[00:45] dialtone (~dialtone@host27-8.pool80116.interbusiness.it) left irc: "Leaving"
[00:49] dialtone (~dialtone@host27-8.pool80116.interbusiness.it) joined #pypy.
----- silence for 1 hr and 36 minutes ----- [02:25] fredrik (fredrik@c83-248-135-181.bredband.comhem.se) left irc: "http://fredrikj.net";
----- silence for 23 minutes ----- [02:48] dialtone (~dialtone@host27-8.pool80116.interbusiness.it) left irc: "This computer has gone to sleep"
----- silence for 5 hr and 14 minutes ----- [08:02] Rhamphoryncus (~rhamph@rhamphoryncus.user) left irc: ""What do you mean, how many hit points do I have?" || Support ISO 8601! http://www.cl.cam.ac.uk/~mgk25/iso-time.html";
[08:15] adim (~adim@logilab.net2.nerim.net) joined #pypy.
----- silence for 31 minutes ----- [08:46] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.
[09:00] ericvrp_ (~chatzilla@ericvrp.demon.nl) joined #pypy.
[09:01] ericvrp_ (~chatzilla@ericvrp.demon.nl) left irc: Client Quit
----- silence for 39 minutes ----- [09:40] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "Eject! Eject! Eject!"
----- silence for 20 minutes ----- [10:00] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.
----- silence for 1 hr and 5 minutes ----- [11:05] dialtone (~dialtone@host27-8.pool80116.interbusiness.it) joined #pypy.
----- silence for 18 minutes ----- [11:23] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "Eject! Eject! Eject!"
[11:35] arigo (~arigo@ratthing-b409.strakt.com) joined #pypy.
[11:37] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.
[11:38] Nick change: hpk_away -> hpk
[11:51] <ericvrp> arigo: goodmorning
[11:51] <arigo> morning
[11:51] <arigo> aleale: are you here?
[11:52] <ericvrp> I am having problems with inplace_pow
[11:53] <arigo> it is a 2-arguments operation only, unlike pow
[11:53] <ericvrp> ok
[11:53] <arigo> (the number of arguments is the 3rd column of the ObjSpace.MethodTable in baseobjspace.py)
[11:55] <ericvrp> I don't think the problem is in rint.py
[11:56] <ericvrp> with i **= n+10 I get the messsage TypeError: inplace_pow() takes exactly 2 arguments (1 given)
[11:58] <arigo> aaah
[11:58] <arigo> it's inplace_pow, not rtype_inplace_pow...
[11:59] <arigo> the annotator is complaining
[11:59] <arigo> the annotator has the bug of the unexpectedly missing 3rd argument in inplace_pow :-)
[12:00] <ericvrp> that's what I figured (I notices that rtype_ was not there)
[12:03] <arigo> the inplace_pow in question is in annotator/binaryop.py
[12:03] <arigo> I fixed it is take only 2 args and provide a SomePBC({None: True}) (i.e. a constant None) to pow()
[12:03] <arigo> writing a test now...
[12:03] <arigo> (always a good policy to write tests)
[12:06] <ericvrp> about writing test: I have to acquire that habbit still :(
[12:06] <arigo> btw, unlike what I suggested, it's probably not a good idea to replace pow(a,b,c) with d=pow(a,b);mod(d,c)
[12:06] <arigo> because of overflow handling
[12:09] <ericvrp> ok
[12:10] <arigo> (checked in)
[12:10] <hpk> hi all
[12:10] <ericvrp> merci
[12:10] <arigo> de rien :-)
[12:10] <arigo> hpk: good morning
[12:10] <ericvrp> hpk: hi
[12:12] <ericvrp> arigo: have you added the test also, can't find it?
[12:12] <ericvrp> never mind
[12:12] <arigo> the annotator tests are in translator/test/ for some reason
[12:13] <ericvrp> makes sense: expected it in annotation/test
[12:13] <arigo> well it makes sense to expected it there :-/
[12:14] <hpk> if nobody tackles the ll-cfg-interpreter then i am going to give it a try next week i guess
[12:16] <arigo> ok
[12:16] <hpk> but i am fine if somebody else can't wait :-)
[12:16] yuuh (ore@82.140.25.212) joined #pypy.
[12:17] <hpk> arigo: are you also relaxing a bit now that you don't escalate so much with Samuele, btw? :-)
[12:17] <arigo> hpk: actually Christian's idea of several people writing the ll-interpreter makes sense
[12:17] <hpk> i thought so too
[12:17] <arigo> hpk: relaxing: yes :-)
[12:18] <arigo> trying to get back to a normal day schedule
[12:18] <hpk> same here, espcecially since the summer now apparently really started
[12:19] <hpk> it's almost too hot to be outside with cloths
[12:19] <arigo> :-)
[12:19] <arigo> still springy around here
[12:19] <hpk> but also, i am somehow pretty confident after our translation session that things will work out ...
[12:19] <hpk> ... nonwithstanding there are many open issues still
[12:21] <hpk> arigo: or do you feel differently?
[12:23] <arigo> not really, that will probably work. the open question is how good the result will be, and how much time we'll have to spend on involved optimizations to get a reasonable performance
[12:29] <hpk> yes, but there seem to be so many optimization possibilities that i am not too worried about it ...
[12:29] <hpk> ... as long as we have a testable decent architecture
[12:29] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) got netsplit.
[12:29] ericvrp (~chatzilla@ericvrp.demon.nl) got netsplit.
[12:30] <arigo> that's a point of view
[12:30] <hpk> yes, my only big worry is that at some point only translating the whole of PyPy and then running things becomes the main way of working
[12:30] <arigo> another one is that there are so many gross inefficiencies right now that we'll be horribly slow before we address on most of them
[12:32] <hpk> yes, but it's also a research project so if we don't fullfill speed goals as promised i don't consider this neccesarily fatal
[12:32] <hpk> unless we are stuck at running 100 times slower than CPython :-)
[12:33] <hpk> hum, did we even promise any speed goals for september?
[12:33] <arigo> actually, my bet is around 10, but then again we need to know
[12:35] ericvrp (~chatzilla@ericvrp.demon.nl) got lost in the net-split.
[12:35] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) got lost in the net-split.
[12:35] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) joined #pypy.
[12:35] ericvrp (~chatzilla@ericvrp.demon.nl) joined #pypy.
[12:35] <ericvrp> arigo: I will remove the pow;mod thing shortly but still would like to know how I can let pow call mod....in type_pow(...) I have put: v_result = rtype_mod((v_result, s_int3)) , but v_result is of type pypy.objspace.flow.model.Variable while s_int should be pypy.rpython.rint.SomeInteger. How do I cast this?
[12:35] <ericvrp> Don't you just love newbies ;)
[12:36] <arigo> ah, that's difficult
[12:36] <arigo> instead, you probably need to generate the mod operation manually with a direct_op('int_mod', ...)
[12:36] <arigo> each rtype_xyz() can generate any number (including zero) operations this way
[12:36] <ericvrp> But then I would have to duplicate the mod code
[12:37] <arigo> or factor out some common part...
[12:37] <arigo> but I see the point
[12:38] <arigo> the problem is that we really don't have a "SomeXyz" value for the intermediate v_result to give to rtype_mod
[12:38] <arigo> additionally, rtype_mod uses receive(..., arg=0/1)
[12:38] <ericvrp> I guess I do know the resulttype of v_result
[12:39] <arigo> the resulttype is Signed or Unsigned, which is not a "SomeXyz" value suitable for s_xyz arguments...
[12:39] Action: arigo leaves for lunch
[12:39] <ericvrp> but I can create I SomeXyz in the if
[12:39] <arigo> (group of people leaving, sorry to hurry away)
[12:39] <ericvrp> thank youm bye
[12:40] <arigo> (the other problem is the receive())
[12:45] ericvrp (~chatzilla@ericvrp.demon.nl) got netsplit.
[12:45] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) got netsplit.
[12:47] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) returned to #pypy.
[12:47] ericvrp (~chatzilla@ericvrp.demon.nl) returned to #pypy.
[12:52] hpk (~hpk@merlinux.de) left irc: Read error: 110 (Connection timed out)
[12:53] hpk (~hpk@merlinux.de) joined #pypy.
----- silence for 21 minutes ----- [13:14] ericvrp (~chatzilla@ericvrp.demon.nl) got netsplit.
[13:14] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) got netsplit.
[13:16] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) returned to #pypy.
[13:16] ericvrp (~chatzilla@ericvrp.demon.nl) returned to #pypy.
[13:28] froosch2 (~frosch@a81-14-145-253.net-htp.de) joined #pypy.
[13:31] froosch2_ (~frosch@a81-14-145-253.net-htp.de) left irc: Read error: 60 (Operation timed out)
----- silence for 18 minutes ----- [13:49] Rhamphoryncus (~rhamph@rhamphoryncus.user) joined #pypy.
----- silence for 50 minutes ----- [14:39] <adim> Hi everyone
[14:39] <adim> Is SVN down ?
[14:41] <arigo> hi
[14:41] <hpk> adim: hi, yes, i am fixing it ...
[14:41] <hpk> ... i think we will make some changes in june to get rid of this problem
[14:41] <adim> hpk: ok, great
[14:41] <adim> oh, is it a recurrent problem ?
[14:42] <hpk> yes
[14:42] <adim> I noticed it was down two or three days ago
[14:42] <hpk> somehow the various accesses to svn manage to mix something up
[14:42] <hpk> there is quite some pressure on codespeak's svn these days
[14:42] <hpk> mostly because the search engines are continously scanning it
[14:43] <adim> I see
[14:43] <arigo> hpk: which means that branches and tags have a "non-constant-time" effect here
[14:43] <hpk> arigo: indeed
[14:43] <hpk> but i am hesitant to restrict the search engines from the repo
[14:43] <arigo> is this a known problem with http views of svn?
[14:43] <hpk> no, it's not really a problem
[14:44] <arigo> I mean, did someone already think about ways of restricting search engines, say, to trunks in the repo?
[14:44] <hpk> (it just that it generates lots of accesses)
[14:44] <hpk> arigo: haven't looked
[14:44] <hpk> last time i did the suggestion was robots.txt blocking the thing
[14:44] <hpk> but maybe generating an appropriate robots.txt could mitigate the problem
[14:45] <hpk> somehow i'd first like to just have it work stable nevertheless :-)
[14:47] <arigo> makes sense :-)
----- silence for 34 minutes ----- [15:21] <arigo> aleale: to run the _codecs CPython tests, should I rename only inprocess__codecs.py, or do I need something else?
[15:35] <ericvrp> arige: you here for a quick Q?
[15:35] <arigo> yes
[15:35] <ericvrp> arigo: you here for a quick Q?
[15:37] <ericvrp> I am working on rfloat.py and would like to do v_float = receive(Float, arg=0) . Float however is not defined as a Primitive in , should I add it or has that sideeffects?
[15:37] <arigo> no, you can add it
[15:38] <arigo> we need a new line in annotation/model.py
[15:38] <arigo> (SomeInteger(), lltype.Signed),
[15:39] <arigo> should add:
[15:39] <arigo> (SomeFloat(), lltype.Float),
[15:39] <arigo> (which must go after the two lines about Signed and Unsigned)
[15:40] <ericvrp> ok, thank you
[15:40] <arigo> this is what allows the annotator to match the SomeFloat annotation with the Float low-level type and vice-versa.
[15:41] <ericvrp> AttributeError: 'float' object has no attribute '_TYPE
[15:41] <arigo> ah yes
[15:42] <arigo> search for Signed in lltype.py, you need to add something similar for Float
[15:42] <ericvrp> ok
[15:43] <ericvrp> one step for mankind...
[15:51] <arigo> I'm considering refactoring a bit the names in rpython/r*.py
[15:52] <arigo> I won't do it while you're editing files, obviously, but I'd like to hear some feedback
[15:53] <arigo> for example, we could rely less on magic global helpers by passing an extra argument to the rtype_xyz methods, just after the 'self' -- 'self' might be a pair so it means after the pair, in this case
[15:54] <arigo> something like rtype_add((s_int1, s_int2), context)
[15:54] <arigo> with a better name than 'context'
[15:56] FiSh3rWoLf`2 (~ciw0-1@2001:5c0:8fff:fffe:0:0:0:2959) joined #pypy.
[15:56] <arigo> to do context.receive() on it
[15:57] <arigo> maybe 'highlevelop', and receive(Signed, arg=0) becoming 'highlevelop.args[0].as(Signed)' or something like this
[15:58] <arigo> I'm keeping the case of 'pow' wanting to call 'mod' in mind
[15:59] <arigo> which might be done with something like
[16:00] <ericvrp> I would like that, perhaps we can query context for the variables. It would be nice to do v_int1, v_int2 = context.receive(2) , receive could figure out the first variable in most cases.
[16:01] Action: ericvrp will commit in 10 minutes
[16:01] <arigo> it's still important to spell out the exact type we expect from the receive(),
[16:01] <ericvrp> maybe (Signed,Signed) instead of 2
[16:01] <arigo> to force explicit conversions
[16:02] <arigo> specially with Signed vs Unsigned
[16:02] <arigo> makes sense
[16:13] <ericvrp> arigo: I have commited, go ahead
[16:14] FiSh3rWoLf` (~ciw0-1@2001:618:400:0:0:0:5138:d930) joined #pypy.
[16:14] <arigo> ok -- though I guess we need some more thinking first
[16:14] <ericvrp> What is the refactoring objective this time?
[16:15] jacob22 (jacob@enzo.strakt.com) joined #pypy.
[16:15] FiSh3rWoLf` (ciw0-1@2001:618:400:0:0:0:5138:d930) left #pypy.
[16:16] <arigo> it would be to find more meaningful names and maybe get rid of the TLS and the global helpers
[16:19] <ericvrp> If you need a sparringpartner for your thought, it would better be someone who is a little less newbie then me
[16:20] <arigo> sure, sorry to bother you about it
[16:20] <ericvrp> not at all, I would love to follow the discussion and contribute when I can. Am just very humble at this time.
[16:21] <arigo> you did fine with rint.py, rfloat.py etc. so you're welcome to give feedback on what looks strange in practice
[16:22] FiSh3rWoLf`2 (~ciw0-1@2001:5c0:8fff:fffe:0:0:0:2959) left irc: Read error: 60 (Operation timed out)
[16:22] <aleale> arigo: Yes You should rename inprogress__codecs.py to _codecs.py. unicodecodec.py is already in the repo
[16:23] <arigo> aleale: ok
[16:24] <arigo> should I worry about codecs.py too?
[16:25] <aleale> arigo: With the latest inprogres... it should work with lib-python/2.3.4/codecs.py
[16:25] <arigo> great
[16:25] <arigo> btw in py.py I get:
[16:25] <arigo> >>>> 'abcdef'.encode('zip')
[16:25] <arigo> LookupError: ('unknown encoding: %s', 'zip')
[16:26] <arigo> apart from the missing '%' instead of ',' in building the error message,
[16:26] <arigo> is that expected
[16:26] <arigo> ?
[16:26] <arigo> oooups sorry. yes, 'import zlib' doesn't work in py.py
[16:28] <aleale> arigo: You had me worried for a minute there - I havent tried all the encodings in encodings
[16:28] <arigo> 'abc'.encode('hex') works as expected, so I guess the whole things basically works (that's just what I wanted to try :-)
[16:29] <arigo> starting the CPython tests now...
[16:33] <aleale> I expect to take a long time (test_unicode took 18 hours)
[16:33] <arigo> argh I see
[16:34] <arigo> hum, test_unicode is currently classified as "non-core" test
[16:35] <arigo> that should probably be changed now, especially if it passed :-)
[16:35] <aleale> ok, I missed that
[16:35] <arigo> well unicode has become core now, I guess.
[16:35] <aleale> I had a few failures
[16:35] <aleale> s/I/it
[16:35] <arigo> mostly passing is already great
[16:36] <arigo> I hope you saved the results somewhere :-)
[16:37] <aleale> I had an accident so no ;-(
[16:39] <aleale> Are you running on your laptop?
[16:39] <aleale> or do you have access to a fast machine?
[16:40] <arigo> no, it's a fast machine
[16:40] <arigo> I still think the core tests should take about 24 hours
[16:41] <arigo> 8 passed tests so far, so no crash-everything kind of bug :-)
[16:43] <aleale> Thats nice - and it wasnt test_unicode it was test_codeccallbacks.py which took 18 ours ( most of the time spend in test_longstrings)
[16:43] <aleale> s/ours/hours
[16:45] <aleale> I wont be able to work much until sunday evening - could you mail the report or post it somewher I can find it ?
[16:46] <arigo> ok, I'll mail you a link to a svn path where I'll have checked in the results
[16:47] <arigo> (I won't check them in the usual place because they would show up on the web test reports and create confusion)
[16:48] <aleale> cool - one of the failures in test_codeccallback is due to exceptions (UnicodeDecodeError taking 5 arguments and such)
[16:49] <arigo> ah, ok
[16:49] <aleale> I commit something last week to enable those checks - but I havent figured out how to generate the exceptions module
[16:49] <aleale> s/commit/committed
[16:50] <arigo> it's missing a comment at the beginning to say it's auto-generated, and how you can regenerate it
[16:51] <aleale> yes - well I have to go.
[16:51] <arigo> ok, see you
[17:04] <ericvrp> arigo: ok if I do another rbool.py and rfloat.py commit now?
[17:04] <arigo> sure, go ahead
[17:05] <arigo> I'll wait for more inspiration another day about the refactoring :-)
[17:06] <ericvrp> done
[17:06] Action: ericvrp leaving to get some dinner
[17:07] <arigo> see you (checkin looks good)
[17:08] senra (~chatzilla@200.172.8.47) joined #pypy.
[17:08] <arigo> we should have no implicit conversion from float to int or bool, though
[17:09] <arigo> only explicit ones using 'is_true' or 'int' ('int' is an explicit operation already at this level)
[17:24] idnar (mithrandi@idnar.user) left irc: Read error: 60 (Operation timed out)
[17:26] idnar (mithrandi@idnar.user) joined #pypy.
[17:32] senra (~chatzilla@200.172.8.47) left irc: Read error: 131 (Connection reset by peer)
[17:32] senra_ (~chatzilla@200.172.8.47) joined #pypy.
[17:32] Nick change: senra_ -> senra
[17:32] aleale (~redorlik@cpe.atm0-0-0-129140.0x3ef2fa3a.bynxx3.customer.tele.dk) left irc: "Eject! Eject! Eject!"
----- silence for 29 minutes ----- [18:01] yuuh (ore@82.140.25.212) left irc: "utz utz utz"
[18:09] arigo (~arigo@ratthing-b409.strakt.com) left irc: Read error: 104 (Connection reset by peer)
----- silence for 43 minutes ----- [18:52] adim (~adim@logilab.net2.nerim.net) left irc: Remote closed the connection
[19:06] senra_ (~chatzilla@200.172.8.47) joined #pypy.
[19:06] senra (~chatzilla@200.172.8.47) left irc: Connection reset by peer
[19:06] Nick change: senra_ -> senra
----- silence for 1 hr and 39 minutes ----- [20:45] senra (~chatzilla@200.172.8.47) left irc: "Chatzilla 0.9.68.5 [Firefox 1.0.4/20050511]"
[00:00] --- Sat May 28 2005