[00:13] yuuh (lpxaqnks@dsl-62-220-8-204.berlikomm.net) left irc: "utz utz utz"
----- silence for 16 minutes ----- [00:29] cfbolz (~cfbolz@hdlb-d9b946c7.pool.mediaWays.net) joined #pypy.
[00:29] <cfbolz> hi!
[00:29] <arigo> hi!
[00:29] <cfbolz> I saw that you worked on test_complex
[00:30] <cfbolz> I tried to fix the last failing test yesterday
[00:30] <cfbolz> how can I check whether float(str) returns inf?
[00:31] <arigo> hum
[00:31] <arigo> I don't know if there is a standard way
[00:32] <cfbolz> that's what the failing test checks:
[00:32] <cfbolz> it calls complex("1" * 800) and expects it to raise a type error
[00:43] fredrik (fredrik@c83-248-135-181.bredband.comhem.se) joined #pypy.
[00:43] <arigo> cfbolz: TypeError?
[00:49] <arigo> it's a bit undefined what occurs in Python when a float() call overflows
[00:49] <arigo> I think it can raise an exception on some platforms
[00:49] <arigo> in pypy, it raises OverflowError and doesn't return 'inf', btw
[00:51] <arigo> ah no,
[00:51] <arigo> it returns an infinite float, but you get an OverflowError whenever you try to print it
[00:51] <arigo> I guess that's not good
[00:58] arigo (~arigo@c-3a8b70d5.022-54-67626719.cust.bredbandsbolaget.se) left irc: "[x]chat"
[01:09] cfbolz (~cfbolz@hdlb-d9b946c7.pool.mediaWays.net) left irc: Read error: 113 (No route to host)
[01:13] stakkars (jpvojlb@dsl-62-220-8-204.berlikomm.net) joined #pypy.
[01:15] <stakkars> hpk: still wake?
[01:15] <hpk> yes, kind of
[01:16] <stakkars> what is the reason why the integer divisions can raise OverflowError?
[01:16] <stakkars> just the specialcase of (-maxint-1) // -1 ?
[01:17] <hpk> don't know
[01:18] <pedronis> stakkars: yes, (-maxint-1)//-1 is an overflow
[01:18] <stakkars> I just checked it --- yes
[01:20] <stakkars> pedronis: I augmented targetpypy1 to test a couple of operations, and I tried to invoke an overflow.
[01:21] <stakkars> pedronis: but there are no w_Exception things implemented. I guess this is by intention, to avoid pulling too much in?
[01:21] <pedronis> no
[01:22] <pedronis> it's probably just that the code is not doing completely the right thing wrt to PBCs attributes
[01:22] <pedronis> the w_* are atttribute on the space that is a PBC
[01:22] <stakkars> I tired to provoke OverflowError generartion, but with no success.
[01:23] <stakkars> ahh!
[01:23] <stakkars> I also tried thingwith -no-a, which generates almost all of the interpreter, but then we have the
[01:24] <stakkars> problem again that we had before -- there is something referenced in the wrong order, and we crash
[01:24] <stakkars> since this takes hours to compute, I ignored it so far.
[01:26] <stakkars> targetpypy1 handles mul, add, sub by now.WOrking on division. Should I check in?
[01:27] <pedronis> yes
[01:27] <stakkars> flow space is a great tool, btw! found two real bugs in longtype.py by it
[01:28] <fredrik> is pypy as inconsistent as cpython when it comes to 10e300**10 vs 10e300*10e10
[01:28] <stakkars> fredrik: why not ask PyPy, yourelf?
[01:29] <stakkars> btw.I guess it is, because we are faking float from CPython.
[01:30] <pedronis> well for float the term is more reusing than faking
[01:30] <stakkars> sorry, I'm partially a bit behind. The target is moving *very* fast
[01:30] <fredrik> it seems pypy raises an exception for both
[01:31] <fredrik> unlike cpython
[01:31] <stakkars> sounds good!
[01:31] <fredrik> it raises different exceptions though
[01:32] <stakkars> OT why are you guys not sleeping right now?
[01:32] <fredrik> isn't that what daytime is for? :)
[01:32] Action: stakkars suffers from the same problem
[01:33] <stakkars> pedronis: I spent quite some time figuring out why the exceptions are not generated and assumed this is
[01:34] <stakkars> because the ocfcheck function is not generated. But it turned out to jnot be the case.
[01:36] <pedronis> well, there are still quite a few missing pieces in genc
[01:36] <stakkars> that's nice to hear. I thought it was my failure.
[01:39] Action: pedronis going to bed
[01:39] pedronis (~Samuele_P@c-3a8b70d5.022-54-67626719.cust.bredbandsbolaget.se) left irc: "Chatzilla 0.9.68a [Firefox 1.0.2/20050317]"
[01:40] <stakkars> checked in
[01:41] <stakkars> too bed :-)
----- silence for 19 minutes ----- [02:00] <stakkars> question to all who are still alive/awake! :-)
[02:01] <stakkars> why does objspace/flow(operation.py
[02:02] <stakkars> exec compile2("""
[02:02] <stakkars> def inplace_truediv(x, y):
[02:02] <stakkars> x /= y
[02:02] <stakkars> return x
[02:02] <stakkars> """, flags=__future__.CO_FUTURE_DIVISION, dont_inherit=1)
[02:02] <stakkars> # makes an INPLACE_TRUE_DIVIDE
[02:12] <stakkars> boohoo!
[02:13] <stakkars> why does floordiv pull in almost allof the interpreter???
[02:18] dialtone (~dialtone@host111-56.pool80117.interbusiness.it) left irc: "This computer has gone to sleep"
----- silence for 1 hr and 55 minutes ----- [04:13] fredrik (fredrik@c83-248-135-181.bredband.comhem.se) left irc: "http://fredrikj.net";
----- silence for 18 minutes ----- [04:31] esteban (~esteban@DWM-203-228.go.retevision.es) joined #pypy.
[04:40] stakkars (jpvojlb@dsl-62-220-8-204.berlikomm.net) left irc: Read error: 60 (Operation timed out)
----- silence for 2 hr and 2 minutes ----- [06:42] esteban (~esteban@DWM-203-228.go.retevision.es) left irc:
----- silence for 33 minutes ----- [07:15] idnar (mithrandi@idnar.user) left irc: Read error: 110 (Connection timed out)
[07:25] sanxiyn (tinuviel@211.104.100.144) joined #pypy.
----- silence for 59 minutes ----- [08:24] dialtone (~dialtone@host111-56.pool80117.interbusiness.it) joined #pypy.
----- silence for 1 hr and 7 minutes ----- [09:31] arre (ac@ratthing-b3fa.strakt.com) joined #pypy.
----- silence for 47 minutes ----- [10:18] aleale (~chatzilla@0x535ca5a5.hrnxx9.adsl-dhcp.tele.dk) joined #pypy.
[10:32] aleale (~chatzilla@0x535ca5a5.hrnxx9.adsl-dhcp.tele.dk) left irc: Read error: 104 (Connection reset by peer)
----- silence for 26 minutes ----- [10:58] sanxiyn (tinuviel@211.104.100.144) left irc: "Bye"
[11:06] aleale (~chatzilla@0x535ca5a5.hrnxx9.adsl-dhcp.tele.dk) joined #pypy.
[11:13] ludal (~ludal@logilab.net2.nerim.net) joined #pypy.
----- silence for 23 minutes ----- [11:36] <hpk> morning all
[11:37] pedronis (pedronis@ratthing-b246.strakt.com) joined #pypy.
[11:37] aleale (~chatzilla@0x535ca5a5.hrnxx9.adsl-dhcp.tele.dk) left irc: Read error: 104 (Connection reset by peer)
----- silence for 31 minutes ----- [12:08] <hpk> pedronis: i refactored the test reporting last night ... there is no quickreport anymore ...
[12:09] <hpk> instead http://codespeak.net/~hpk/pypy-testresult/ is regenereated (in seconds) on each testresult-commit
[12:10] <hpk> you can also generate html pages locally with 'genreportdata.py' which will give you an index.html and .cache full of per-run html files
----- silence for 18 minutes ----- [12:28] <pedronis> hpk: hi. I see.
[12:28] aleale (~chatzilla@0x535ca5a5.hrnxx9.adsl-dhcp.tele.dk) joined #pypy.
[12:28] <pedronis> are some tests missing? I don't see test_new for example?
[12:28] <hpk> i probably didn't run it
[12:29] <hpk> (all the old test reports are ignored)
[12:32] <hpk> hum, some tests are failing because we are running verbose mode
[12:43] arigo (~arigo@c-3a8b70d5.022-54-67626719.cust.bredbandsbolaget.se) joined #pypy.
[12:46] dialtone (~dialtone@host111-56.pool80117.interbusiness.it) left irc: "This computer has gone to sleep"
[12:59] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) joined #pypy.
[12:59] <cfbolz> hi!
[12:59] <arigo> hi!
[13:00] <arigo> I'm writing one line about LLVM for the talk proposal,
[13:00] Action: hpk hi's around
[13:00] <arigo> does this look good to you:
[13:00] <cfbolz> ok. what are you writing?
[13:00] <arigo> - the LLVM back-end (targetting a virtual machine with a different model and
[13:00] <arigo> different trade-offs than plain C).
[13:00] <cfbolz> yes. that's fine.
[13:01] <cfbolz> btw: should I bring genllvm into any particular shape before the release?
[13:02] <arigo> probably not
[13:02] <arigo> the release will not focus on the translator
[13:02] <cfbolz> yes, I know.
[13:02] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) left irc: Client Quit
[13:03] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) joined #pypy.
----- silence for 22 minutes ----- [13:25] yuuh (gtrivq@82.140.29.224) joined #pypy.
[13:32] <cfbolz> would it be worth to really wait for the tests that timeout to finish (I could just run them on a server here)?
[13:34] <arigo> cfbolz: yes, probably, but:
[13:34] <arigo> some of them might really take ages
[13:34] <cfbolz> what's ages: a week?
[13:34] <arigo> and we don't have so far a nice way to avoid anyone else running the test after you to overwrite the result with a timeout result
[13:35] <arigo> "ages" might not be more than a day, though
[13:35] <arigo> usually it's better to see if there is an obvious very long loop in the test, and make it shorter.
[13:36] <cfbolz> ok. I think I'll just do a test run with a really high number of pystones
[13:37] <cfbolz> how do I do that now?
[13:37] <arigo> --timeout=
[13:37] <arigo> I think the default is --timeout=100mp
[13:37] <cfbolz> no, the whole command
[13:37] <arigo> cd lib-python/2.3.4/test
[13:37] <arigo> py.test --timeout=1000mp
[13:37] <cfbolz> thanks. I couldn't keep track of the recent changes
[13:38] <arigo> sure, it was changing daily :-)
[13:38] <arigo> the above command only run the tests flagged "core", which are the ones we are interested in passing at the moment
[13:40] <cfbolz> ok. it generates a cfbolz@whatever directory in result, right?
[13:40] <arigo> yes
[13:41] <arigo> you have to check out something as specified in the message after you try py.test first
[13:41] <arigo> then you can check in the new cfbolz@xyz directory at any time
[13:41] <cfbolz> ok
[13:42] <cfbolz> I get an ImportError: No module named test.regrtest when doing py.test in lib-python/2.3.4/test
[13:46] <cfbolz> seems to be a problem with python 2.4 only, it works with python 2.3
[13:50] <cfbolz> ok, have to go.
[13:50] <cfbolz> see you later
[13:50] <hpk> see you and beware of python 2.4 :-)
[13:51] <cfbolz> yes. I notice that quite regularly
[13:51] <cfbolz> regexes in PyPy don't work under Python 2.4 due to the changing of some magic numver in _sre
[13:51] <hpk> yes
[13:52] <hpk> after the may release we want to move to python2.4 as quick as possible
[13:52] <cfbolz> good. bye
[13:52] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) left irc: "Leaving"
----- silence for 23 minutes ----- [14:15] ludal (~ludal@logilab.net2.nerim.net) left irc: "Download Gaim: http://gaim.sourceforge.net/";
[14:17] ludal (~ludal@logilab.net2.nerim.net) joined #pypy.
[14:26] arigo (~arigo@c-3a8b70d5.022-54-67626719.cust.bredbandsbolaget.se) left irc: "[x]chat"
[14:28] dialtone (~dialtone@host118-0.pool21345.interbusiness.it) joined #pypy.
----- silence for 33 minutes ----- [15:01] stakkars (vhvpum@82.140.29.224) joined #pypy.
[15:03] <stakkars> hi
[15:04] <stakkars> did we plan for some meeting, today?
[15:06] arigo (~arigo@ratthing-b401.strakt.com) joined #pypy.
[15:09] <arigo> hpk: ping?
[15:10] <stakkars> hi, do we have a date at 3pm, or was it another day?
[15:10] <hpk> stakkars: on the 4th IIRC
[15:10] <hpk> arigo: wait a moment
[15:25] <hpk> arigo: yes?
[15:26] <stakkars> arigo: I have a few small questions
[15:26] <arigo> the results shown in the html test report have several oddities
[15:26] <arigo> e.g. test_new is nowhere to be found
[15:27] <arigo> test___all__ times out after 14 seconds
[15:27] <arigo> test_longexp failed strangely (but works just fine when I try it on various machines)
[15:28] <hpk> let's discuss one by one
[15:28] <hpk> test_new was probably just not run by me
[15:28] <hpk> note that all old test results are not parsed
[15:28] <hpk> only testreport-version >= 1.1
[15:29] <arigo> sure
[15:29] <arigo> how did you run the tests then?
[15:29] <hpk> don't know, it was late :-)
[15:29] <arigo> ah
[15:29] <hpk> well, probably 'py.test' simply
[15:30] <arigo> then it's strange
[15:30] <hpk> and note that KeyboardInterrupts are noted as Timeouts
[15:30] <arigo> so you probably Ctrl-C'ed test___all__?
[15:30] <hpk> so if you issue 'py.test' and then hit control-c you will get that reported as a timeout
[15:30] <hpk> arigo: probably
[15:30] <hpk> note that py.test does not overwrite a non-timeout result with a time-out one anymore
[15:30] <arigo> ah, good
[15:30] <hpk> so once we have non-timeout results they should stick
[15:31] <arigo> can you check if test_new.txt is here but not checked in, or something like that?
[15:33] <hpk> it wasn't there i think
[15:33] <stakkars> arigo: why does floordiv pull almost the whole interpreter in, while div does not?
[15:34] <arigo> stakkars: --verbose please?
[15:34] <stakkars> I extended targetpypy1 to pick MM operations in a loop and to call them all.
[15:35] <arigo> hpk: also, it was nice to see the last line in the html summary for crashing tests
[15:35] <stakkars> this works fine for mul, add, sub, div
[15:35] <hpk> arigo: i am going to lunch right now, just change it in genreportdata.py
[15:35] <stakkars> but when I try floordiv, I get almost all of the interpreter.
[15:35] <hpk> arigo: i added a test_new run
[15:36] <arigo> hpk: ok, the real question was to understand why it was missing and if other tests are missing
[15:37] <hpk> arigo: i wouldn't think so, i guess i just didn't systematically run all the tests
[15:37] <arigo> samuele is regenerating all core tests too, we'll check if more things are missing afterwards
[15:37] <arigo> stakkars: no clue...
[15:37] <hpk> arigo: i have an 11 hour test run of all core tests which hopefully finishes soon (i am running with a high timeout of 1500 MP)
[15:37] <hpk> arigo: you did get the MP thingie (mega-pystones), right?
[15:37] <arigo> yes
[15:38] <stakkars> what is MP, please?
[15:38] Action: hpk is off
[15:38] <arigo> the time-out when running the tests with py.test in lib-python/2.3.4/
[15:39] <arigo> it's not an amount of seconds, but an amount of "pystone times"
[15:39] <stakkars> hmm
[15:39] <arigo> so 100 MP on fast and slow machines gives approximately the same chance to each test
[15:39] <stakkars> ah, I see.
[15:40] <stakkars> are the tests now distributed over machines, and coordinated? The reports looked like this
[15:41] <stakkars> ok, later
[15:42] Action: stakkars goes into the sun
----- silence for 30 minutes ----- [16:12] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) joined #pypy.
[16:12] <cfbolz> hi!
[16:13] <arigo> hi!
[16:13] <cfbolz> py.test in lib-python/2.3.4/test still doesn't work for me
[16:13] <cfbolz> ImportError: No module named test.regrtest
[16:14] <cfbolz> do you have anything special in your importpath or something?
[16:14] Action: hpk -> back
[16:14] <hpk> cfbolz: no
[16:15] <hpk> cfbolz: which python version are you using?
[16:15] <cfbolz> 2.3.4
[16:16] <cfbolz> I thought it was an 2.4 problem but it was unrelated
[16:17] <hpk> can you login to codespeak and repeat the steps to see if it is your usage pattern or your local setup that is the problem?
[16:18] <hpk> i am a bit at a loss of guessing
[16:18] <hpk> because apparently people are successfully using it on several machines
[16:18] <hpk> arigo: are you changing the "finalline" thingie?
[16:18] <arigo> not at the moment
[16:18] <hpk> ok, then i go for it
[16:19] dialtone (~dialtone@host118-0.pool21345.interbusiness.it) left irc: "This computer has gone to sleep"
[16:19] <arigo> thanks
[16:20] <hpk> arigo: anything else on your immediate wish list or does the html report make sense otherwise?
[16:20] stakkars (vhvpum@82.140.29.224) left irc: Read error: 60 (Operation timed out)
[16:21] <arigo> no, it looks good
[16:25] fredrik (fredrik@c83-248-135-181.bredband.comhem.se) joined #pypy.
[16:29] <hpk> pedronis: your test run looks strange
[16:31] <pedronis> there maybe some old results for non-core tests
[16:32] <hpk> if you look at http://codespeak.net/~hpk/pypy-testresult/
[16:32] <hpk> then all the first tests fail very quickly
[16:32] <hpk> do you have an unmodified tree?
[16:33] <hpk> completely updated, including an updated testresult directory (easy to forget the latter)
[16:34] Action: hpk is probably going to move the testresult code to pypy/tool/testsupport/ and let the testresult only contain test results and nothing else
[16:35] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) left irc: Remote closed the connection
[16:39] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) joined #pypy.
[16:40] <cfbolz> hpk: it works on codespeak.net. It's probably some problem of mine, sorry for the fuss.
[16:40] <hpk> cfbolz: no, it's fine
[16:40] <hpk> cfbolz: it should work on your machine
[16:40] arre (ac@ratthing-b3fa.strakt.com) left irc: "using sirc version 2.211+KSIRC/1.3.11"
[16:40] <hpk> cfbolz: can we do the following:
[16:41] <hpk> cfbolz: issue a 'screen -x hpk/hpk'
[16:41] <hpk> cfbolz: and then login to the machine where the test fails (if you can)
[16:42] <cfbolz> sorry, away for a sec. I'll do that in a few minutes.
[16:43] <hpk> cfbolz: just ping me when you are back
[16:47] <cfbolz> hpk: back
[16:54] <cfbolz> hpk: thanks a lot. It works now.
----- silence for 35 minutes ----- [17:29] <hpk> pedronis: i am pretty sure that your recent tests are borked
[17:30] <hpk> i just reran some and your '0.3 second execution time' errors just run fine, e.g. test_syntax.py
[17:33] <arigo> he's about to svn rm them
[17:34] <arigo> our best guess is that he started working in his WC before the tests finished
[17:34] <hpk> yes, makes sense
[17:35] <hpk> a very interesting piece of info is if errorring test once passed ...
[17:35] <arigo> yes
[17:38] <arigo> indeed.
[17:38] <hpk> i guess that working on improving all this is still more important than the dev-tracker
[17:40] <hpk> do you agree?
[17:40] <hpk> side note: some tests are failing (like test_math.py) which don'T fail on fine grained mode ...
[17:40] <arigo> dev-tracker?
[17:41] <hpk> bug/issue tracker
[17:41] <arigo> ah, yes
[17:41] <hpk> ... because fine-grained switches test_support.verbose off when running output tests
[17:41] <hpk> and for our coarse-grained there is no easy way to do that
[17:41] <hpk> or should we simply copy test_support to modified?
[17:41] <arigo> argh
[17:42] <arigo> so these output tests must run non-verbose, but run verbose if we just start them from the cmdline
[17:43] <hpk> you mean if we start them with 'py.py test_math.py' from the command line
[17:43] <hpk> or with py.test ?
[17:43] <arigo> I'm trying to understand what you mean :-) I mean 'py.py test_math.py' yes
[17:44] <hpk> ok, as you know, we run coarse-grained tests completely unmodified
[17:44] <hpk> so we would have to add a py.py cmdline-switch for that, right?
[17:45] <arigo> I don't know, I'm asking if I understood you correctly saying 'py.py test_math.py' now runs verbosely and the expected output is for non-verbose
[17:46] <hpk> yes
[17:46] <arigo> ok
[17:46] <hpk> let me summarize:
[17:46] <hpk> py.test test_math.py # fails because it runs in verbose mode by default
[17:46] <hpk> py.test -E test_math.py # works because fine-grained can modify test_support.verbose
[17:47] <hpk> py.py test_math.py # gives verbose output
[17:47] <hpk> now coarse-grained cannot modify the remote process's applevel test_support.verbose
[17:47] <hpk> unless you have a better idea than adding a cmdline switch
[17:48] <hpk> am i making sense?
[17:48] <arigo> yes
[17:49] Action: hpk is probably renaming to '-F' for fine-grained soon, "-E" sounds too much like error :-)
[17:50] <arigo> I don't see a nice way to fix the verbose problem
[17:51] <hpk> me neither
[17:51] <arigo> maybe patching test_support to look for command-line options
[17:51] <hpk> argh
[17:51] <hpk> i think adding a py.py cmdline option is the cleanest solution
[17:52] <hpk> --pypy-regrtest-setverbose
[17:52] <arigo> well it's just obscure
[17:52] <arigo> when you see that in py.py --help
[17:53] <hpk> ok, then we should write a regrrun.py script which calls into py.py's mechanisms
[17:53] <hpk> that's not too bad i think
[17:54] Action: hpk hears arigo hum'ming
[17:54] <arigo> I see
[17:55] <arigo> I've done a few more horrible hacks to run the CPython test suite with Psyco
[17:55] <arigo> we could even run py.py regrrun.py test_xyz.py
[17:56] <arigo> where regrrun.py is an app-level thingy doing
[17:56] <arigo> import test.regrtest; test.regrtest.verbose=0
[17:56] <arigo> execfile(sys.argv[1])
[17:56] <hpk> [1:], yes
[17:56] <hpk> s/[1:]//
[17:59] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) left irc: "Leaving"
[18:00] <hpk> arigo: so should i do it this way?
[18:00] <arigo> probably
[18:01] <arigo> I mean, yes I think so
[18:12] <hpk> ok, works
[18:12] <arigo> nice
[18:12] dialtone (~dialtone@host118-0.pool21345.interbusiness.it) joined #pypy.
[18:14] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) joined #pypy.
[18:21] <hpk> cfbolz: if we do the mini-sprint 15th-20th of this month, would you be interested/available to come at all?
[18:22] <cfbolz> I'm very interested -- unfortunately I have another exam on the 24th and I really should study a bit for that
[18:22] <cfbolz> :-(
[18:23] <hpk> cfbolz: bad luck
[18:23] <cfbolz> well yes. I fixed the date before I knew about the minisprint, otherwise I would have taken it at another point
[18:23] <cfbolz> but I don't want to reschedule it now
[18:24] <hpk> sure, it's not even 100% that we do the sprint
[18:24] <hpk> although i am currently thinking, we probably should
[18:24] <cfbolz> ok. Were would it take place?
[18:24] <hpk> either goetheborg or hildesheim as it seems
[18:25] <hpk> i guess that hildesheim would be better for you to reach on short notice (it has an ICE station)
[18:25] <hpk> we'll have a meeting on the 4th where we decide that
[18:25] <cfbolz> Yes, a lot better. But Goetheborg has more pypyers :-)
[18:26] <hpk> yes :-)
[18:26] <hpk> well, it's about equal, germany is gaining :-)
[18:26] <cfbolz> right.
----- silence for 18 minutes ----- [18:44] <hpk> arigo: you may be interested in the new py.test '-k' option
[18:44] <hpk> http://codespeak.net/py/current/doc/test.html#selecting-tests-by-keyword
[18:44] <arigo> :-)
[18:44] <hpk> it works with '-E' as well of course
[18:44] <arigo> at the moment I'm buried in strange C-ish Python code
[18:45] <arigo> because I'm writing execnet-over-hacked-rxvt-terminal
[18:45] <hpk> uh, ok, just wanted to let you know that you can now isolate regrtests pretty nicely
[18:45] <arigo> sure, -k looks very promizing
[18:53] dialtone (~dialtone@host118-0.pool21345.interbusiness.it) left irc: "Leaving"
[19:00] idnar (mithrandi@idnar.user) joined #pypy.
[19:02] dialtone (~dialtone@host118-0.pool21345.interbusiness.it) joined #pypy.
[19:12] dialtone (~dialtone@host118-0.pool21345.interbusiness.it) left irc: "Leaving"
[19:15] dialtone (~dialtone@host118-0.pool21345.interbusiness.it) joined #pypy.
[19:21] dialtone (~dialtone@host118-0.pool21345.interbusiness.it) left irc: "This computer has gone to sleep"
[19:24] ludal (~ludal@logilab.net2.nerim.net) left irc: "Download Gaim: http://gaim.sourceforge.net/";
----- silence for 37 minutes ----- [20:01] <cfbolz> I'm going home. See you all.
[20:01] <hpk> see you
[20:01] <arigo> see you
[20:01] <cfbolz> bye
[20:01] cfbolz (~cfbolz@edv01.physi.uni-heidelberg.de) left irc: "Leaving"
----- silence for 18 minutes ----- [20:19] <aleale> hi
[20:19] <arigo> hi Anders!
[20:19] <aleale> I want to ask about the dates for the Europython Sprint
[20:19] <aleale> Is it before or after the conference?
[20:20] <arigo> the main sprint is after the conference
[20:20] <arigo> there are 4 "core-people-hacking" days before
[20:20] <arigo> and some number of days of a more normal sprint after
[20:21] <aleale> As a sidenote I will be able to attend the mini-sprint if it flyes
[20:21] <arigo> nice
[20:21] <aleale> I have a problem with the after the confernce sprint
[20:22] <aleale> My daughter is going to play handball in Hungary during Europython
[20:23] <aleale> and we have planned to go there for our vacation
[20:23] <aleale> So how should "core-people hacking" be interpreted?
[20:23] <arigo> I see the problem
[20:24] <arigo> you're welcome to attend the shorter pre-conference sprint
[20:24] <arigo> the idea was to avoid having new people show up day after day and not knowing what to do
[20:25] <aleale> ok I see
[20:27] pedronis (pedronis@ratthing-b246.strakt.com) left irc: "Chatzilla 0.9.67 [Firefox 1.0.2/20050325]"
[20:35] <arigo> today is the day where strange stuff happens
[20:35] <arigo> some tests, randomly apparently, have an unknown rev number
[20:35] <arigo> test_builtin doesn't work on Samuele's machine, but works everywhere else we tried
[20:44] <aleale> Sounds a bit like my day ( if you put a laptop turning off several times without warning)
[20:45] <aleale> on top
[20:51] <hpk> arigo: can samuele on his machine do 'py.path.svnwc(.../pypy).info().rev' ?
[20:51] <arigo> ...test_builtin works on Samuele's machine now
[20:53] Action: hpk is rerunning all currently timeouted core tests
[20:53] <hpk> (except for test_mutants :-)
----- silence for 39 minutes ----- [21:32] <hpk> arigo: i am a bit unclear about the criteria you choose for selecting the core tests
[21:32] <hpk> for example why is test_cookie.py in there?
[21:34] <arigo> food time
[21:34] <arigo> basically, core tests are the ones that don't depend on external extension modules
[21:35] <arigo> or, in another point of view, tests that have a good chance to pass on Jython too
[21:35] <arigo> Cookie.py is a pure Python module, so it's a good stress-test of core things only
[21:35] <hpk> ok
[21:36] <hpk> test_cookie.py depends on insignificant pickling details ...
[21:36] <arigo> well most tests depend on insignificant <insert name here> details :-(
[21:36] <hpk> sure
[21:36] <hpk> that's a bit bad about doctests that they test string-representations
[21:36] <arigo> for <insert name here> in {pickling, doctest output, regexpr, Attribute vs TypeError, Memory vs OverflowError, ...}
[21:37] <arigo> yes.
[21:37] <arigo> for example test_extcall
[21:37] <hpk> yes, that is horrible
[21:37] <arigo> has probably no chance to pass
[21:37] <hpk> i've looked at it once but quickly gave up
[21:37] <arigo> we can probably remove it from the list at some point
[21:37] <arigo> the list is "optimistic"
[21:38] <arigo> it's a list of tests that we could pass if they were nicely written, basically
[21:38] <arigo> we could write nice replacements for some of these tests, actually, and propose them to CPython
[21:38] <hpk> with fine-grained we can list (at least the unit-) tests that we want skipped for this reason
[21:38] <hpk> yes
[21:39] Action: arigo -> dinner
[21:39] <arigo> see you
[21:39] arigo (~arigo@ratthing-b401.strakt.com) left irc: "[x]chat"
[21:39] <hpk> aleale: you are reading pypy-dev, right?
[21:41] <hpk> aleale: in particular i am asking because of the non-unicode-fake branch
[21:42] <aleale> Yes I've noticed
[21:43] <aleale> I looked at the branch and it seemed that it was about an interpreterlevel implementation of unicode objects
[21:44] <aleale> I have concentrated on the codecs part (and at applevel)
[21:44] <aleale> So I dont expect any dupplication of work
[21:45] <hpk> ok
[21:45] <hpk> aleale: welcome, btw!
[21:46] <aleale> When Anders from sweeden needs to grow .encode/decode/translate methods on the unicode objects I guess we can figure out how to reuse the stuff I am working on
[21:47] <aleale> And I am looking forward to have non faked unicode objects as I think they are breaking some of the test
[21:47] <aleale> I might be wrong though
[21:48] <aleale> And thank you Holger.
[21:48] <aleale> What are the rules for the three "lib" folders (lib,lib-python/2.3.4, lib-python/modified-2.3.4 )in the pypy tree
[21:48] <hpk> armin wrote a mail ... wait a moment, i am updating the documentation ...
[21:49] <aleale> I have discovered a need to modify 6 files in the encodings package
[21:50] <aleale> Should I check those files in in "lib" or "modified-2.3.4" ? (actually the whole package)
[21:50] <hpk> you probably need to "svn cp" things from lib-python/2.3.4 to lib-python/modified-2.3.4
[21:51] <hpk> i guess the whole 'encodings' package, unfortunately
[21:51] <hpk> as we can't partially modify the package in modified-2.3.4
[21:52] <hpk> am i making sense?
[21:57] <aleale> Yes - well I have tried copying, but I think that svn remembers where the files came from
[21:57] <hpk> yes, that's the intention
[21:57] <hpk> we want this trace
[21:58] <aleale> But when I check them in wont change the originals ?
[21:58] <aleale> But when I check them in wont I change the originals ?
[21:59] <hpk> if you do svn cp X Y
[21:59] <hpk> and modify Y, then you will not change X implicitely
[21:59] <hpk> the log message will show "copied modified from X"
[21:59] <aleale> Ok I wil try it (we can always roll back I guess)
[22:00] <hpk> yes, but try hard not to change anything in lib/2.3.4 by accident
[22:00] <aleale> Thats why I am asking
[22:00] <hpk> sure
[22:06] <aleale> ok here it goes
----- silence for 21 minutes ----- [22:27] dialtone (~dialtone@host118-0.pool21345.interbusiness.it) joined #pypy.
----- silence for 17 minutes ----- [22:44] <hpk> aleale: i wrote some documentation here: http://codespeak.net/pypy/index.cgi?doc/coding-style.html#id10
[00:00] --- Tue May 3 2005