==== Channel ##pypy: 05/21/05 ====

[00:07] <cfbolz> ok. I stop working on anything not math related immediately and wish everyone a nice (and relaxing!) weekend.

[00:07] <hpk> cfbolz: good idea

[00:07] <cfbolz> And Congratulations!!!

[00:08] <hpk> YES!!!!!

[00:08] <hpk> to you too, i mean!!

[00:08] <cfbolz> I feel very relieved!

[00:08] <hpk> you can tell

[00:08] <cfbolz> too bad we can't celebrate a bit

[00:09] <hpk> we can celebrate 0.7 at ep :-)

[00:09] <cfbolz> ok :-) see you then!

[00:09] <stakkars> yes that's fine, but still it feelslike a gap

[00:09] <cfbolz> right

[00:10] <stakkars> hmm, once that I started writing, I can't stop it.

[00:10] <stakkars> adding to translation.txt

[00:10] <hpk> well, be aware that people are surfing these pages right now

[00:10] <hpk> and nobody is there to proof-read anything

[00:11] <cfbolz> goodnight

[00:11] <stakkars> not checking in, then.

[00:11] cfbolz (~cfbolz@hdlb-d9b95ec8.pool.mediaWays.net) left irc: "Verlassend"

[00:11] <hpk> then again i shouldn't stop you from writing documentation :-)

[00:12] <hpk> it's tough enough to get everyone beaten to it and you are one of the toughest :-)

[00:12] <stakkars> I want to explain how all the litle code snippets are handled,cached and such.

[00:12] <stakkars> Actually that shouldgo into the gateway documentation, but for now I think it makes sense.

[00:12] <hpk> the current documentation has been proof-read a couple of times by various people

[00:13] <hpk> and we definitely need an interpreter-chapter that we can reference

[00:13] <stakkars> no chance,I always find another typo

[00:13] <hpk> but i am really stopping now ...

[00:13] <hpk> see you!

[00:13] <stakkars> bye

[00:13] Nick change: hpk -> hpk_away

[00:26] Action: stakkars is thinking if it would be better to have a way to freeze the docs as well

[00:29] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

[00:38] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "BitchX sucks"

[00:45] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

[00:45] <aleale> Hi stakkars, are you there?

[00:48] <aleale> I've got 81 failures with test_all.py and 32 with py.test - trying to delete pyc's

[00:49] <aleale> ImportError no module named _formatting

[00:56] stakkars (~tismer@rosine27.inf.fu-berlin.de) left irc: Read error: 113 (No route to host)

[00:57] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "*monkey grin*"

[01:07] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.

[01:08] <aleale> sorry - all tests passes. Forgot to delete the cache. goodnight

[01:08] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: Client Quit

----- silence for 5 hr and 21 minutes -----

[06:29] tea (~tea@cap049-090.kcn.ne.jp) left irc: Remote closed the connection

----- silence for 2 hr and 54 minutes -----

[09:23] dialtone (~user@host111-56.pool80117.interbusiness.it) left irc: Read error: 113 (No route to host)

----- silence for 1 hr and 45 minutes -----

[11:08] Nick change: hpk_away -> hpk

----- silence for 1 hr and 9 minutes -----

[12:17] cfbolz (~cfbolz@hdlb-d9b946a7.pool.mediaWays.net) joined #pypy.

[12:17] <cfbolz> morning

[12:17] <cfbolz> hpk: Is this only in dist or also in the release?

[12:17] <hpk> cfbolz: not clear yet

[12:17] <hpk> hi carl :-)

[12:17] Action: hpk is 5 minutes busy

[12:17] <cfbolz> hi!

[12:18] <cfbolz> I'm away instantly anyway :-)

[12:20] <cfbolz> arghh! It's in the release!

[12:23] <hpk> what exactly is in the release?

[12:23] <hpk> the _formatting problem?

[12:25] <cfbolz> yes

[12:27] <hpk> i see, starting py.py twice is the culprit

[12:27] <cfbolz> You mean python py.py py.py

[12:27] <cfbolz> or in two sessions

[12:27] <cfbolz> ?

[12:28] <hpk> i mean python py.py

[12:28] <hpk> then "print %s" % 3

[12:28] <hpk> then quit

[12:28] <hpk> then python py.py

[12:28] <hpk> print "%s" % §3 -> crash

[12:28] <cfbolz> ouch. That's _very_ bad!

[12:29] <hpk> indeed

[12:29] <cfbolz> shit shit shit shit.

[12:29] <cfbolz> sorry

[12:30] <hpk> i agree anyway

[12:30] <cfbolz> I thought you were joking yesterday!

[12:30] <hpk> me too

[12:30] <cfbolz> :-(

[12:30] <cfbolz> What do we do now?

[12:30] <hpk> i am tempted to just revert to the rev three days ago

[12:30] <hpk> and copy the current situation to some branch

[12:31] <cfbolz> hmm.

[12:31] <cfbolz> what happens if we you throw the switch christian built in?

[12:31] <hpk> the releaseed pypy seems more unstable than the dist-one during the last couple month

[12:32] <cfbolz> I'd just revert to the pre print works in genc thing

[12:32] <hpk> i am going to wait for a few more people dropping by

[12:33] <cfbolz> Did you check the download count?

[12:33] <hpk> which download count? :-)

[12:33] <cfbolz> you're not counting downloads?

[12:33] arigo (~arigo@c-3a8b70d5.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.

[12:33] <hpk> cfbolz: not directly

[12:33] <hpk> cfbolz: but you can grep through log files

[12:33] <hpk> arigo: hi armin

[12:34] <arigo> hi!

[12:34] <cfbolz> hi

[12:34] <hpk> arigo: did you already hear that the release is broken?

[12:34] <cfbolz> hpk: that's good enough :-)

[12:35] <hpk> cfbolz: i think the count is >100 (but i would need to look at all the log files, don't know if they got rotated)

[12:36] <cfbolz> that's bad

[12:36] <hpk> arigo: do twice: python py.py -c "print '%s' % 3" on the released pypy

[12:36] <hpk> cfbolz: yip

[12:37] <hpk> so i am indeed thinking of a 0.6.1 that we put on the web page and at least announce to comp.lang.announce

[12:37] <arigo> hpk: I see

[12:38] <hpk> and i'd like 0.6.1 to be a known working pypy not some added hack to what we have now

[12:38] <hpk> the current dist should be copied to a branch

[12:38] <hpk> hum, but then it's not so easy, unfortunately :-(

[12:39] <arigo> hpk: we can copy release/0.6 and make the bug fix in that branch

[12:40] <hpk> arigo: you mean copy release/0.6 to 0.6.1 ?

[12:40] <arigo> maybe directly, yes (I thought about a branch inbetween)

[12:40] <hpk> yes makes sense

[12:40] <hpk> release/0.6 -> branch/0.6.x

[12:40] <arigo> yes

[12:40] <hpk> i am doing it

[12:41] <arigo> next time we need such a branch a few days in advance, so that normal work can continue to go to dist/ or trunk/

[12:42] <hpk> yeah, we need to think a bit about the exact details

[12:42] <arigo> yes.

[12:42] <hpk> also with respect to documentation which is currently coming from dist

[12:42] <cfbolz> sorry, I have to go (there's probably not much I can do anyway)

[12:42] <arigo> yes

[12:42] <arigo> see you

[12:43] <hpk> cfbolz: sure, you learn for your exams :-)

[12:43] <hpk> cfbolz: see you

[12:43] <cfbolz> yes

[12:43] cfbolz (~cfbolz@hdlb-d9b946a7.pool.mediaWays.net) left irc: "Verlassend"

[12:43] <hpk> arigo: opened the 0.6.x branch

[12:44] <hpk> do you have an idea how to fix things already?

[12:44] <hpk> i didn't really follow the last two days core changes due to all the other stuff

[12:45] <arigo> yes, the problem is obvious :-(

[12:45] <arigo> stringobject.py contains "import _formatting" in an app-level helper

[12:45] <arigo> which gets geninterp'ed

[12:45] <arigo> into a version that contains a CPython-level "import _formatting"

[12:45] <hpk> hua

[12:46] <arigo> indeed

[12:46] Action: hpk is meanwhile trying to fix all the references to 0.6 to become 0.6.1

[12:47] <hpk> hum, not sure though how to go about that exactly

[12:47] Action: hpk is out 5 minutes

[12:47] <lac-out4dinner> do we know how many people dl 0.6 ? It might be easier to just name what Armin fixes as 0.6 ....

[12:48] Nick change: lac-out4dinner -> lac

[12:52] ericvrp (~chatzilla@ericvrp.demon.nl) joined #pypy.

[12:53] ericvrp (~chatzilla@ericvrp.demon.nl) left irc: Client Quit

[12:54] ericvrp (~chatzilla@ericvrp.demon.nl) joined #pypy.

[12:55] <hpk> arigo: hum, i am tempted to just fix the web page to contain downloads for 0.6.1

[12:55] <hpk> and send a custom announcement/correction to the lists

[12:55] <arigo> yes -- what else?

[12:55] <hpk> well, our documents speak often of the "0.6" release everywhere

[12:55] <arigo> ah, ok

[12:56] <arigo> no, I guess it's fine to just fix the direct links

[12:56] Action: hpk wasn'T prepared to do any work today let alone work that requires thinking

[13:01] <arigo> should work now, testing...

[13:07] Action: hpk also has a strange autopath problem with pypy/bin/py.py

[13:07] <hpk> which gets fixed if i insert a __init__.py into pypy/bin/

[13:08] <hpk> too many people hacked on autopath.py

[13:10] <hpk> arigo: i packaged 0.6.1 for testing purposes with your fixes

[13:10] <hpk> http://codespeak.net/download/pypy/

[13:14] <hpk> arigo: so far it looks good

[13:15] <hpk> did you also try?

[13:15] <arigo> yes

[13:15] <arigo> autopath.py is incredible

[13:15] <hpk> yes

[13:15] <hpk> but it didn'ts tart like this

[13:15] <hpk> bob added a couple of obscure hacks

[13:16] <hpk> for example, autopath originally put the pypy-containing at the beginning of sys.path IIRC

[13:17] <hpk> well, but anyway

[13:18] <hpk> should we put the packages and links online now?

[13:18] <arigo> what is your autopath problem?

[13:19] <hpk> pypy/tool/sourcetools.py sees pypy/bin/py.py for "import py"

[13:19] <hpk> i am not yet exactly sure how so

[13:19] <arigo> in which situation?

[13:19] <hpk> but i think that autopath should ensure that the determined pypy-dir is really first on the path

[13:20] <hpk> python pypy/bin/py.py

[13:20] <arigo> I mean, I cannot reproduce the problem

[13:20] <hpk> yes, i can only reproduce it on codespeak

[13:20] <hpk> i guess it's some custom path setting

[13:20] <arigo> PYTHONPATH?

[13:21] <hpk> no, something else

[13:21] <arigo> (or just do it in screen)

[13:29] <arigo> ok, it's still a bit mysterious but just a dev-issue for later...

[13:29] <hpk> yes, i guess so

[13:29] <arigo> figuring out why windows end-of-lines confuse Unix py.py.

[13:29] <arigo> including autopath, which is currently above my guesses...

[13:30] <arigo> maybe put a note in getting-started, that the zip version doesn't work on Unix and probably vice-versa

[13:33] <hpk> website is updated

[13:34] <arigo> ok

[13:35] <hpk> so we don't care that the README and friends talk of 0.6 all the time, i guess

[13:35] <arigo> no

[13:36] Action: arigo forward-ports the bug fix to pypy/dist/

[13:36] <hpk> good idea

[13:37] <arigo> the testresult page looks very bad right now...

[13:37] <hpk> i was thinking of that in fact

[13:37] <hpk> (that was the first thing i noticed this morning, while enjoying a nice beginning of day :-)

[13:37] <arigo> fun fun fun

[13:37] <hpk> just revert the last round of test results?

[13:37] <arigo> yes, if easy

[13:38] <arigo> svn rm them, maybe

[13:39] <hpk> did that

[13:40] <hpk> better, but not good

[13:40] <hpk> apparently i was the only one runnig tests lately

[13:40] <arigo> ah ha

[13:40] <arigo> yes, I don't run them all daily, far from it

[13:41] <arigo> checked in

[13:42] <arigo> maybe you should revert your directory to the status of last week...

[13:42] <hpk> yeah, i geuss so :-(

[13:43] <arigo> uh, no

[13:43] <arigo> Carl's numerous failures are from the 17th of May

[13:43] <hpk> it's better now

[13:43] <hpk> look again

[13:43] <arigo> yes

[13:44] <arigo> very strange

[13:45] <arigo> we need a warning mail to pypy-dev when one or more tests go from OK to ERR

[13:45] <hpk> or just file an issue

[13:46] <arigo> :-)

[13:46] <hpk> spamming pypy-dev with almost 200 subscribers might be a bit much :-)

[13:46] <arigo> right

[13:46] <hpk> and at best assign it to the person who caused it :-)

[13:46] <arigo> :-)

[13:46] <arigo> btw, can we file a new issue by e-mail?

[13:46] <hpk> yes

[13:46] <hpk> simply mail there

[13:46] <hpk> i mean the pypy-dev-issue at codespeak.net

[13:47] <arigo> ok

[13:47] <hpk> i guess you can set properties in the subject line as usual

[13:47] <arigo> good

[13:47] <hpk> although i would generally prefer if roundup had a way to change things in the body

[13:47] <hpk> so that one could uniformly used it from svn-checkin and mail

[13:48] <hpk> so where do we mail?

[13:50] <arigo> pypy-dev, python-announce?

[13:52] <hpk> i am mailing python-announce

[13:52] <hpk> you reply to your message on pypy-dev?

[13:52] <arigo> ok, that leaves pypy-dev for me, yes

[13:53] <hpk> you can also mail python-announce but it's maybe confusing if the messages come from different persons

[13:53] <hpk> people might think that we are arguing :-)

[13:53] <arigo> no, they wouldn't think that :-)

[13:54] <arigo> (here you must answer "yes, they would")

[13:54] <hpk> they certainly would

[13:56] <hpk> btw, if we had advertised pypy/dist as being the latest released version ....

[13:56] <hpk> then i could simply recommend "svn up"

[13:57] <hpk> the "btw" relates to the argument thingie of course :-)

[13:58] <arigo> well :-) we could advertise the branch/0.6.x then ?

[13:58] <hpk> not really, because that is moving in steps possibly as well

[13:59] <hpk> and next month it's going to be 0.7

[13:59] <arigo> yes, I see the point

[13:59] <hpk> :-)

[14:00] <arigo> can you easily prevent check-ins into, say, pypy/dist, while allowing removing-and-replacing the dist directory itself?

[14:00] <hpk> not sure, i can certainly restrict to very few people or just a single person

[14:00] <hpk> which is probably good enough

[14:01] <arigo> right

[14:01] <arigo> there are more issues, though

[14:01] <arigo> e.g. which version of the documentation/ should the website show?

[14:02] pedronis (~Samuele_P@c-3a8b70d5.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.

[14:02] <arigo> and against which version should the CPython tests run daily?

[14:02] <hpk> is this subject line for python-announce-list ok:

[14:02] <hpk> PyPy 0.6.1 bugfix release / 0.6 broken

[14:02] <hpk> ?

[14:03] <arigo> is "0.6 broken" needed?

[14:03] lypanov (~alex@a80-126-190-213.adsl.xs4all.nl) joined #pypy.

[14:03] <hpk> well, it was broken, wasn't it?

[14:03] <lypanov> congratz you guys :)

[14:03] <arigo> yes, but so soon after the 0.6, it's obvious that 0.6.1 is a needed bugfix

[14:03] <hpk> lypanov: thanks :-)

[14:03] <hpk> arigo: ok

[14:03] <lypanov> (and gals in fact ;))

[14:04] <arigo> lypanov: thanks -- though I guess you're talking about 0.6 and don't know yet that 0.6.1 has been released

[14:04] <lypanov> arigo: hehe. actually i just read your mail which reminded me to sign in at a sane time ;)

[14:04] <lypanov> (like, now as opposed to 3 am last night)

[14:05] dialtone (~user@host51-0.pool21345.interbusiness.it) joined #pypy.

[14:12] <arigo> hpk: just a note, Christian found out about the problem and mention it in a 5am check-in message... and nowhere else

[14:12] <hpk> i see

[14:12] <hpk> with "deliverable" he means the release

[14:13] <arigo> yes, it's exactly the problem I fixed too

[14:13] <lypanov> arigo: side note, nothing automated will fix the need for a .1 if you don't do pre releases to a larger audience...

[14:13] <lypanov> something *always* goes wrong

[14:14] <hpk> probably

[14:14] <pedronis> we could have done a release candidate

[14:14] <pedronis> anyway large changes that needs 2 days of bug fixing should be a no-no the next time around

[14:15] <hpk> i guess we may want to appoint a person who controls the release tree starting two/three days before the release

[14:15] <hpk> but let's discuss this or other suggestions later

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

[14:36] dialtone (~user@host51-0.pool21345.interbusiness.it) left irc: Read error: 145 (Connection timed out)

[14:47] Rhamphoryncus (~rhamph@rhamphoryncus.user) joined #pypy.

[14:47] <Rhamphoryncus> hmm surprisingly empty

[14:47] <hpk> what did you expect?

[14:47] <hpk> everybody is released

[14:47] <hpk> kind of

[14:48] <Rhamphoryncus> huh?

[14:48] <Rhamphoryncus> I'd expect the release to *increase* activity, atleast with all the tourists :)

[14:48] Action: lypanov hasn't been released yet

----- silence for 44 minutes -----

[15:32] dialtone (~user@host51-0.pool21345.interbusiness.it) joined #pypy.

[15:32] <lypanov> Rhamphoryncus: its a very very quiet room generally anyway :)

[15:32] <lypanov> Rhamphoryncus: and not many people even understand what pypy is for ;)

[15:36] <Rhamphoryncus> heh

[15:36] <Rhamphoryncus> their loss :)

[15:37] <lypanov> yup. but then the lack of understanding is already pretty much there loss :)

[15:37] <lypanov> s/ere/eir/

[15:43] Nick change: hpk -> hpk_away

[15:46] dialtone (~user@host51-0.pool21345.interbusiness.it) left irc: Read error: 145 (Connection timed out)

[15:46] <arigo> as a side note, I am back to working on C code generation now

[16:01] Gromit (~bear@A3692.a.pppool.de) joined #pypy.

----- silence for 1 hr and 21 minutes -----

[17:22] Gromit (~bear@A3692.a.pppool.de) left irc: Read error: 110 (Connection timed out)

[17:22] Gromit (~bear@A3682.a.pppool.de) joined #pypy.

[17:28] dialtone (~user@host72-0.pool21345.interbusiness.it) joined #pypy.

[17:43] Nick change: hpk_away -> hpk

[17:43] <hpk> arigo: you don't even feel like resting, i guess

[17:46] <hpk> pedronis: ayt?

[17:47] <pedronis> yes

[17:47] <pedronis> hpk: yes

[17:47] <hpk> i think it would be a good idea to mail squeak-dev

[17:47] <hpk> we talked about this earlier

[17:48] <pedronis> I see

[17:48] <hpk> do you agree?

[17:49] <pedronis> yes, but I would contact Markus (?) or the Impara people to figure out an intro to annoucenment

[17:50] <hpk> pedronis: i could mail Markus and ask him to forward it appropriately

[17:50] <pedronis> yes

[17:50] <hpk> are you subscribed to squeak-dev?

[17:50] <pedronis> no

[17:50] <pedronis> sometimes I skim it

[17:51] <pedronis> I'm just aware that some part of the smalltalk communities are less friendly to different languages that for example the python community itself

[17:51] dialtone (~user@host72-0.pool21345.interbusiness.it) left irc: Read error: 145 (Connection timed out)

[17:51] <pedronis> (disclaimer: generalizing etc)

[17:54] Action: lypanov grins

[17:54] <lypanov> one random guy doing a ruby iirc (peter suk)

[17:55] <lypanov> seems to be an intelligent guy... but... hell, i place my trust in pypy and my own project :)

[17:55] <hpk> lypanov: hey, but you may forward the announcement to some appropriate ruby-list :-)

[17:56] <hpk> quite a bit of ou architecture should be kind of language independent anyway

[17:56] Action: lypanov isn't actually on any ruby list anymore

[17:56] <lypanov> hehe. kind of? :P

[17:57] <lypanov> (the ruby lists are hellish wrt s/n)

[17:57] Gromit (bear@A3682.a.pppool.de) left #pypy ("Client Exiting").

[18:00] <hpk> s/b ?

[18:00] <hpk> s/n?

[18:01] <pedronis> signal/noise ratio

[18:10] <hpk> i sent the mail to Marcus

[18:15] <hpk> arigo: just to be sure and before i forget it: you did see that i changed the architecture document after your recent mission statement checkins?

[18:19] <arigo> I remember you were changing it, yes -- didn't look so far

----- silence for 1 hr and 3 minutes -----

[19:22] dialtone (~user@host111-56.pool80117.interbusiness.it) joined #pypy.

----- silence for 1 hr and 3 minutes -----

[20:25] -deedra (dmwaters@dmwaters-osuosl.staff.freenode) to $*- {global notice} Hi all! In a little more then 12 hours, I will begin major maintenence on the network. For more information, please see 'http://www.freenode.net/news.shtml' Thank you for your patience, and thank you for using freenode!

----- silence for 49 minutes -----

[21:14] ericvrp (~chatzilla@ericvrp.demon.nl) left irc: "Chatzilla 0.9.68.3 [Firefox 1.0.4/20050511]"

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

[21:34] bbailey (~bbailey@adsl-146-20-126.mia.bellsouth.net) left irc: "Leaving"

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

[21:50] bbailey (~bbailey@adsl-146-20-126.mia.bellsouth.net) joined #pypy.

[21:54] cfbolz (~cfbolz@hdlb-d9b95eec.pool.mediaWays.net) joined #pypy.

[21:54] <cfbolz> hi!

[21:54] <cfbolz> hpk: you there?

[21:54] <hpk> hi carl!

[21:54] <hpk> i think so

[21:54] <hpk> i am at this minute reading your posting on llvm-dev :-)

[21:54] <hpk> telepathy seams to work reasonably well

[21:54] <lypanov> hehe

[21:54] <cfbolz> :-)

[21:55] <cfbolz> did you see the announcement of yet another testing thing called PyTest?

[21:55] <hpk> i am actually thinking of joining you in the LLVM efforts :-)

[21:55] <hpk> yes, i did

[21:55] <hpk> unfortunate naming clash

[21:55] <cfbolz> cool!

[21:55] <cfbolz> sorry, I mean you joining me of course

[21:55] <hpk> i thought so

[21:56] <hpk> it's the second time somebody releases something called pytest

[21:56] <hpk> i guess i really need to do a fully official release

[21:56] <cfbolz> yes.

[21:56] <hpk> yesterday during the release frenzy a gentoo guy came by and wanted to package it

[21:56] <hpk> yet another person pressing me to do a release :-)

[21:56] bbailey (~bbailey@adsl-146-20-126.mia.bellsouth.net) left irc: "Leaving"

[21:57] <cfbolz> but you're probably having enough of releases lately

[21:57] <hpk> not really

[21:57] <hpk> in some ways i think it also moves things forward

[21:57] <hpk> more than what would happen usually

[21:57] <cfbolz> especially documentation

[21:58] <hpk> yes, but writing documentation also servers to reflect the current situation

[21:58] <hpk> and how reasonable it is to explain it

[21:58] <hpk> and it serves to share a common picture

[21:58] <hpk> including details

[21:58] <hpk> "reflect" in the ense of thinking

[21:58] <hpk> the german sense in fact :-)

[21:59] <cfbolz> :-)

[21:59] <cfbolz> yes, that's true. I learned a lot about other parts of PyPy

[21:59] <cfbolz> especially geninterp

[22:01] <cfbolz> "reflect" kann mean reflektieren, too

[22:01] <cfbolz> can, of course :-)

[22:01] <hpk> yip

[22:02] <hpk> well, anyway, i'll still want to re-find a way into the translation part

[22:02] <hpk> i have been out of that loop a bit lately

[22:02] <hpk> btw, we use SSA in translation.txt as single style assignment

[22:02] <hpk> but that is found nowhere on the web

[22:02] <hpk> where its always static single assignment

[22:03] <cfbolz> right

[22:03] <hpk> that's a bug

[22:03] <hpk> i guess

[22:03] <hpk> and finally i am understanding that phi-function stuff :-)

[22:04] <hpk> i mean, i likely learned that in my compiler courses at some point

[22:04] <hpk> but well

[22:04] <cfbolz> yes, our links make that very easy to understand

[22:05] <cfbolz> I think our university doesn't even offer compiler courses :-)

[22:06] <hpk> we don't have any SSA references

[22:06] <hpk> and deliver the wrong wording for it, btw :-)

[22:07] <hpk> do you know if "straight-line code" has any formal definition?

[22:07] <hpk> it's used a lot

[22:07] <cfbolz> I don't understand it

[22:07] <hpk> for example: http://www.flex-compiler.lcs.mit.edu/Harpoon/quads/node3.html

[22:08] <cfbolz> The wikipedia article about SSA is very good, btw.

[22:08] <cfbolz> straight-line code sounds like "without branching constructs"

[22:09] <hpk> it's probably some banality like this

[22:10] <cfbolz> I'm quite sure, because it is mentioned as the contrary of "branching code"

[22:10] <hpk> makes sense

[22:12] <cfbolz> I'm going back to math-land.

[22:12] <cfbolz> Let's discuss genllvm next week.

[22:12] <hpk> sure

[22:13] <hpk> good luck

[22:13] <cfbolz> see you

[22:13] <cfbolz> Yes, I'll really need that

[22:13] <cfbolz> (the prof told me last time I had "Mut zur Luecke" :-) )

[22:14] <hpk> :-)

[22:14] <cfbolz> bye

[22:14] cfbolz (~cfbolz@hdlb-d9b95eec.pool.mediaWays.net) left irc: "Verlassend"

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

[22:32] fredrik (fredrik@c83-248-135-181.bredband.comhem.se) joined #pypy.

[22:37] -lilo (lilo@levin-pdpc.staff.freenode) to $*- [Global Notice] Hi all. As reported previously, we're going to have major downtime in about 11 hours.... for an update on the downtime and on fundraising, please take a look at http://freenode.net/news.shtml .... thank you for your patience, and thank you for using freenode!

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

[23:17] <lypanov> k. gtg :)

[23:17] lypanov (alex@a80-126-190-213.adsl.xs4all.nl) left #pypy.

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

[23:36] stakkars (sjsjoubq@i528C13E8.versanet.de) joined #pypy.

[23:40] <arigo> hi!

[23:40] <stakkars> hi!

[23:41] <stakkars> that _formatting problem (caused by the quick hack at importing) -- I thought I checked the correction in last night?

[23:42] <arigo> stakkars: yes, sorry

[23:42] <arigo> what we did was:

[23:42] <arigo> copy release/0.6 into a new "branch/0.6.x"

[23:43] <arigo> fix this specific problem (and touch nothing else)

[23:43] <arigo> make the release 0.6.1

[23:43] <stakkars> without checking that I fixed this already

[23:43] <arigo> yes

[23:43] <arigo> then I merged the change to the dist, and found out just after that you already did, so cancelled by changes.

[23:43] <stakkars> ah ok

[23:43] <arigo> stakkars: anyway we couldn't have used your changes easily, as we wanted *only* a minimal fix

[23:44] <arigo> it could have been separated. the truth is just that we didn't notice.

[23:45] <arigo> (you might have sent a warning e-mail to pypy-dev and/or opened an issue, though, when you noticed)

[23:45] <stakkars> sure (and hpk's reaction is not applicable)

[23:45] <arigo> I'm not sure what you are referring to

[23:45] <stakkars> oh, I thought the checkin was enough. Will do next time.

[23:46] <arigo> well, we gave the release a lot of "meaning". being broken was extremely bad, in the opinion of not just me.

[23:47] <stakkars> the fact is that we just found a problem too late that I caused days before.

[23:47] <arigo> days == 2

[23:47] <arigo> if anything it showed that we need a release branch at least three days before the release, with *no* changes in it but important fixes.

[23:47] bbailey (~bbailey@adsl-146-20-126.mia.bellsouth.net) joined #pypy.

[23:48] <arigo> I'll stress again that you shouldn't have gone off and work on "print" for so long without giving or asking for feedback...

[23:48] <stakkars> so you are in fact mad at me.

[23:48] <arigo> no

[23:48] <arigo> the result is great.

[23:48] <arigo> it's the style and timing...

[23:49] <stakkars> then we take it again asalesson andlook forward?

[23:49] <arigo> definitely

[23:49] <arigo> we have been working today on C generation again, you might want to have a look too

[23:50] <arigo> we can generate static structures for any kind of global data, translator/c/test/test_database.py

[23:50] <stakkars> absolutely! I have though about things quite a lot, and I think the rtyper approach is

[23:51] <stakkars> absolutely applicable to geninterp, too.

[23:51] <stakkars> btw. I stepped into a problem with genc when it tries to create an empty tuple.

[23:51] <stakkars> did you see this before?

[23:52] <stakkars> (looking into svn updates)

[23:54] <arigo> hum, it's somehow bad that we didn't meet in person after all

[23:54] <arigo> discussing these things face-to-face would have made much more sense

[23:55] <arigo> at this point, genc is going away, and building geninterp on top of rtyper seems to be a far-away goal --

[23:55] <arigo> if I may say so, I'd think that you shouldn't try to go in that direction at the moment

[23:56] <stakkars> whatever the replacement of genc does, it replaces things with simpler things, which I would like to do, too.

[23:56] <stakkars> I mean, is the direction wrong?

[23:57] <arigo> in the long-term I have no clue -- maybe not. right now, probably.

[23:57] <arigo> you can't just drop rtyper in geninterp, you need to annotate app-level code in the middle -- and that's a lot of new problems

[23:57] <stakkars> what I mean is: the genc (or what is it now) takes what we write in interp-level and produces great code.

[23:58] <stakkars> what I produce is just bad because I don't do annotation.

[23:58] <arigo> the annotator isn't meant for applevel code

[23:58] <stakkars> If I did, the produced code could be as good as the handwritten stuff that you can digest so well.

[23:58] <stakkars> why is there any difference at all?

[23:58] <arigo> the flowspace wasn't meant for applevel, and you managed to use it nevertheless (with sys.path hacks (!))

[23:58] <arigo> the annotator is much more complicated than the flowspace

[23:59] <arigo> so I'm very very afraid, at the moment

[23:59] <stakkars> there is one thing that's definately applevel and a simple application at the same time.

[23:59] <stakkars> the rpystone program is such a beast that runs great with the annotator.

[00:00] --- Sun May 22 2005