==== Channel ##pypy: 05/17/05 ====

[00:23] aleale (~chatzilla@0x535ca5a5.hrnxx9.adsl-dhcp.tele.dk) joined #pypy.

[00:33] <hpk> aleale: did you want to get auto-nosied on tracker issues as well?

[00:33] <aleale> Sure

[00:34] <aleale> But I already get a lot of issues sent to me ?

[00:34] <hpk> yes, you get top-level creations of issues

[00:34] <stakkars_> when they are generated first time

[00:35] <aleale> Ok but it is nice to get updated

[00:36] <aleale> Does anyone know if there is still work done on the non-fake-unicode branch?

[00:36] <aleale> or should I take a stab at it ?

[00:36] <hpk> aleale: so i add you to auto-nosy for the time being

[00:37] <hpk> aleale: talk to Arre, really

[00:37] <aleale> hpk: thanks

[00:37] <aleale> Ok

[00:38] <aleale> Will try to talk to himm in the morning

[00:38] <aleale> Good night

[00:38] <hpk> good night

[00:38] aleale (~chatzilla@0x535ca5a5.hrnxx9.adsl-dhcp.tele.dk) left irc: "Chatzilla 0.9.68a [Firefox 1.0.3/20050414]"

----- silence for 21 minutes -----

[00:59] cfbolz (~cfbolz@hdlb-d9b946cc.pool.mediaWays.net) joined #pypy.

[00:59] <cfbolz> hpk: still awake?

[00:59] <hpk> cfbolz: yip, barely

[00:59] <cfbolz> :-)

[00:59] <cfbolz> are you working on getting_started at the moment?

[00:59] <hpk> no

[00:59] <hpk> just finished

[01:00] <hpk> (i did a couple of checkins though)

[01:00] <cfbolz> otherwise I would try to do a round of improvements myself

[01:00] <hpk> great

[01:00] <cfbolz> yes, I saw these

[01:00] <cfbolz> I agree that the subversion stuff should go

[01:00] <cfbolz> (but I won't do that tonight)

[01:01] <cfbolz> I guess getting started is really one of the most important things to make the release a success

[01:01] <hpk> indeed

[01:02] <hpk> we should move the subversion chapter to 'svn_help.txt' and reference it from getting_started i guess

[01:03] <hpk> it contains a bunch of useful information that one cannot easily recover

[01:03] <hpk> i also thought about moving it to the coding-guide

[01:03] <cfbolz> yes, but it's not really appropriate anymore in getting_started since we don't expect the majority of the people reading getting_started to become contributors

[01:04] <hpk> true

[01:04] <hpk> just go ahead and refactor things

[01:04] <hpk> i guess i am going to sleep and will continue tomorrow

[01:05] <cfbolz> sleep well!

[01:06] <cfbolz> bye

[01:06] cfbolz (~cfbolz@hdlb-d9b946cc.pool.mediaWays.net) left irc: "Verlassend"

----- silence for 1 hr and 39 minutes -----

[02:45] dialtone (~dialtone@host111-56.pool80117.interbusiness.it) left irc: "Leaving"

[02:54] dialtone (~dialtone@host111-56.pool80117.interbusiness.it) joined #pypy.

----- silence for 5 hr and 11 minutes -----

[08:05] stakkars_ (yolexnp@dsl-62-220-13-97.berlikomm.net) left irc: Read error: 60 (Operation timed out)

----- silence for 1 hr and 12 minutes -----

[09:17] arre (ac@ratthing-b3fa.strakt.com) joined #pypy.

[09:23] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

[09:23] <aleale> hi

[09:24] <hpk> aleale: morning (i am a bit off-keyboard currently)

[09:24] <aleale> arre: can I do something to help out with the non-faked-unicode branch ?

----- silence for 50 minutes -----

[10:14] adim (~adim@logilab.net2.nerim.net) joined #pypy.

[10:22] Gromit (~bear@B7494.b.pppool.de) joined #pypy.

[10:33] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "Eject! Eject! Eject!"

[10:35] dialtone (~dialtone@host111-56.pool80117.interbusiness.it) left irc: "Leaving"

[10:36] dialtone (~user@host111-56.pool80117.interbusiness.it) joined #pypy.

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

[10:53] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

[10:59] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) joined #pypy.

[10:59] <cfbolz> hi!

[11:01] <aleale> hi

----- silence for 21 minutes -----

[11:22] mwh (~user@modem-35.argonath.dialup.pol.co.uk) joined #pypy.

[11:31] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) left irc: "Leaving"

[11:35] sanxiyn (~tinuviel@211.104.100.115) joined #pypy.

[11:35] cfbolz (~bolz@zenon.physi.uni-heidelberg.de) joined #pypy.

[11:36] <sanxiyn> Hello

[11:36] <cfbolz> hi!

[11:36] mwh (~user@modem-35.argonath.dialup.pol.co.uk) left irc: "bye"

[11:42] arigo (~arigo@ratthing-b407.strakt.com) joined #pypy.

[11:46] <cfbolz> hi!

[11:47] <arigo> hi!

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

[12:04] <cfbolz> should we put a link to the irc logs on our contact page?

[12:05] <hpk> cfbolz: yes

[12:06] <cfbolz> ok.

[12:06] <cfbolz> sorry for the dead link yesterday

[12:07] <hpk> np

[12:07] <hpk> but you didn't run the tests, did you?

[12:08] Gromit (~bear@B7494.b.pppool.de) left irc: Read error: 104 (Connection reset by peer)

[12:08] <cfbolz> no, I did run a py.test -R and didn't see anything

[12:08] <hpk> ups, strange

[12:08] <cfbolz> but it could have been my tired eyes

[12:08] <hpk> :-)

[12:08] <cfbolz> don't know exactly

[12:08] <hpk> hi all btw :-)

[12:08] <cfbolz> what does the additional -v do ?

[12:09] <hpk> verbose mode, you then see which links are currently being checked

[12:09] <cfbolz> ah, cool (and kind of obvious :-) )

[12:13] <cfbolz> btw: the link pliant.cx in index.txt doesn't seem to be reachable

[12:16] <arigo> it's pingable, so maybe it's only temporary

[12:16] idnar (mithrandi@idnar.user) left irc: Read error: 60 (Operation timed out)

[12:18] <cfbolz> ok

[12:18] idnar (mithrandi@idnar.user) joined #pypy.

[12:22] <arigo> may I raise again the question of what exactly we target in the 0.6 release?

[12:23] <arigo> I'm confused now if things like "make 'print' statements work in RPython for user programs" go in 0.6, or are postponed.

[12:24] <cfbolz> postponed, I'd say

[12:24] <cfbolz> There's really some work to be done in documentation I think

[12:25] <arigo> ok

[12:27] <cfbolz> hey, that's just *my* opinion :-)

[12:27] <cfbolz> maybe everyone else disagrees

[12:34] <arigo> hpk: any chance at all that having a different password for the issue tracker is easy?

[12:34] <hpk> arigo: did you try?

[12:34] <hpk> i think i hacked things so that it should be possible

[12:35] <hpk> although i am not entirely sure it's a good idea

[12:35] <hpk> basically _both_ passwords will probably work

[12:35] <arigo> links is giving me troubles, I want to be able to start a new links in a single keypress and be logged in

[12:36] <arigo> I can do that by encoding the name and password in the URL

[12:37] <arigo> works, thanks!

[12:38] <hpk> so you are having two passwords now, right?

[12:38] <arigo> yes, both work.

[12:40] <hpk> hum, i am guessin this will break at some point.

[12:40] <hpk> (the next time anyone changes a password on codespeak in which case things might get reset)

[12:40] <hpk> i am trying to prevent this now, so file a bug if it does still break

[12:40] <arigo> ok

[12:41] <aleale> Is arre online ?

[12:42] <hpk> arigo, cfbolz: i think we should go through all issues at some point and decide about M0.5 <-> M1.0

[12:42] <hpk> and assign people to each of them

[12:42] <cfbolz> now?

[12:42] <hpk> maybe it's good if christian and samuele were there too

[12:42] <cfbolz> right

[12:43] <hpk> or we simply assign everything to them and go home

[12:43] <cfbolz> good idea

[12:44] <hpk> arigo: did you notice that i assigned the front-page issue to you?

[12:44] <arigo> (sorry back in a min)

[12:44] arigo (~arigo@ratthing-b407.strakt.com) left irc: "[x]chat"

[12:46] <cfbolz> I'm getting some errors when trying the PyPy tests with Python2.4

[12:46] <cfbolz> module/parser/test/test_parser.py FAILED TO LOAD MODULE

[12:46] <cfbolz> there are more of these (failed to load...)

[12:47] arigo (~arigo@ratthing-b407.strakt.com) joined #pypy.

[12:48] <arigo> hpk: yes

[12:49] <hpk> cfbolz: can you see if these failures are shallow?

[12:49] <hpk> i have the feeling that trying to get python2.4 work completely for M0.5 might be too much

[12:49] <cfbolz> yes.

[12:50] <arigo> I don't think there is any serious problem with Python 2.4

[12:50] <cfbolz> wait. at least some of them are due to missing distutils --> suse problem

[12:50] <hpk> cfbolz: again a suse or debian machine? :-)

[12:50] <hpk> arigo: probably not, but having all tests pass and all translator examples work might be problematic

[12:50] <cfbolz> yes, the institute uses only suse

[12:50] <hpk> cfbolz: then you have to install some python-development package or something

[12:50] <hpk> likely

[12:51] <cfbolz> yes, I'm doing that at the moment

[12:55] Action: hpk -> lunch

[12:59] <arigo> hpk: I see only one extra test failure with 2.4, in recparser's test_samples.py

[12:59] <cfbolz> me too after having installed distutils

[12:59] <arigo> I don't really see why it would make the translator unhappy

[13:00] <arigo> it's somehow the whole point of PyPy to be able to emulate one specific version of Python over a range of versions

[13:00] <cfbolz> yes

[13:00] <arigo> in my opinion it's barely ok to drop 2.2 support, but not 2.4 and 2.5 support...

[13:00] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "Eject! Eject! Eject!"

[13:00] <sanxiyn> arigo: I think it's perfectly okay to drop 2.2 support.

[13:00] <arigo> sanxiyn: hi :-) ok

[13:00] <cfbolz> Should I try to run CPython's tests against CPython2.4 also?

[13:01] <cfbolz> using PyPy of course

[13:01] <arigo> maybe -- that would let us now at least if ignoring _sre.MAGIC works :-)

[13:01] <arigo> s/now/know

[13:01] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

[13:01] Action: arigo -> lunch too

[13:01] arigo (~arigo@ratthing-b407.strakt.com) left irc: "[x]chat"

[13:02] <sanxiyn> cfbolz: Hm, how do I do that?

[13:02] <sanxiyn> (py.test option?)

[13:02] <cfbolz> I just do python2.4 test_all.py .....

[13:02] <sanxiyn> Ah, sure.

[13:02] <sanxiyn> ;;;

[13:03] <cfbolz> arigo: it's somehow troubling that there is no PyPy test that finds the sre problem

[13:14] <cfbolz> sanxyin: are you running the CPython regression tests? could you check in your testresult directory afterwards (I can't do it here at the moment)?

[13:15] <sanxiyn> cfbolz: With 2.4?

[13:15] <cfbolz> yes

[13:15] <sanxiyn> I can fire test now, but this is a slow machine.

[13:15] <cfbolz> ok. Don't do it if it's too much trouble.

[13:16] <cfbolz> I can do it tommorrow or something like that

----- silence for 28 minutes -----

[13:44] yuuh (~yuuhu@i3ED6B51A.versanet.de) joined #pypy.

----- silence for 24 minutes -----

[14:08] <cfbolz> I'm going home. see you all

[14:08] mwh (~user@pc150.maths.bris.ac.uk) joined #pypy.

[14:08] arigo (~arigo@ratthing-b407.strakt.com) joined #pypy.

[14:09] <arigo> (dropping in for a planning note: Samuele is at the company meeting, and I'm out for fresh air; we should be back in an hour or so)

[14:09] arigo (~arigo@ratthing-b407.strakt.com) left irc: Remote closed the connection

[14:10] <cfbolz> (I will try to be there in 1.5 hours. See you)

[14:10] cfbolz (~bolz@zenon.physi.uni-heidelberg.de) left irc: "Leaving"

[14:12] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "Eject! Eject! Eject!"

[14:18] sanxiyn (tinuviel@211.104.100.115) left #pypy ("Bye").

----- silence for 25 minutes -----

[14:43] stakkars (bygjng@i3ED6B51A.versanet.de) joined #pypy.

[14:46] Gromit (~bear@B74d4.b.pppool.de) joined #pypy.

[14:49] pedronis (pedronis@ratthing-b246.strakt.com) joined #pypy.

[14:51] arigo (~arigo@ratthing-b407.strakt.com) joined #pypy.

[15:02] redorlik (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

[15:04] <hpk> arigo: so you are making a distinction between current and M0.5 as it seems

[15:05] <hpk> might make sense although i would probably put it in M1.0

[15:10] cfbolz (~cfbolz@hdlb-d9b946c1.pool.mediaWays.net) joined #pypy.

[15:13] redorlik (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "Eject! Eject! Eject!"

[15:19] <arigo> yes I'm thinking "current" == "unclassified"

[15:19] <hpk> ok, i guess i am going to modify the tracker webpage slightly then

[15:20] <hpk> to have an item 'show M0.5' issues and 'show current' or something

[15:20] <hpk> i'd like to have a focus on M0.5 issues

[15:20] <arigo> the default view is ok enough, I think

[15:21] <arigo> maybe renaming "current" to "unclassified" would make it at the end of the alphabet and the web page, and be fine for now

[15:21] <cfbolz> yes but you have to know that current does not mean as fast as possible

[15:21] <hpk> cfbolz: exactly

[15:21] <arigo> sure, I just mean we can already use views and queries etc.; the default view is good (after renaming)

[15:22] <cfbolz> of course

[15:22] <hpk> ok, i try that at some point (unfortunately roundup has a pretty 'distributed configuration')

[15:23] <cfbolz> I hate that in programs

[15:23] <hpk> web apps often have this problem

[15:23] <cfbolz> yep

[15:23] <hpk> but you can only rewrite that much softare at a time :-)

[15:23] <cfbolz> :-)

[15:24] <arigo> :-)

[15:24] <cfbolz> should I mark Toplevel README (issue5) as closed now that getting_started is getting improved?

[15:24] <hpk> i thought about this

[15:24] <hpk> not sure that we wouldn't want a toplevel README in pypy/dist

[15:24] <hpk> anyway

[15:24] <cfbolz> but mainly as a pointer to getting_started?

[15:25] <hpk> not sure, but a few pointers into the documentation, yes

[15:25] <hpk> (i don't like it if a README from a just downloaded package starts off telling how to download the package)

[15:25] Gromit (~bear@B74d4.b.pppool.de) left irc: Read error: 110 (Connection timed out)

[15:26] <cfbolz> I agree

[15:26] <hpk> basically i'd like the README to contain the Mission statement, the release it belongs to and further reading

[15:27] <hpk> let's extend the README to 'add README and LICENSE'

[15:27] <hpk> because it's also interesting to have a copy of the license (MIT-license)

[15:27] <cfbolz> absolutely. Are you doing it (or should I)?

[15:27] <hpk> question is: who is the copyright holder?

[15:28] <stakkars> hpk: oops, I marked an issue as resolved, but this was an error. How can I find it again?

[15:28] <hpk> enter the issue number for example

[15:29] <hpk> (you can also search for it if you have no clue about the issue number)

[15:30] <stakkars> thanks

[15:32] <hpk> stakkars: so you think that supporting print statements is neccessary for M0.5?

[15:33] <hpk> (i mean RPython-level ones, of course)

[15:33] <cfbolz> no, lets not support print at all. Nobody needs that anyway

[15:34] <cfbolz> :-)

[15:34] <hpk> the "print sucks" release :-)

[15:37] <cfbolz> aren't we all here now? should we just go through all M0.5 issues and decide what to do with them?

[15:37] <hpk> makes sense to me

[15:38] <cfbolz> since the others aren't objecting they can't complain later :-)

[15:38] <hpk> i'd like them to participate, actually :-)

[15:38] <cfbolz> that's what the smiley was supposed to mean

[15:38] <hpk> arigo, pedronis, stakkars: ok to have a round of looking at all M0.5 issues?

[15:38] <hpk> cfbolz: sorry about that, i am missing that always

[15:39] <hpk> cfbolz: i am more the serious guy, y'know

[15:39] <hpk> cfbolz: :-) of course

[15:39] <cfbolz> hpk: I wouldn't have believed you even without the smiley :-)

[15:40] Gromit (~bear@B74e0.b.pppool.de) joined #pypy.

[15:41] stakkars (bygjng@i3ED6B51A.versanet.de) left irc: Read error: 148 (No route to host)

[15:43] <cfbolz> btw: python2.4 seems to be no real problem after I installed distutils

[15:55] <arigo> hpk: back, sorry, yes, let's assign and make more precise our M0.5 issues...

[15:57] <hpk> ok

[15:57] <hpk> let's first look through current, though

[15:58] <hpk> and consider current > M0.5

[15:58] <hpk> *flowgraph viewing*

[15:59] <hpk> this is not M0.5 is it?

[16:03] <arigo> no

[16:04] <hpk> ok, let's just talk about the ones that could be M0.5

[16:04] <hpk> have a doc chapter for CPython-dev'ers?

[16:04] <arigo> maybe

[16:05] <arigo> "We basically have information here"

[16:05] <arigo> where?

[16:05] <arigo> we don't compare that much

[16:06] <hpk> not really compare but we relate to it

[16:06] <hpk> e.g. in architecture.txt "It helps to remember how CPython deals with the same issue ..."

[16:06] Gromit (~bear@B74e0.b.pppool.de) left irc: Read error: 60 (Operation timed out)

[16:06] <arigo> well we say here and there "remember RPython is like C in CPython"

[16:07] <hpk> yes, we don't go to the frame/bytecode level etc.pp. although that is rather easy

[16:07] <arigo> that's not really relating to the CPython implementation, in itself

[16:07] <hpk> ok

[16:07] <hpk> anyway, i think writing such a chapter would be a worthwhile addition to the documentation

[16:07] <arigo> yes, definitely

[16:07] <hpk> because i expect that many cpython-dev'ers would be tempted to read it

[16:07] <arigo> I'm just not sure how much efforts it requires

[16:08] <hpk> i would say, about 1-2 hours for something helpful

[16:08] <arigo> maaaaaybe

[16:08] <hpk> just pointing pyopcode.py and pyframe and how ceval<->pypy relate would be already helpful

[16:09] <arigo> that's the very tip of the iceberg of differences below the water surface...

[16:09] <hpk> sure, i'd say focus on the common-ness instead :-)

[16:09] <arigo> makes sense, I guess

[16:09] <hpk> of course, one can get arbitrarily detailed and write a whole book about it

[16:10] <arigo> doesn't make sense now

[16:10] <hpk> the main difference is: PyPy has more documentation :-)

[16:10] <arigo> i still have that feeling about how CPython developers consider us,

[16:11] <arigo> that we're just too abstract for now

[16:11] <hpk> yes, i agree

[16:11] <hpk> don't you think that such a chapter could help to show that it's more concrete?

[16:12] <arigo> no

[16:12] <arigo> I think we won't attract more "US python-dev" kind of people before we have a good translation, and even then, probably not before we are really holding our promises of doing better than CPython in some major ways

[16:12] <arigo> until then we're just a fine useless experiment

[16:12] <hpk> you are probably right

[16:13] <arigo> I'm not saying we shouldn't have such a chapter, just that it probably won't attract much people by itself now.

[16:13] <hpk> If such a chapter gets us someone like Carl then writing it is well worth it :-)

[16:14] <arigo> ?

[16:14] cfbolz (~cfbolz@hdlb-d9b946c1.pool.mediaWays.net) left irc: Read error: 113 (No route to host)

[16:14] <hpk> arigo: i mean if such a chapter interests someone

[16:14] Gromit (~bear@A36dc.a.pppool.de) joined #pypy.

[16:14] <hpk> enough to get starting playing around

[16:15] <hpk> not making sense still?

[16:15] <arigo> the good persons don't really need that -- Carl witness :-)

[16:15] <hpk> hehe, good point :-)

[16:15] cfbolz (~cfbolz@hdlb-d9b946e7.pool.mediaWays.net) joined #pypy.

[16:16] <hpk> OTOH, Michael Chermside did read the documentation before coming to the Pycon sprint

[16:16] <arigo> well, somehow it even makes sense to remember not to be *too* beta-user-friendly at this point

[16:16] <arigo> the goal is still to interest people that really want to play around, not just try it and run away because it's so slooooow

[16:17] <hpk> ok, i am convinced to forget it for M0.5

[16:17] <hpk> app-level docstrings?

[16:17] <arigo> no

[16:18] <arigo> that's a good introduction job for a would-be PyPy developer :-)

[16:18] <arigo> (including thinking about how to do it half-automatically)

[16:18] Action: hpk notices that having an 'easy/medium/hard' qualification on the issues would be nice

[16:18] <arigo> right

[16:18] <hpk> arigo: ok, then

[16:18] <hpk> current seems to be all non-M0.5 related then

[16:19] <hpk> (dictionary iterators are not M0.5, are they?)

[16:19] <arigo> (no)

[16:23] <cfbolz> on to the M0.5 issues then?

[16:23] <arigo> yes

[16:23] <hpk> yip

[16:23] <hpk> clean up and document files

[16:23] <hpk> i think we should go for Armin's suggestion to have some place that describes all the directories

[16:23] <hpk> (or was it you, carl? anyway)

[16:24] <cfbolz> both

[16:24] <cfbolz> I think so too

[16:24] <hpk> so we should open a new issue accordingly

[16:24] <hpk> and move that one to M1.0

[16:25] <cfbolz> yes. We can't really clean up everything and a description of dirs is important

[16:25] <arigo> ok

[16:25] Action: hpk is doing that

[16:25] <hpk> you can continue ...

[16:26] <arigo> cleaning up "tool": should probably be done now

[16:26] <arigo> but how?

[16:27] <cfbolz> how about a bin directory or similar

[16:27] <arigo> at least to regroup all our executable scripts?

[16:27] <cfbolz> yes

[16:28] <cfbolz> I think py.py doesn't really belong to interpreter

[16:28] <arigo> yes

[16:29] <cfbolz> so we rename the issue?

[16:30] <arigo> well

[16:30] <arigo> there is a lot of strange stuff in "tool"

[16:30] <arigo> mostly ONLY strange stuff

[16:31] <arigo> maybe we can go over the list, adds a comment to what it's used for

[16:31] <cfbolz> what list?

[16:31] <cfbolz> the files in tool?

[16:31] <arigo> yes

[16:32] <arigo> half of them are executables, but definitely not the kind we *want* people to use

[16:32] <hpk> the only easy enough fix i can think of is

[16:32] <hpk> to put things in subdirs in tool

[16:33] <hpk> fix == slight cleanup

[16:33] <cfbolz> yes, but I think a note at the top of each file would also help

[16:33] <arigo> definitely

[16:34] <hpk> ok, then who to assign?

[16:35] <cfbolz> not me. I find the tool directory very mistifying myself :-)

[16:35] <hpk> arigo: you start off and hand it over to me or someone else at some point?

[16:35] <arigo> ok

[16:36] <arigo> I'm minded to merge some files (if not their functionality for now), like compile.py template.py sourcetools.py

[16:36] <arigo> and hack.py

[16:36] <hpk> probably, yes

[16:37] <hpk> just try it and run tests (i really hope all those tools are taking part in testing somehow)

[16:37] Action: hpk assigned the issue to arigo for now, let's move on

[16:37] <arigo> ok

[16:37] <cfbolz> finalizing getting-started document

[16:38] <hpk> i'd like to keep this open a bit still

[16:38] <cfbolz> yes. me too. But there has been some good progress

[16:38] <hpk> yes, i still want to do a few minor things though

[16:38] <cfbolz> ok. I'm going to continually look through it

[16:39] <hpk> cfbolz: can you add a good link for installing LLVM?

[16:39] <hpk> just pointing at the home page is lazy :-)

[16:40] <hpk> http://llvm.cs.uiuc.edu/docs/GettingStarted.html

[16:40] <hpk> for example

[16:40] <cfbolz> ok. I can do that.

[16:41] <hpk> maybe also hint at platforms where you know it's more difficult or plain not working

[16:41] <cfbolz> Yes. Windows seems to be difficult

[16:41] <cfbolz> (veeeeery difficult)

[16:41] <hpk> ick

[16:41] <hpk> even just the minimal backend?

[16:41] <cfbolz> well. We always say it's experimental

[16:42] <cfbolz> ah, no. you're right

[16:42] <cfbolz> genllvm should work without the C front end

[16:42] <hpk> rigt, so just hint that on win32 people should install only care about the LLVM backend part

[16:42] <cfbolz> ok. And I'll write a warning that it might be difficult

[16:42] <hpk> and a mailing list where they can complain :-)

[16:43] <hpk> ok, let's move on

[16:43] <cfbolz> LLVM-dev is very helpful so that's a good idea.

[16:43] <cfbolz> ok

[16:43] <hpk> support print statements in RPython

[16:43] <cfbolz> support print statements in RPython

[16:43] <hpk> support print statements in RPython

[16:43] <cfbolz> is stakkars here?

[16:43] <arigo> apparently not

[16:44] <hpk> why would it be neccessary for M0.5, to prevent people from trying the translator with simple snippets and failing?

[16:44] <arigo> yes, that's the idea

[16:45] <arigo> the problem is that it means more strange hacks

[16:45] <cfbolz> but we could just clearly say that print is not supported

[16:45] <hpk> can't we just add a proper prominent warning

[16:45] <cfbolz> :-)

[16:45] <hpk> at the place where we explain how to use it?

[16:45] <hpk> :-)

[16:45] <arigo> yes

[16:46] <hpk> ok, side note: can i file a new issue 'remove Pyrex' references, including the top-level Pyrex directory?

[16:46] <arigo> I think I'll try to discuss this with Christian in the issue of prints

[16:46] <arigo> in the issue tracker I mean

[16:47] <hpk> ok

[16:47] <arigo> I don't think we should remove Pyrex now

[16:47] <cfbolz> but at least getting_started should probably not advertise it as the main backend

[16:47] <arigo> yes

[16:48] <hpk> yes

[16:48] <hpk> that was my starting point for the suggestion

[16:48] <arigo> it should be un-documented, probably

[16:48] <hpk> arigo: but removing pyrex is pretty easy

[16:48] <arigo> well, keeping it is pretty easy too

[16:48] <hpk> i don't see why we want to continue having it around at top-level and all that

[16:48] tremolo (~tremolo@82.144.60.19) joined #pypy.

[16:48] <hpk> arigo: reducing/preventing confusion

[16:48] <cfbolz> ok. I can do it when I add the LLVM links to getting_started.

[16:49] <hpk> cfbolz: ok, assigned the issue to you

[16:49] <cfbolz> ok

[16:49] <hpk> just assign it back to me when you are done

[16:49] <arigo> we can move all Pyrex-related stuff to its own subdir of translator

[16:49] <hpk> that is more work than just getting rid of it once and for all

[16:50] <cfbolz> actually, genllvm uses pyrex

[16:50] <hpk> in which way?

[16:50] <cfbolz> to create the wrapper around the LLVM function

[16:51] <hpk> oh

[16:51] <cfbolz> that probably has to be fixed anyway.

[16:51] <hpk> but then i agree that it is not a M0.5 issue, i guess

[16:51] <cfbolz> yes

[16:51] <arigo> yes, for now it's fine to have "external dependencies" like Pyrex and the py lib at the same level

[16:51] <hpk> should we still move Pyrex out of top level?

[16:53] <arigo> I can try to do that and fix the relevant places

[16:53] <hpk> ok

[16:53] <cfbolz> hum. I guess build_llvm_module.py could be problematic

[16:54] <cfbolz> I can try to fix that tonight, but no guarantee

[16:54] <hpk> don'T worry too much, let's just try to shift Pyrex (and forget all about it if it's too complicated)

[16:54] <arigo> hpk: please add an issue about Pyrex, as you intended to, and assign to me for now

[16:55] <cfbolz> ok. next one?

[16:55] <cfbolz> CPython 2.4 compatibility

[16:56] <cfbolz> I guess that looks pretty good now, I'm running the tests and have not encountered problems

[16:56] <cfbolz> the magic number in SRE seems to be the biggest problem.

[16:56] <cfbolz> and that's fixed now

[16:57] <pedronis> recently I have always used 2.4 on my laptop with PyPy

[16:57] <pedronis> the only issue was the sre magic number

[16:57] <cfbolz> good. So I'm closing it?

[16:58] <hpk> yes

[16:58] <cfbolz> ok.

[16:58] <pedronis> if the 2.4 re compiled form is backward compatible with 2.3 one, things should work

[16:58] <pedronis> the only way to know is to run some related CPython tests on 2.4

[16:59] <cfbolz> How would I do that?

[16:59] <cfbolz> just try python24 test_blabla?

[16:59] <hpk> python pypy test_blablabla

[16:59] <hpk> py.py actually

[17:00] <cfbolz> well, I'm running the tests with py.test under CPython2.4 allready. Is that enough?

[17:00] <hpk> compliance tests or pypy tests?

[17:00] <cfbolz> meaning CPython's regression tests

[17:00] <hpk> that should be enough, yes

[17:00] <cfbolz> pypy tests are ok, too

[17:00] <hpk> i think that's good enough for closing the issue for now

[17:00] <cfbolz> ok. doing that.

[17:01] <hpk> we will likely be moving to python2.4 wholesale in the near term, anyway

[17:01] <hpk> *Refactor PyPy's entry points*

[17:04] <pedronis> mmh

[17:04] <cfbolz> we need at least documentation for the tracing.

[17:04] <arigo> __pytrace__ = 1

[17:04] <pedronis> translator docstring refers to accessing cl and pyrex backends only

[17:04] <cfbolz> I can fix that, too.

[17:04] <arigo> pedronis: I'll fix that while moving Pyrex somewhere else

[17:05] <cfbolz> ok

[17:05] <arigo> py.py docs should also mention Ctrl-C --> interplevel console

[17:05] <cfbolz> should I write about pytrace and ctrl-c in getting-started?

[17:05] <hpk> arigo, cfbolz: merged yes

[17:05] <cfbolz> hu?

[17:06] <hpk> yes to both

[17:06] <cfbolz> ok

[17:06] <hpk> do we leave py.py in pypy/interpreter?

[17:07] <arigo> make a dist/pypy/bin directory?

[17:07] <cfbolz> I would say no

[17:07] <cfbolz> arigo: yes

[17:07] <hpk> what else do we put there if anything?

[17:08] <arigo> the interactive part of translator.py

[17:08] <arigo> ?

[17:08] Gromit (~bear@A36dc.a.pppool.de) left irc: Read error: 104 (Connection reset by peer)

[17:08] <hpk> makes some sense, we have to grep through the documentation to eliminate wrong refs then

[17:09] <arigo> yes

[17:09] <hpk> pedronis: would you like to get an assignment? :-)

[17:09] <arigo> I'm fine with that

[17:10] <hpk> you mean doing it?

[17:10] <arigo> yes

[17:10] <arigo> that's somehow part too of cleaning up pypy/tool

[17:10] <cfbolz> yes

[17:10] <hpk> ok, then create an issue for that, anyway

[17:10] <arigo> why not use issue "Refactor PyPy's entry points?"

[17:11] <pedronis> what do we do about goal dir?

[17:11] <hpk> pedronis: good point

[17:11] <cfbolz> arigo: if you create the interactive part of translator.py, would you import genllvm.llvmcompile, too?

[17:11] <arigo> indeed

[17:11] <hpk> arigo: good point

[17:11] <arigo> cfbolz: right, yes, we should

[17:11] <arigo> cfbolz: btw these import statements in translator.py should be done lazily

[17:11] <cfbolz> I don't want to make it part of the translator, though

[17:11] <hpk> arigo: maybe separate out your recent suggestion to the entry points (unless you can quicklky hack it)

[17:12] <cfbolz> arigo: which imports?

[17:12] <arigo> cfbolz: translator.py (it's already interactive)

[17:12] <cfbolz> sorry, I don't get it.

[17:13] Action: hpk wistles something about lazy imports

[17:13] <arigo> cfbolz: translator.py imports genc, genpyrex, etc. just to implement t.pyrex(), t.ccompile(), etc

[17:13] <cfbolz> okay

[17:13] <arigo> cfbolz: I'm just suggesting to make these imports local to the t.pyrex(), t.ccompile() etc functions

[17:13] <arigo> cfbolz: and then add a t.llvmcompile()

[17:14] <cfbolz> aaah. Makes sense. The problem is just that you can't really call t.llvmcompile twice on the same translator

[17:14] <arigo> I think genc has the same problem, actually :-)

[17:14] <cfbolz> :-)

[17:14] <arigo> you're supposed to fill the translator with functions and annotations, and then only call ccompile() once at the end, basically.

[17:14] <cfbolz> ok, then it makes sense to add it.

[17:15] <cfbolz> (it == llvmcompile)

[17:15] <arigo> ok

[17:15] <pedronis> well, althuogh a check about double calling is probably required

[17:15] <hpk> can it be made so that is refuses early to double-compile?

[17:16] <cfbolz> wait. I can make doublecompiling safe.

[17:16] <cfbolz> I'll create an issue.

[17:16] <pedronis> hpk: yes, we want something like that because the error you get with genc I think are pretty confusing

[17:16] <pedronis> (I mean if you try to compile a 2nd time)

[17:16] <arigo> ok, will try that...

[17:16] <hpk> arigo: do you put that into your refactor PyPy's entry points issue? (that i just assigned to you)

[17:17] <cfbolz> I'm creating a fix translator issue and put it in there

[17:17] <cfbolz> ok?

[17:18] <hpk> just the LLVM part for now

[17:18] <cfbolz> ok

[17:18] <hpk> genc can fail early instead i'd say

[17:18] <hpk> ok, can we continue with the list?

[17:18] <arigo> yes

[17:18] <hpk> roadmap

[17:19] <hpk> i'd like to add that to the documentation pages (also on the left navigation side)

[17:19] <hpk> fine for M0.5?

[17:20] <pedronis> don't know, I think it can be postponed for around europython (having a roadmap)

[17:21] <hpk> i think we have the roadmap pretty fixed because of the EU project and it's interesting information (and not too hard to infer from the EU proposal)

[17:21] <cfbolz> I agree

[17:21] <cfbolz> It's really of interest for people downloading the release: when can we expect this to be fast, etc.

[17:23] <hpk> pedronis: or do you think it's too early for a roadmap?

[17:23] <pedronis> maybe, the fact that we ard funded does not make people believe in our claims

[17:23] <cfbolz> I

[17:23] <arigo> I tend to agree with Carl on this point

[17:24] <cfbolz> pedronis: I'm not too sure about that: you hear "maybe PyPy can do XXX, they got EU funding" quite regularly

[17:24] <arigo> this document should not mention funding or the EU at all

[17:24] <arigo> ah

[17:24] <cfbolz> that wasn't my point

[17:24] <arigo> (hpk: svn down...)

[17:24] <pedronis> it depends, people have various level of tolerance to vapourware announcement

[17:25] <pedronis> various -> varying

[17:26] <cfbolz> the roadmap would just be some kind of "we expect to be there then and then..."

[17:26] <hpk> (arigo: fixed)

[17:27] <arigo> (thanks)

[17:27] <hpk> pedronis: i see your point but i doubt that not having a roadmap at all improves the perception

[17:27] <arigo> (hpk: issue tracker down :-( )

[17:28] <pedronis> my point is that a roadmap talking about 2.0 make more sense around europython/ july

[17:28] <pedronis> right we should do expectations management

[17:28] <hpk> (arigo: fixed, damn it :-)

[17:28] <pedronis> that means attest that 0.5 is slow but

[17:28] <pedronis> and say that we are working on full translation etc and we are able to annotate all of PyPy

[17:28] <hpk> pedronis: this is all good release announcement stuff

[17:30] <pedronis> hpk: yes, I don't think right now we need a roadmap, I think we just have to do good expectations management in the announcement etc

[17:30] <arigo> would it make sense to have a top-level link naively called "download" that points only to (1) the release annoucement with this information, (2) getting_started, (3) zip/tgz files

[17:31] <arigo> i.e. use this proeminently placed release annoucement instead of having something called "road map" for now?

[17:31] <hpk> pedronis: works for me if you insist on it :-)

[17:31] <hpk> arigo: carl made the good point that 'download' leads people to assume you can use it for productive usage

[17:31] <hpk> getting-started reflects better the current state -> expectation management

[17:32] <arigo> ok, but it would be good to have some kind of "release 0.6" page easily available too

[17:32] <hpk> the front page + refering the full release announcement from the main doc-page?

[17:33] <hpk> pedronis: would you be ok to write the release announcement, btw?

[17:33] <arigo> the release annoucement itself is fine for that role, yes

[17:34] <cfbolz> and maybe adding a sentence to getting_started (where we talk about the slowness)

[17:35] <pedronis> hpk: I could start a draft but I think we should really all agree and maybe do a screen session to finalize it

[17:35] <hpk> pedronis: just start off and let us all freely improve it (if necessary :-)

[17:36] <hpk> pedronis: ok?

[17:36] <hpk> (i am not too fond of screen sessions for writing documentation/prosaic texts)

[17:37] <pedronis> I see, well assign it to me and can try

[17:37] <hpk> great, doing it

[17:39] <hpk> *architecture.txt refactoring*

[17:39] <arigo> I guess that's the piece of documentation I'd me most easily convinceable to do,

[17:39] <hpk> arigo: fine

[17:39] <arigo> but I don't want to steal interesting bits from other people if they want to have it :-)

[17:40] <cfbolz> I knew there was a but coming

[17:40] <cfbolz> :-)

[17:40] <hpk> arigo: linking it a bit more to the rest of the documentation would be nice, as well

[17:40] <arigo> it's a positive "but" :-)

[17:41] <arigo> yes

[17:41] <hpk> although i generally recommend to keep an eye on the "levels of links", i.e.

[17:41] <hpk> not mixing links to low level details and high level views without making it obvious.

[17:42] <arigo> makes sense

[17:42] <hpk> ok, assigning it to you, you could set it to testing if you are done

[17:42] <arigo> I'll try to see if the links *to* architecture.txt from other places violate this rule, too.

[17:43] <hpk> pedronis: again, if you want to be 'autonosy', drop me a note or say you are not interseted to stop me from asking :-)

[17:43] <hpk> geninterp documentation

[17:46] <arigo> except MissingStakkarsError: pass

[17:46] <cfbolz> :-)

[17:47] <hpk> we could say that it's ok to postpone it past M0.5

[17:47] <cfbolz> It probably is. The docs don't mention geninterp anyway, so nobody will notice :-)

[17:48] <arigo> hum :-)

[17:48] <hpk> cfbolz: sure, it's just that at some point i'd like to understand it better :-)

[17:48] <cfbolz> hpk: yes, me too.

[17:49] arre (ac@ratthing-b3fa.strakt.com) left irc: "using sirc version 2.211+KSIRC/1.3.11"

[17:49] <hpk> let's leave it for now, and nag christian when he is back

[17:50] <hpk> ok?

[17:50] <arigo> yes

[17:50] <cfbolz> better mission statement for homepage

[17:50] <hpk> *better mission statement*

[17:50] <cfbolz> ok, I'll leave it to you from now on :-)

[17:50] <hpk> nono, just go ahead :-)

[17:50] <hpk> this is the ethernet protocol

[17:51] <cfbolz> :-)

[17:51] <cfbolz> we need some master of ceremonies (with a nice funny hat)

[17:51] <hpk> i think the mission statement is clearly M0.5 and easy enough, shall we move on?

[17:51] <arigo> *-^-<:-)

[17:51] <arigo> yes

[17:52] Action: hpk is out 5 minutes, but says from the off that he considers web site navigation closed

[17:52] <pedronis> (hpk: autonosy: yes)

[17:53] <arigo> let's make it resolved yes.

[17:53] <pedronis> backtrack: what about goal dir?

[17:53] <arigo> s/yes/then

[17:53] <cfbolz> I'm away for a few minutes, too

[17:53] cfbolz (~cfbolz@hdlb-d9b946e7.pool.mediaWays.net) left irc: "Verlassend"

[17:54] <arigo> ok, time for a few minutes' break then

[18:00] <arigo> goal/ could be moved to pypy/translator/goal for translate_pypy and friends,

[18:00] <arigo> and then we should see if we remove or keep some of the other files as tests

[18:00] <arigo> basically we are past the phase of a directory with small snippets of code that we are trying to interpret correctly...

[18:01] <arigo> (now we need obscure detail-testing CPython tests as examples of code that we are trying to interpret "correctly" :-)

[18:02] cfbolz (~cfbolz@hdlb-d9b95ec4.pool.mediaWays.net) joined #pypy.

[18:06] <arigo> goal dir: assigned to me

[18:06] <arigo> (issue47)

[18:07] <arigo> remaining issues:

[18:08] <arigo> traceobjspace refinements and docs

[18:08] <arigo> (wait more for news from Richard before deciding on this?)

[18:08] <arigo> integrate exceptions geninterping

[18:08] <arigo> (same, see if Christian wants to do it or if we postpone it)

[18:08] <arigo> _file.py needs more tests

[18:09] <arigo> (adding some now, probably postpone then)

[18:09] <arigo> General documentation overhaul

[18:10] <cfbolz> I guess that's getting better with architecture and getting_started being worked on

[18:10] <arigo> ok

[18:10] <arigo> Toplevel README

[18:11] <arigo> (kind of resolved unless I'm mistaken)

[18:11] <arigo> -- end of list --

[18:11] <cfbolz> README: Holger wanted to do something about it, so don't close it just now

[18:12] <arigo> ok

[18:12] Action: hpk back for a minute

[18:12] <arigo> do we need an issue for Holger, "write docs about modules" ?

[18:12] <hpk> indeed, README -> README + LICENSE and i think it's needed for dist/

[18:13] <hpk> arigo: yes, i also did a bit on coding-guide and would move it to become another chapter in coding-guide

[18:13] <hpk> _file.py doesn't have an assignee

[18:14] <arigo> I'm taking it and setting done_cbb for now

[18:14] tremolo (~tremolo@82.144.60.19) left irc: Read error: 60 (Operation timed out)

[18:14] <hpk> right, and move it to current or M1.0

[18:14] <hpk> (done-cbb issues still show up in the default view)

[18:15] <hpk> any remarks regarding my README+LICENSE remarks?

[18:15] <cfbolz> have we found out who the license holder is?

[18:16] <hpk> i could write a script and list everyone with >10 commits or something

[18:17] <hpk> at best this would be per file

[18:17] <pedronis> hpk: I fear this should be discussed with the consortium

[18:17] <hpk> pedronis: well

[18:18] <hpk> i think we say in the consortium agreement that copyrights remains with the writers

[18:19] <cfbolz> I think it would be nice to list contributors anyway

[18:19] <hpk> i agree

[18:19] <cfbolz> (which doesn't solve the license question)

[18:19] <hpk> pedronis: what specifically are you thinking off?

[18:19] <pedronis> you are right but propably a bad idea long term, there's a reason why open source project are moving away from the simple model

[18:19] <hpk> linux is not

[18:19] <hpk> is it?

[18:20] <pedronis> no, but linux has big backers with huge patents portfolios

[18:21] <hpk> i think going for mostly "automatated" copyright computations is the fairest it can get right now (and it's a repetetive process)

[18:22] <hpk> arigo: are you creating that module issue?

[18:22] <arigo> ...ok doing it now

[18:23] <cfbolz> are we finished with the issues, then?

[18:23] <pedronis> hpk: there are still issues, I don't think I have the copyright on code that I wrote while working for Strakt

[18:24] <pedronis> (in fact I'm pretty sure I don't)

[18:25] <hpk> pedronis: depends on your contract, but indeed

[18:25] <arigo> cfbolz: it seems so

[18:25] <hpk> except for the README + LICENSE one

[18:25] <hpk> (and what about the general doc overhaul?)

[18:26] <hpk> cfbolz: but feel free to drop out :-)

[18:26] <arigo> (I suggested to postpone it until we have written some more of the doc we promized)

[18:26] <pedronis> universities also have varying rules

[18:26] <hpk> (ok, let's reflect that decision in the tracker)

[18:26] <cfbolz> yes. the most severest problems seem to be addressed by various other issues.

[18:27] <hpk> pedronis: yes, but what do you think for M0.5 now, we don't want to wait for a consortium meeting, do we? (and even then it owludn't be clear we have an immediate solution)

[18:28] <pedronis> I think you may want to mail pypy-funding and see what attention you get

[18:29] <hpk> hum, let me suggest something that has more chances of getting resolved quickly (i strongly doubt we will get enough responses to have pypy-funding interrogation become meaningful)

[18:30] <hpk> we list for everyone the entry points into companies/universities

[18:30] <hpk> and i write a script that considers all files, walks through the logs and does te right thing (tm)

[18:31] <hpk> so pedronis before 1st January would become Samuele Pedronis (private) and after Samuele Pedronis (AB Strakt)

[18:31] Nick change: lac -> lac-afk

[18:31] <hpk> that's an approximation that we can refine over time

[18:32] <hpk> or do you think that it is not enough?

[18:33] <cfbolz> hpk: should I assign README to you?

[18:33] <arigo> bad idea of Samuele _Pedroni_ you to choose pedronis as a nickname :-)

[18:34] <hpk> oh, sorry sorry sorry, i am always getting confused

[18:34] <hpk> I can't think in addition to talking :-)

[18:34] <pedronis> I did not choose it, it was chosen at the dawn of times

[18:34] <pedronis> and stuck

[18:35] <hpk> cfbolz: i guess so, LICENSE becomes a different issue because of that damned copyright issue

[18:35] <hpk> let's also not forget that the MIT license basically allows to do everything with the source

[18:36] <hpk> so basically i wouldn't try to emphasize this discussion too much but simply deploy a half-sensible approach and invite the consortium to make sugestions for M1.0

[18:38] <cfbolz> the license itself isn't under debate anyway, right?

[18:38] <hpk> right

[18:38] <cfbolz> good

[18:38] <hpk> i suggest just listing all >10 committers in a top-level LICENSE and be done for now

[18:38] <hpk> and include all the companies and universities as well

[18:39] <hpk> but if anyone else has a suggestion short off postponing the whole release: i am all ears

[18:39] <arigo> not me

[18:40] <cfbolz> me neither

[18:41] <hpk> arigo, cfbolz: does that mean that my last or former suggestion would work for you?

[18:41] <cfbolz> I didn't sign anything about code I write, if that's your question.

[18:42] <cfbolz> so you can just write "private" for me

[18:42] <hpk> cfbolz: you will sign something when you sign the EU accession/contract stuff

[18:42] <hpk> cfbolz: you will remain always with private copyrights for now

[18:42] <cfbolz> ok. whatever.

[18:43] <hpk> pedronis: so if we choose my last suggestion and i sent a mail to pypy-funding telling them about it, would that work for you?

[18:43] <cfbolz> as long as it will always be avaiable via MIT or similar

[18:43] <hpk> cfbolz: you only need to agree to license your stuff via MIT

[18:44] <hpk> cfbolz: you can sell it to MS later, still :-)

[18:44] <pedronis> hpk: ok

[18:44] <cfbolz> hpk: hah

[18:45] <pedronis> hpk, cfbolz: yes, the individual partner agreement has a part about agreeing on relasing under MIT license

[18:45] <cfbolz> yes, that's fine

[18:45] <hpk> and just so i know: would everyone also agree doing copyright-computations on a per-file level?

[18:45] <cfbolz> you mean instead of a per LOC level?

[18:46] <cfbolz> and what do you mean by per-file level? Many people work on a file.

[18:46] <hpk> no my very last suggestion was to just mention copyright holders in the toplevel LICENSE file

[18:46] <cfbolz> yes, that's clear.

[18:47] <cfbolz> but what about "doing copyright-computations on a per-file level"

[18:47] <cfbolz> I didn't get that

[18:47] <hpk> a script that visit each file separately instead of dist/pypy wholesale

[18:48] <cfbolz> so everyone that did more than n commits gets listed?

[18:48] <hpk> yes, basically

[18:48] <cfbolz> are there people with only very few commits?

[18:48] <cfbolz> otherwise this point is moot anyway

[18:49] <hpk> no clue to be honest

[18:49] <cfbolz> just choose n small enough to enclude me, then it's fine with me :-)

[18:50] adim (adim@logilab.net2.nerim.net) left #pypy.

[18:50] <cfbolz> s/enclude/include

[18:50] <hpk> i am eluding you, np

[18:50] <hpk> :-)

[18:50] <cfbolz> :-)

[18:51] <hpk> ok, so i am putting up a LICENSE issue and assign me to it, feel free to add comments

[18:51] <cfbolz> Is there anything else? Otherwise I would drop off

[18:51] <cfbolz> ok

[18:51] Action: hpk needs a thorough break as well

[18:51] <cfbolz> I think it was pretty productive

[18:52] <hpk> pedronis: please feel very free to add comments to the LICENSE issue, i'd like to send pypy-funding mail tomorrow sometime

[18:52] <hpk> cfbolz: yes, indeed!

[18:52] <cfbolz> see you, then!

[18:52] <arigo> see you!

[18:52] <hpk> see you!

[18:52] cfbolz (~cfbolz@hdlb-d9b95ec4.pool.mediaWays.net) left irc: "Verlassend"

----- silence for 20 minutes -----

[19:12] <hpk> pedronis: i added you to autonosy, should have done it earlier

[19:12] <hpk> now all issues are more or less created, sorry

[19:13] Action: hpk -> taking a larger break

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

[20:13] rxe (~rxe@host-212-158-203-122.bulldogdsl.com) joined #pypy.

[20:13] <rxe> Hi

[20:13] <arigo> hi!

[20:13] pedronis (pedronis@ratthing-b246.strakt.com) left irc: "Chatzilla 0.9.67 [Firefox 1.0.2/20050325]"

[20:14] <rxe> I am bit confused as to what to checkout these day

[20:14] <arigo> for PyPy? it is http://codespeak.net/svn/pypy/dist

[20:15] <rxe> yes that is what I have - do I need anything else for more docs? And for testing?

[20:15] <arigo> not essentially

[20:15] <arigo> if you want to run py.test inside lib-python/2.3.4/test, then you'll get a message about checking out an optional directory containing CPython test results

[20:16] <arigo> (we save them to svn because they take ages (one day!) to run)

[20:17] <rxe> ah... that is http://codespeak.net/svn/pypy/testresult

[20:17] <arigo> indeed.

[20:17] <rxe> ok - that's good

[20:17] <rxe> what is milestone .5?

[20:17] <arigo> friday, first release of PyPy :-)

[20:18] <rxe> ah There is an issue against me for then :-)

[20:18] <hpk> hi richard

[20:19] <hpk> "against me" is a good expression :-)

[20:19] <rxe> hi holger

[20:19] <rxe> hpk: :-)

[20:20] <rxe> hpk: can you try something for me... traceobjectspace and too many ops being displayed

[20:20] <rxe> i think what you are seeing is uncaching opertations

[20:20] <rxe> maybe?

[20:21] <rxe> that if you repeat then same cmd, there wont be so many operations

[20:22] <hpk> it looks much better compared to two days ago

[20:22] <rxe> guess the question is how to trigger the cache so you dont see those operations when you type commands

[20:23] <rxe> ie if you do range(5) first time then you see loading of the cache

[20:23] <arigo> rxe: maybe enter_cache_building_mode() gives the hint you need

[20:23] <arigo> this and leave_cache_building_mode() are new methods on the space

[20:23] <rxe> ahhh! thankyou! :-)

[20:24] <arigo> we recently added them for objspace/flow/objspace.py

[20:24] <arigo> there used to be a different way of doing this, I guess we broke this part of the tracing when we changed it.

[20:24] <hpk> i am wondering if we couldn't represent return values nicer for tracespace

[20:25] <rxe> i was playing last nice with using the objects repr - except it can very slow

[20:25] <rxe> last night

[20:25] <hpk> i am more refering to the formatting

[20:26] <hpk> i'd probably try to append the return value to the actual operation with like

[20:26] <hpk> -> <Function range>

[20:26] <rxe> i see - what about exceptions?

[20:26] <arigo> another new interface, w_objects all have a getname(space, 'defaultvalue')

[20:27] <arigo> might be a good compromize if you don't want to use repr

[20:28] <rxe> arigo: cool - thanks.

[20:29] <hpk> rxe: exceptions: maybe something like '-> Exception: ...' ?

[20:29] <rxe> hpk: at the moment if the operation has a result we have a new line with ":= xxx" . if the operation raises an exception, then it " x= xxx"

[20:29] <rxe> hpk: yes that is nicer

[20:29] <rxe> and more obvious

[20:30] <hpk> the x= is confusing, especially since i just tried 'int("x")' :-)

[20:30] <hpk> other than that it seems just nice

[20:30] <hpk> if we were to take screenshots for the release tracespace would be worth one :-)

[20:31] <rxe> :-) Yes. Lines can be long which maybe is one reason for new lines. But if we dont use wrapped reprs everywhere, then they will be shorter.

[20:32] <hpk> yes, and even then you cannot always avoid long lines

[20:32] <hpk> well, one idea would be to indent the space-ops with just four spaces compared to the bytecodes

[20:33] <hpk> and then even forget about the ">>" to reduce character garbage :-)

[20:33] <hpk> i guess that isn't ahrd to change (i haven't looked at the code for some time)

[20:33] <rxe> wow - 48 seconds to start py.py ?

[20:34] <hpk> caching?

[20:35] <rxe> not sure... my laptop isnt in great health.

[20:35] <rxe> Yes I think the indenting of operations is not such a good idea at the moment.

[20:36] <arigo> rxe: strange. from scratch, I get 16 seconds for start-up on my old laptop.

[20:36] <arigo> < 3 sec the next times.

[20:37] <rxe> I need to reboot the laptop, that time was 16 seconds.

[20:38] <hpk> there could be something with the caching mechanism though

[20:38] <rxe> i did svn st, I have't changed anything. Has caching changed today?

[20:39] <hpk> no, i don't think so

[20:39] <hpk> but in general managing file-caches transparently is a delicate issue (and our applevel -> interplevel translation is doing that)

[20:39] arigo (~arigo@ratthing-b407.strakt.com) left irc: "see you later"

[20:40] <hpk> if your second time takes longer than the first one, it sounds odd

[20:40] <hpk> but anyway, feel free to reboot

[20:40] <rxe> 48 seconds was the first time i ran after a big svn up

[20:40] <hpk> ah, and then 16 seconds

[20:40] <rxe> yup... makes sense

[20:40] <hpk> sorry, than i misunderstood your reply to armin

[20:41] <rxe> is issue25 really important for milestone0.5?

[20:42] <rxe> :-)

[20:42] <hpk> yes, definitely :-)

[20:42] <hpk> i'll buy you an extra beer in g-borg :-)

[20:42] <rxe> I cant see a problem other than my laptop!

[20:42] <rxe> thanks! :-)

[20:43] <rxe> I am about to go to "go"

[20:43] <rxe> Thanks for the feedback...

[20:44] <rxe> bbye

[20:44] <hpk> sure, see you!

[20:44] rxe (~rxe@host-212-158-203-122.bulldogdsl.com) left irc: "Download Gaim: http://gaim.sourceforge.net/";

----- silence for 41 minutes -----

[21:25] yuuh (~yuuhu@i3ED6B51A.versanet.de) left irc: "utz utz utz"

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

[21:56] yuuh (pjaivjqn@i3ED6B51A.versanet.de) joined #pypy.

[00:00] --- Wed May 18 2005