joined #pypy.

[15:18] <cfbolz> hi all!

[15:18] <hpk> cfbolz: hi carl!

[15:18] <hpk> there is not much of "all" here, i guess

[15:19] Action: hpk is happily creating yet more issues

[15:19] <cfbolz> what? nobody else is working?

[15:19] <hpk> no clue

[15:19] <hpk> we usually don't report to each other low-level details :-)

[15:20] <cfbolz> well, now is not usually :-)

[15:20] <hpk> good point :-)

[15:20] <hpk> i guess christian, armin and samuele chickened out to work on translation

[15:21] <hpk> let's complain a bit then so they can read it in the logs :-)

[15:22] <cfbolz> well, translation is probably more fun. But not two days before a release.

[15:22] <hpk> actually i think that everybody thinks now that the issues are well resolvable before friday

[15:22] <hpk> and rests a bit

[15:23] <cfbolz> I understand. I'm not doing too much myself :-)

[15:23] <hpk> although i wouldn't mind getting done earlier instead of postponing to thursday night :-)

[15:24] <cfbolz> do you think the tracing will change before fryday?

[15:27] Action: hpk just "finished" module documentation, uff, took almost the whole day so far

[15:27] <hpk> which tracing do you mean, the space or the mechanism?

[15:28] <cfbolz> I mean the output if you do __pytrace__ = 1

[15:28] <cfbolz> what richard promised to do

[15:28] <hpk> ok, the space then

[15:28] <hpk> i think richard wants to work a bit on it

[15:29] <cfbolz> ok. Then I have to change that bit in getting_started. Otherwise I think I done with it (modulo errors)

[15:30] <hpk> set the issue to testing

[15:30] <cfbolz> ok.

[15:30] <cfbolz> I find you're module docs very interesting: I didn't know anything about it yet.

[15:31] <hpk> i actually had to fix a couple of stuff to make the documentation correct :-)

[15:31] <cfbolz> That's another good part about writing docs

[15:40] <cfbolz> exactly when on fryday will the release be made? when everythings done?

[15:41] <cfbolz> s/fryday/friday

[15:42] <hpk> i guess so

[15:43] <hpk> afternoon i guess

[15:43] <cfbolz> 'kay

[15:44] stakkars (~tismer@i528C13CD.versanet.de) joined #pypy.

[15:45] <cfbolz> hi!

[15:46] <hpk> stakkars: hi chris

[15:46] Action: hpk -> sun

[15:46] pedronis (pedronis@ratthing-b246.strakt.com) joined #pypy.

[15:46] <stakkars> hi!

[15:48] <cfbolz> I'm going. bye

[15:48] cfbolz (~cfbolz@hdlb-d9b95ee3.pool.mediaWays.net) left irc: "Verlassend"

[15:48] <stakkars> pedronis: I managed to get print to work in flowspace.

[15:49] <stakkars> after all, it was verysimple,and I'm wondering why we needed so long to do it.

[15:53] yuuh (gyxqmedk@i528C13CD.versanet.de) joined #pypy.

[16:04] jacob22|office (jacob@enzo.strakt.com) left irc: Remote closed the connection

----- silence for 30 minutes -----

[16:34] dialtone` (~user@host111-56.pool80117.interbusiness.it) joined #pypy.

[16:34] dialtone (~user@host111-56.pool80117.interbusiness.it) left irc: Nick collision from services.

[16:34] Nick change: dialtone` -> dialtone

[16:46] pedronis (pedronis@ratthing-b246.strakt.com) left irc: "Chatzilla 0.9.67 [Firefox 1.0.2/20050325]"

[16:48] arigo (~arigo@ratthing-b407.strakt.com) joined #pypy.

[16:49] lac-afk (lac@c-51c6e055.1321-1-64736c11.cust.bredbandsbolaget.se) left #pypy.

[17:04] stakkars (~tismer@i528C13CD.versanet.de) left irc: Read error: 113 (No route to host)

[17:16] Gromit (~bear@B74bc.b.pppool.de) joined #pypy.

----- silence for 27 minutes -----

[17:43] pedronis (pedronis@ratthing-b246.strakt.com) joined #pypy.

----- silence for 17 minutes -----

[18:00] <arigo> hi

[18:01] <arigo> what about some kind of 'demo' directory, for sharedref.py and some of our goal/*.py ?

[18:02] <arigo> maybe also containing scripts that call the translator and show nice flow graphs?

[18:07] Action: arigo suggests pypy/bin/demo/

[18:11] <hpk> probably more a toplevel demo directory, no?

[18:11] <hpk> pedronis: sorry, i am still not understanding: so we should only comment and not modify the release-announcement?

[18:11] <pedronis> hpk: review, edit whatever

[18:12] <hpk> ok, fine

[18:12] <arigo> hpk: for now we'll have executable demo files, mostly

[18:13] <arigo> but I don't mind pypy/demo/, probably better in the long run.

[18:13] <hpk> i thought dist/demo

[18:13] <arigo> ah

[18:13] <arigo> makes sense too

[18:13] <arigo> ok.

[18:15] Action: pedronis has a bit of head-ache

[18:16] Action: hpk would like to see if we can reuse arigo's upcoming mission-statement in the release announcement

[18:16] Action: arigo didn't start thinking about it so far

[18:16] <hpk> regarding release-0.6: i think we should mention that pypy 0.6 has much improved documentation at this point

[18:17] <hpk> and hint at the already-flexibility-showing points we have in getting-started

[18:18] <pedronis> there are hints at thunks and tracing, they may need more underlining

[18:19] <pedronis> it's also a matter the an announcemnt is quite long as it is

[18:25] <hpk> it could use sub-headers

[18:26] <hpk> http://codespeak.net/pypy/index.cgi?doc/release-0.6.html

[18:26] <hpk> is a way to access the web-page version already

----- silence for 23 minutes -----

[18:49] adim (~adim@logilab.net2.nerim.net) left irc: Remote closed the connection

[19:04] <hpk> do we explain anywhere what 'faking' is?

[19:04] <arigo> good question

[19:07] Action: hpk is currently annotating the directory structure and creating issues as he goes :-)

[19:08] <arigo> I'm about to check in the reorganisation of issue47

[19:09] <hpk> i shouldn't be affected much

[19:14] Action: arigo fixing getting_started for the new layout

[19:18] cfbolz (~cfbolz@hdlb-d9b95eb5.pool.mediaWays.net) joined #pypy.

[19:18] <cfbolz> arigo: do you want a.specialize not to be mentioned at all?

[19:18] <cfbolz> because it is in getting_started, too

[19:23] <arigo> right, I don't know exactly

[19:24] <hpk> getting consistent documentation is not much easier than getting consistent code actually :-)

[19:24] <arigo> no, I think getting_started.txt is fine

[19:24] <arigo> let's re-add a.specialize() to the docstring :-)

[19:24] <cfbolz> yes, I think so too. It's cool to see the int_add's

[19:27] cfbolz_ (~cfbolz@hdlb-d9b95ea0.pool.mediaWays.net) joined #pypy.

[19:31] cfbolz_ (~cfbolz@hdlb-d9b95ea0.pool.mediaWays.net) left irc: Client Quit

----- silence for 16 minutes -----

[19:47] cfbolz (~cfbolz@hdlb-d9b95eb5.pool.mediaWays.net) left irc: Read error: 113 (No route to host)

----- silence for 27 minutes -----

[20:14] pedronis (pedronis@ratthing-b246.strakt.com) left irc: Remote closed the connection

[20:17] pedronis (pedronis@ratthing-b246.strakt.com) joined #pypy.

[20:17] Gromit (~bear@B74bc.b.pppool.de) left irc: Read error: 110 (Connection timed out)

[20:29] <hpk> arigo: you broke revreport.py

[20:30] <hpk> hum, or is it a damn pyc file? lemme check

[20:30] <arigo> "interpreter/py.pyc"

[20:30] <hpk> i guess so

[20:30] <arigo> if we had a way to remove that one specifically in an "svn up" it would be great

[20:31] <hpk> i often wished for a local (python)-post-process script for svn-up

----- silence for 17 minutes -----

[20:48] <arigo> guess what I find in pypy.tool.opcode?

[20:48] <arigo> IEXEC_THROW, END_IEXEC, SETUP_IEXEC

[20:52] <hpk> upsie

[20:52] <hpk> on codespeak i guess

[20:53] <hpk> maybe i should implement them for PyPy to be done with that :-)

[20:53] <arigo> :-)

[20:53] <hpk> could be a use case for a flexible parser, actually

[20:54] <hpk> basically we could think about having a nice way to have various PEP 340 implementations

[20:54] <hpk> because that is something that is hard to do in CPython just so

[20:54] <hpk> would be a pretty cool sprint topic actually

[20:54] <hpk> (if the parser were far enough)

[20:55] Action: hpk is proud of: http://codespeak.net/pypy/index.cgi?doc/coding-guide.html#annotated-directory-structure-of-pypy

[20:55] <hpk> it looks like a simple thing but took me forever :-)

[20:56] <arigo> a directory verview :-)

[20:57] <arigo> it looks just great

[20:58] <hpk> i was thinking of linking this or even putting this in getting-started

[20:58] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

[21:00] <arigo> yes, I believe it would make more sense in getting_started

[21:00] <arigo> this said, "Annotated" has become a reserved keyword in PyPy :-)

[21:00] <hpk> hehe

[21:01] Action: hpk jumps in thinking: we should probably have an intermediate milestone M0.7 for EuroPython

[21:02] <hpk> aleale: you could help tomorrow if you started reading all documentation once we think we are basically finished :-)

[21:02] Action: hpk is getting even more jumpy

[21:12] <aleale> I am already reading it - just hard to keep up with the changes

[21:12] <hpk> yes, that's why i was suggesting that you read it when we are finished

[21:12] <hpk> you then have a fresh view

[21:12] <aleale> Ok

[21:13] <hpk> there are still some codecs related problems in the testreport i think

[21:13] <hpk> (linked from the 'doc' page)

[21:13] <aleale> Do you have time for a technical questions

[21:13] <hpk> sure (although i am currently busy checking stuff in before someone else commits :-)

[21:13] <aleale> regarding integrating unicode and codecs module

[21:15] <hpk> hua, i defer to arigo :-)

[21:15] <aleale> It seems that both me and arre is relying on each other to implement the basics of unichr and unicode

[21:16] <aleale> so when trying to merge I end up in endless recursions

[21:17] <aleale> by calling unichr -> which calls unicode -> encode -> which uses unichr

[21:18] <arigo> it's probably a question of level

[21:19] <arigo> there is no avoiding to use unichr() at interp-level, ultimately

[21:19] <pedronis> unichr should be a primitive and not call unicode btw

[21:20] <pedronis> yes, at interp level unichr is probably required

[21:20] <pedronis> but it's something the target should supply

[21:20] <aleale> target?

[21:21] <hpk> translation target i guess

[21:23] <hpk> pedronis: i'd try to avoid ReST-links as much as possible (in favour of plain http-links) for the release-0.6 document

[21:23] <pedronis> hpk: right

[21:24] <pedronis> I think arre implentation uses 1 unicode char as the primitive repr elements, these and a way to convert an integer to such a thing needs to be supplied

[21:25] <pedronis> by the translation target

[21:26] <aleale> unichr is defined in /builtin/app_misc.

[21:27] <aleale> How does unichr get information from the translation target

[21:27] <pedronis> that's the app-level unichr

[21:27] <aleale> Is this a necessary dependency?

[21:27] <pedronis> I was talking about the interp-level one

[21:28] <pedronis> it depends, if the final platform has a preferred way of representing unicode code-points the abstraction makes sense

[21:29] <hpk> arigo: was it on purpose that you indented "the py.py interpreter" in getting-started?

[21:30] <pedronis> the implementation in app_misc of unichr is not that much sane

[21:30] <arigo> it was on purpose that I did not dedent it, yes

[21:31] <arigo> it used to be a numbered list, so it was indented already

[21:31] <pedronis> it should probaly be implemented at app-level

[21:31] <pedronis> sorry interp-level

[21:32] <pedronis> this would remove the cirucularity too

[21:32] <pedronis> with encodings

[21:39] <hpk> arigo: you removed example_pytest.py instead of copying it to tool/pytest ?

[21:40] <arigo> yes, though I wasn't too sure about that

[21:40] <hpk> it's a good way to see if reporting of interp/app level tests works

[21:40] <arigo> ah

[21:40] <hpk> should be called pypy_test_demo.py or something

[21:41] <arigo> ok, bringing it back under that name...

[21:41] <hpk> pypy_test_failure_demo.py

[21:41] <hpk> even while you are at it

[21:42] <hpk> arigo: indentation: it doesn't look sensible in my browser (it's all italics)

[21:42] <arigo> oh, ok

[21:44] <arigo> sorry about example_pytest, I thought it was an old demo for how to write tests -- didn't realize they were all failing ones.

[21:44] <hpk> np

[21:44] <hpk> it's not that i am personally related to that file item somewhere in svn's revision tree :-)

[21:45] <hpk> hey, LLVM got a AMD64 backend!

[21:46] Action: hpk is reading the LLVM 1.5 announcement

[21:48] <hpk> it also seems to have a powerpc backend ...

[21:49] <hpk> the changenotes look pretty impressive

[21:49] <arigo> what exactly is translator/test/run_snippet.py ?

[21:51] <hpk> some arcane code that can go i think

[21:52] Action: hpk wishes for a characterlevel blame command

----- silence for 18 minutes -----

[22:10] <hpk> pedronis: i'd like to do some refactoring of release-0.6 at some point, probably tomorrow, if you don't mind

[22:11] <hpk> arigo: do you mind if i hack a bit on getting-started regarding indentation and link cleanup?

[22:11] <arigo> not at all

[22:11] <pedronis> hpk: that's ok

[22:13] <hpk> pedronis: i am probably going to try to beef it up a bit (but i'll try to hard not to raise wrong expectations)

[22:14] <pedronis> hpk: I see, remember that it is already quite long and people glaze over long announcement

[22:14] <hpk> pedronis: yes

[22:22] <arigo> grumble! pyc files.

[22:26] <arigo> we should patch the svn client to recognize

[22:26] <arigo> svn:kill *.pyc

[22:26] <arigo> properties

[22:26] <hpk> py.svn up ?

[22:26] <arigo> a variant of svn:ignore, more aggressive

[22:29] <arigo> if I had suggested a patch to CPython to disable all pyc file handling,

[22:29] <arigo> would you have answered: py.python ?

[22:29] <arigo> :-)

[22:30] <mwh> arigo: delete what you like

[22:30] <arigo> hi :-) thanks :-)

[22:31] <hpk> arigo: hehe, i was actually thinking of turning 'py' into a runner of python programs that can e.g. just have 'def main' among other niceties <hpk ducks and runs>

[22:31] <mwh> my email has been rather screwy of late

[22:31] <arigo> hpk: yes, and of course I was thinking of a site.py that inserts a custom import hook that completely replaces the built-in one <arigo ducks and run in the opposite direction>

[22:42] Action: pedronis leaving

[22:42] pedronis (pedronis@ratthing-b246.strakt.com) left irc: "Chatzilla 0.9.67 [Firefox 1.0.2/20050325]"

[22:42] Action: arigo leaving too

[22:43] arigo (~arigo@ratthing-b407.strakt.com) left irc: Remote closed the connection

----- silence for 38 minutes -----

[23:21] pedronis (~Samuele_P@c-3a8b70d5.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.

----- silence for 18 minutes -----

[23:39] Nick change: hpk -> hpk_away

[00:00] --- Thu May 19 2005