[00:23] hpk_ (~holger_kr@mail.trillke.de) joined #pypy.
[00:33] _hannes (qimzniss@i528C114B.versanet.de) left irc: "utz utz utz"
----- silence for 21 minutes ----- [00:54] idnar (mithrandi@idnar.user) left irc: Read error: 60 (Operation timed out)
----- silence for 16 minutes ----- [01:10] hpk_ (~holger_kr@mail.trillke.de) left irc: "ChatZilla 0.9.61 [Mozilla rv:1.7.5/20041217]"
[01:13] idnar (mithrandi@idnar.user) joined #pypy.
----- silence for 42 minutes ----- [01:55] tea (~tea@cap058-158.kcn.ne.jp) left irc: "using sirc version 2.211+KSIRC/1.3.10"
----- silence for 20 minutes ----- [02:15] pedronis (~Samuele_P@c-2b8b70d5.022-54-67626719.cust.bredbandsbolaget.se) left irc: "Chatzilla 0.9.68a [Firefox 1.0.2/20050317]"
----- silence for 31 minutes ----- [02:46] fredrik (fredrik@c83-248-135-181.bredband.comhem.se) left irc: "http://fredrikj.net";
----- silence for 38 minutes ----- [03:24] dialtone (~dialtone@host111-56.pool80117.interbusiness.it) left irc: Remote closed the connection
[03:25] dialtone (~dialtone@host111-56.pool80117.interbusiness.it) joined #pypy.
----- silence for 3 hr and 49 minutes ----- [07:14] idnar (mithrandi@idnar.user) left irc: Connection timed out
----- silence for 2 hr and 7 minutes ----- [09:21] idnar (mithy@idnar.user) joined #pypy.
----- silence for 40 minutes ----- [10:01] arre (ac@ratthing-b3fa.strakt.com) joined #pypy.
----- silence for 48 minutes ----- [10:49] hpk (~hpk@merlinux.de) left irc: Remote closed the connection
----- silence for 56 minutes ----- [11:45] Gromit (~bear@B74c2.b.pppool.de) joined #pypy.
----- silence for 37 minutes ----- [12:22] Arafat (~bear@A36bc.a.pppool.de) joined #pypy.
[12:35] hpk (~hpk@merlinux.de) joined #pypy.
[12:44] Gromit (~bear@B74c2.b.pppool.de) left irc: Read error: 110 (Connection timed out)
[12:53] Nick change: Arafat -> Gromit
----- silence for 19 minutes ----- [13:12] Gromit (~bear@A36bc.a.pppool.de) left irc: Read error: 60 (Operation timed out)
----- silence for 1 hr and 30 minutes ----- [14:42] _hannes (ylhhicq@i528C155E.versanet.de) joined #pypy.
----- silence for 58 minutes ----- [15:40] pedronis (pedronis@ratthing-b246.strakt.com) joined #pypy.
[15:44] tea (~tea@cap006-049.kcn.ne.jp) joined #pypy.
----- silence for 18 minutes ----- [16:02] stakkars (~tismer@rosine120.inf.fu-berlin.de) joined #pypy.
[16:10] <stakkars> hi!
[16:18] <stakkars> hpk: are you there?
[16:18] <hpk> yes kind of
[16:19] <stakkars> something weird is happening. I've rewritten my cache stuff to use single files.
[16:19] <stakkars> after the cache has been built once, overything is fine, 1165 tests passed and one failed.
[16:19] <stakkars> but if test_all triggers compilation, I get hit by something:
[16:20] <stakkars> problem is that I need to run flow space with the lib folder in sys.path,
[16:20] <stakkars> which interferes with your testing code.
[16:20] <stakkars> The interesting thing is that this happens after 1131 tests, gives a warning and says one failed.
[16:21] <stakkars> It looks like you are removing something from the path but then something just stops ???
[16:22] <stakkars> here the end of the messages:
[16:22] <stakkars> [D:\pypy\dist\pypy\lib\test2\conftest.py:30]
[16:22] <stakkars> [testcode : D:\pypy\dist\pypy\lib:2147483648]
[16:22] <stakkars> [modulepath: ]
[16:22] <stakkars> - - - - - - - - - - - - - - - recorded stdout - - - - - - - - - - - - - - -
[16:22] <stakkars> warning, D:\pypy\dist\pypy\lib on sys.path at cpython level, removing it
[16:22] <hpk> i have a look into that later, i am busy right now (a friend is here)
[16:23] <stakkars> thanks. I think I'll check this in, anyway, with the error. Otherwise you cannot reproduce it.
[16:23] <stakkars> quick question: how can I trick the import to allow me to have lib in the path while I'm flowing?
[16:31] tea (tea@cap006-049.kcn.ne.jp) left #pypy.
[16:36] <stakkars> ok, I think I got it:
[16:36] <stakkars> in some situation you remove lib from sys.path, then one of my imports failes and things stopp.
[16:36] <stakkars> but the problem is that I don'z get my error message, but your warning instead.
[16:43] arre (ac@ratthing-b3fa.strakt.com) left irc: Remote closed the connection
----- silence for 39 minutes ----- [17:22] idnar (mithy@idnar.user) left irc: "kbye"
----- silence for 1 hr and 6 minutes ----- [18:28] <hpk> stakkars: hi i am kind of back now (friend is gone)
[18:28] <hpk> stakkars: indeed, pypy/lib is only supposed to be imported at app level
[18:29] <stakkars> sure. But I could work around it, actually my wayof restoring the path was wrong and your message was right.
[18:29] <stakkars> What's not right that I actually got an import error, but your warning message instead.
[18:32] <pedronis> hpk: hi, it's there a problem with pypy-svn, also the archive is stuck at rev 10949 and there have been commits after that
[18:33] Action: hpk goes to check ...
[18:33] <hpk> i at least got non-pypyp commit mails recently
[18:34] <pedronis> strange, http://codespeak.net/pipermail/pypy-svn/2005-April/date.html, for example no sign of Christian commit
[18:35] <hpk> i see
[18:35] <hpk> ok, there must be something wrong with respect to mail sending then ...
[18:37] <pedronis> I'm leaving for now, see you
[18:37] pedronis (pedronis@ratthing-b246.strakt.com) left irc: "Chatzilla 0.9.67 [Firefox 1.0.2/20050325]"
[18:41] <hpk> stakkars: so you need pypy/lib in sys.path at cpython level or did i misunderstand something?
----- silence for 20 minutes ----- [19:01] idnar (mithrandi@idnar.user) joined #pypy.
----- silence for 17 minutes ----- [19:18] tic (~vision@c-ba6e73d5.019-35-67626717.cust.bredbandsbolaget.se) left irc: Read error: 54 (Connection reset by peer)
[19:30] tic (~vision@c-ba6e73d5.019-35-67626717.cust.bredbandsbolaget.se) joined #pypy.
[19:39] tic (~vision@c-ba6e73d5.019-35-67626717.cust.bredbandsbolaget.se) left irc: Read error: 113 (No route to host)
[19:49] tic (~vision@c-ba6e73d5.019-35-67626717.cust.bredbandsbolaget.se) joined #pypy.
[19:54] tic (~vision@c-ba6e73d5.019-35-67626717.cust.bredbandsbolaget.se) left irc: Read error: 104 (Connection reset by peer)
[19:55] tic (~vision@c-ba6e73d5.019-35-67626717.cust.bredbandsbolaget.se) joined #pypy.
----- silence for 23 minutes ----- [20:18] <stakkars> hpk: yes,but I'm doing this "right", now. you don't get involved.
[20:18] <hpk> i freely admit that the removal thingie is a gross hack
[20:18] <stakkars> the remaining problem is, as I mentioned, that I got an import error, but instead of the message I got your warning.
[20:19] <stakkars> what I did around my code generation was
[20:19] <stakkars> hold = sys.path ; sys.path.insert(xxx)
[20:19] <stakkars> then my stuff
[20:19] <stakkars> then sys.path = hold
[20:19] <stakkars> and *that was bad because I forgot to makeacopy.
[20:20] <stakkars> your hack is fine. the error handling is just a bit buggy.
[20:20] <hpk> the hack is only neutralizing another hack, actually :-)
[20:20] <hpk> (i mean my hack is neutralizing another hack of mine)
----- silence for 34 minutes ----- [20:54] <stakkars> hpk: again: this is not about that hack. it is about error handling if I cannot import, your message wins!
[20:54] <hpk> :-)
[20:55] Gluon (~patrick@d149-67-55-168.col.wideopenwest.com) joined #pypy.
[20:55] <Gluon> hi
[20:56] <hpk> stakkars: but let me please continue to dislike my hack which is - at least - strange as all sys.path hacks are kind of strange
[20:56] <hpk> Gluon: hi
[20:58] <stakkars> hpk: fine. I just wanted to make sure that you see what I think needs fixing
[20:58] <Gluon> I had the idea of making a Python system for CL yesterday (I hadn't heard of it before). My motivation was using something like CMUCL for performance, and also to have python modules available from CL. I have just been doing some web browsing and came across PyPy.
[20:58] <hpk> ah!
[20:58] <hpk> yes, indeed sanxiyn (a southkorean guy) has been hacking on a CL backend for pypy
[20:59] <hpk> it is not uptodate with our current type inference techniques IIRC
[21:00] <Gluon> Does pypy have a public repository?
[21:00] <hpk> sure
[21:00] <hpk> did you see http://codespeak.net/pypy/index.cgi?doc
[21:00] <Gluon> I missed it on the web site
[21:00] <Gluon> thanks
[21:00] <hpk> and the getting-started link?
[21:00] <Gluon> ?
[21:00] <Gluon> no
[21:01] <hpk> sorry, what do you mean with 'no' here?
[21:01] <hpk> ah, ok, no to the question, sorry
[21:02] <Gluon> I didn't see those, but now I see the getting-started link thanks.
[21:02] <hpk> we updated and cleaned up the documentation a few days ago ...
[21:02] <Gluon> heh. yeah I'm just blind.
[21:03] <Gluon> (not really)
[21:03] <Gluon> I was thinking that implementations like CMUCL have a lot of type inferencing already in them.
[21:04] <Gluon> but I guess that doesn't fit with other target languages
[21:05] <hpk> right, we are basically targetting more lower level languages
[21:06] <hpk> but for CMUCL one may take a different approach
[21:06] <Gluon> so does sanxiyn stop by this channel?
[21:06] <hpk> yes, usually he does
[21:06] <Gluon> cool
[21:06] <hpk> otherwise you could just ask on pypy-dev@codespeak.net for some more background on the current clisp/backend code
[21:07] <stakkars> hpk: did you like the new _cache thing?
[21:07] <Gluon> thanks. I'll give pypy a spin here so I know something about what I'm asking about first.
[21:08] <hpk> Gluon: great
[21:08] <hpk> stakkars: i haven' looked at it yet (the commit messages only arrives an hour ago when Samuele told me there was a problem, btw)
[21:09] <hpk> but now that you talk about it
[21:09] <hpk> i have done a hack with armin that allows to do "virtual remote imports" over py.execnet
[21:09] <hpk> this is a nice possibly building block for doing remote testing
[21:10] <hpk> you generate .py's in the filesystem and import them, right?
[21:11] <stakkars> yes,almost like before with the one file.ö Now they are many, and I don't get duplicates on recursive testing,also not all is imported at once.
[21:12] <stakkars> and you can do specific testing by removing some or all. default action is self-destruct :-)
[21:12] <stakkars> virtual remote import sounds funny
[21:13] <hpk> it really works and it is quite easy
[21:13] <hpk> it works as least as long as you don't inspect a module's __file__ that is :-)
[21:13] <stakkars> you mean you would do remote testing without remotely installing pypy?
[21:13] <hpk> exactly
[21:14] <hpk> it even works for the py lib with its special importing/packaging techniques
[21:14] <stakkars> and no python?
[21:14] <hpk> you need python on the remote end
[21:14] <hpk> i didn't get around to rewrite Python in Bash yet :-)
[21:14] <stakkars> that's bad
[21:14] <Gluon> haha
[21:15] <hpk> other than an ssh-login and a python version to run you need nothing (nada) on the remote end
[21:16] <hpk> especially no preinstalled software or service
[21:16] <stakkars> I'm wondering how to do the scheduling. I would need to use python clients which run on all the testing
[21:17] <stakkars> machines and try to "get the next working package"
[21:17] <stakkars> from the server. Well,might work.
[21:18] <stakkars> in a sense, the problem seems to be reverse. Instead of pushing to many servers, I want them to pull from one.
[21:19] <hpk> but then you have to schedule on all the clients instead of just one server
[21:19] <hpk> however, i do see uses for both directions
[21:19] <hpk> but here i am first talking about interactive usage aka:
[21:19] <hpk> py.test --remotessh=hpük@codespeak.net pypy
[21:21] <hpk> for the other directions i am considering some '--sendreport' option where a detailed report is sent to a central server
[21:21] <hpk> which can then display results by platform/python-version
[21:23] <hpk> the _cache commit seems nice, btw
[21:23] <stakkars> well, scheduling on the clients is the best possible load-balancing scheme I ever found
[21:23] <stakkars> thx
[21:25] <hpk> i think both scenarios have their uses
[21:25] <stakkars> python in bash, well, that would be a nice bootstraping target!
[21:25] <hpk> i wonder how large files bash can take, though :-)
[21:27] <stakkars> well, we could download a python dist and install it,locally, then run the tests for instance
[21:27] <stakkars> on the other hand, checking svn out on every machine would do just fine, and you could do this via execnet as well
[21:27] <hpk> :-)
[21:37] <Gluon> What does it mean when it says it is faking a type?
[21:38] <stakkars> then this type wasn't implemented in pypy, yet,and the real one is stolen from C python
[21:38] <Gluon> ok I was starting to guess that. but I wasn't sure. thanks.
----- silence for 1 hr and 18 minutes ----- [22:56] stakkars (~tismer@rosine120.inf.fu-berlin.de) left irc:
[23:11] Gluon (patrick@d149-67-55-168.col.wideopenwest.com) left #pypy.
[23:15] Gluon (~patrick@d149-67-55-168.col.wideopenwest.com) joined #pypy.
[23:15] <Gluon> Well it looks cool hpk
[23:16] <hpk> :-)
[23:16] <hpk> i just spent an hour improving the documentation because i reread it myself :-)
[23:16] <Gluon> The object spaces kinda makes my head spin
[23:16] <Gluon> but then I skipped lunch :)
[23:16] <hpk> we sometimes think we should just have called them "object library"
[23:17] <hpk> i think it's the 'space' that makes spins off heads :-)
[23:18] <Gluon> so if you replace all the core C modules with just python, what about all the user C modules?
[23:18] <hpk> we could build a bridge to extension modules later
[23:18] <Gluon> ok
[23:18] <hpk> we need some kind of foreign function interface anyway
[23:18] <Gluon> yeah
[23:19] <Gluon> I'll take a look into that, although I have a bunch to learn
[23:20] <hpk> i guess you know about CTypes
[23:20] <Gluon> I came across it yesterday
[23:20] <Gluon> haven't used it
[23:20] <hpk> lots of new stuff these days, eh? :-)
[23:20] <Gluon> yep.
[23:22] <Gluon> I see gencl.py in the translator dir. Is that what you were referring to before?
[23:22] <hpk> yes
[23:24] <hpk> the test/test_cltrans.py still passes
[23:24] <hpk> but it is not as advanced as the C or LLVM backend
[23:36] <Gluon> To properly run the tests in test_all.py I can just do python test_all.py right? I don't have to do python interpreter/py.py testall.py ... do I?
[23:36] <hpk> no, just test_all.py
[23:36] <Gluon> thanks
[23:39] <Gluon> If I would like to contribute a C module translated to pure python, is there a priority list of the remaining ones?
[23:46] <Gluon> well I'm gonna get some dinner. Thanks for putting up with my pestering hpk. Ciao.
[23:46] <hpk> see you!
[23:46] Gluon (patrick@d149-67-55-168.col.wideopenwest.com) left #pypy.
[00:00] --- Fri Apr 22 2005