MoinMoin: MoinMoinChat/Logs/moin-dev/2008-08-22

2008-08-22T00:09:09  <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4074:5df48eca0179 1.8-ldapgroups-mmihaljevic/MoinMoin/config/multiconfig.py: really fix group_manager config default
2008-08-22T00:59:03  <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4075:d97254aeb952 1.8-ldapgroups-mmihaljevic/MoinMoin/script/maint/mkpagepacks.py: mkpagepacks: use new group code
2008-08-22T00:59:04  <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4076:882126dbafe8 1.8-ldapgroups-mmihaljevic/MoinMoin/ (5 files in 4 dirs): make more use of new group code
2008-08-22T01:22:45  *** irc.freenode.net sets mode: +o ChanServ
2008-08-22T08:41:36  <dreimark> moin
2008-08-22T08:53:34  <ThomasWaldmann>  moin
2008-08-22T09:30:23  * ThomasWaldmann reads GSOC applications and deliverables
2008-08-22T10:42:52  <xorAxAx> TheSheep, ThomasWaldmann -> #moin-soc
2008-08-22T10:43:11  <xorAxAx> (even though there are still 15 min. left :))
2008-08-22T10:56:50  <ThomasWaldmann> PawelPacana: http://moinmo.in/ChristopherDenter/HelpOnStorageConfiguration can you fill in the hg bits?
2008-08-22T11:00:53  <ThomasWaldmann> TheSheep: please join us :)
2008-08-22T11:55:24  <xorAxAx> waldi: please see idea on http://moinmo.in/BastianBlank
2008-08-22T12:00:01  <waldi> xorAxAx: http://moinmo.in/BastianBlank/TreeOutputFormatter/Spec, not yet finished
2008-08-22T12:01:19  <xorAxAx> waldi: uh, is that define stuff some known scheme language?
2008-08-22T12:01:26  <xorAxAx> or your invention
2008-08-22T12:01:40  <xorAxAx> in any case, please explain it on a metalevel or refer to the name of it :)
2008-08-22T12:03:17  <waldi> yep
2008-08-22T12:03:25  <waldi> this is relax-ng
2008-08-22T12:06:52  <xorAxAx> then please add the name to the page :)
2008-08-22T12:07:25  <xorAxAx> also i am missing the define for page:page
2008-08-22T14:54:20  <zenhase> dreimark: i still haven't found a clue for why the tests are running slow
2008-08-22T14:55:16  <zenhase> actually in a case of heisenbug my test-runtimes jumped from about 120 seconds to 60 seconds
2008-08-22T14:55:30  <zenhase> i don't know of anything i have changed
2008-08-22T14:56:42  <dreimark> zenhase: I look later again in it, do you have some profiling code?
2008-08-22T15:03:35  * dreimark has go now 
2008-08-22T15:03:49  <dreimark> to
2008-08-22T15:03:53  <dreimark> bbl
2008-08-22T15:04:40  <dennda> google forgot an option in their survey :-)
2008-08-22T15:06:05  <dreimark> ?
2008-08-22T15:07:41  <xorAxAx> "do you want the money? [ ] no" :)
2008-08-22T15:12:56  <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4077:dd2570665f5b 1.8-ldapgroups-mmihaljevic/MoinMoin/ (5 files in 2 dirs): streamline group_manager initialisation (TODO: does not work with tests, but with standalone server)
2008-08-22T15:14:15  <dennda> No, wrt to finishing the goals you set
2008-08-22T15:14:21  <dennda> s/to//
2008-08-22T15:16:54  <xorAxAx> in which sense? you mean overfulfillment?
2008-08-22T15:18:00  <ThomasWaldmann> xorAxAx: please have a look at the changset above
2008-08-22T15:18:45  <ThomasWaldmann> i supect i am missing something rather easy there, maybe staticmethod() at the right place, but i did some experiments and it never worked for both cases
2008-08-22T15:19:41  <xorAxAx> who instantiates the lambda in the ordinary case?
2008-08-22T15:20:11  <ThomasWaldmann> you mean calling it?
2008-08-22T15:20:43  <zenhase> dreimark: http://paste.pocoo.org/show/83007/
2008-08-22T15:20:46  <xorAxAx> yes
2008-08-22T15:20:54  <ThomasWaldmann> request base init
2008-08-22T15:21:32  <ThomasWaldmann> (but I want to be flexible, so i didn't want to require that self MUST BE Request object)
2008-08-22T15:22:20  <xorAxAx> i suggest that you fetch another request object after generating the test config
2008-08-22T15:22:32  <ThomasWaldmann> and defining the lambda dummy, request: ... is ugly
2008-08-22T15:22:48  <xorAxAx> well
2008-08-22T15:23:01  <xorAxAx> if you getattr it from the class, it will wrap it in a bound method
2008-08-22T15:23:07  <xorAxAx> so it will get a self argument
2008-08-22T15:23:14  <xorAxAx> s/class/instance/
2008-08-22T15:23:49  <ThomasWaldmann> yes, that's the problem :)
2008-08-22T15:31:58  <dennda> No. If you didn't fulfill the original tasks, how much percent of the revised goals did you achieve? That misses the "all" option
2008-08-22T15:32:56  <xorAxAx> dennda: uh
2008-08-22T15:33:03  * xorAxAx checks
2008-08-22T15:33:21  <xorAxAx> " Student has completed all of the revised project goals."
2008-08-22T15:33:25  <xorAxAx> is on my list
2008-08-22T15:33:37  * dennda checks again
2008-08-22T15:33:59  <ThomasWaldmann> maybe wait a day until other people have found the bugs :)
2008-08-22T15:34:05  <dennda> not on mine
2008-08-22T15:34:19  <ThomasWaldmann> (or report them yourself :)
2008-08-22T15:34:29  <dennda> ThomasWaldmann: I already have. I spotted it yesterday and was told I was not the first who discovered that :-)
2008-08-22T15:34:42  <dennda> Because my feeling is, that that option would be the correct one for me
2008-08-22T15:35:06  <xorAxAx> well
2008-08-22T15:35:09  <xorAxAx> it wont be fixed
2008-08-22T15:35:18  <dennda> of course not
2008-08-22T15:35:20  <xorAxAx> and next year there will be the new FOSS version
2008-08-22T15:35:28  <dennda> foss version? :-)
2008-08-22T15:35:36  <ThomasWaldmann> with new bugs :))
2008-08-22T15:35:57  <dennda> for other organizations like microsoft to have a summer of code themselves without reinventing the wheel? :-)
2008-08-22T15:36:45  <ThomasWaldmann> that's NFOSS rather
2008-08-22T15:37:12  <xorAxAx> dennda: no, to let students fix the bugs
2008-08-22T15:39:38  <dennda> ThomasWaldmann: Oh, they do have a FOSS Research Center
2008-08-22T15:39:52  <dennda> This is it: http://www.meine-urlaubsbilder.de/media/520/huette070404110427d4.jpg
2008-08-22T15:45:51  <ThomasWaldmann> dennda: yeah, likely to spy on what FOSS is doing, find out weaknesses, etc...
2008-08-22T15:46:57  <ThomasWaldmann> i guess they need at least power and internet connectivity :)
2008-08-22T15:47:37  <ThomasWaldmann> hmm, i guess what we really need is a improved conftest
2008-08-22T15:48:34  <ThomasWaldmann> esp. TestConfig
2008-08-22T15:52:48  <xorAxAx> well, testconfig is kind of legacy
2008-08-22T15:52:54  <xorAxAx> and was broken before
2008-08-22T17:28:25  <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4155:477528bcbdc0 1.8-ldapgroups-mmihaljevic/ (1013 files in 185 dirs): merged moin/1.8 repo
2008-08-22T17:55:05  <ThomasWaldmann> zenhase: what did you try yet to find why the tests are slow?
2008-08-22T18:40:17  <zenhase> ThomasWaldmann: the problem is searching for something, that doesn't seem to occur here :/
2008-08-22T18:40:32  <zenhase> i try it out on other boxes right now
2008-08-22T18:43:01  <ThomasWaldmann> what machine does it not occur on/
2008-08-22T18:43:13  <ThomasWaldmann> hm, shift out of order
2008-08-22T18:44:23  <zenhase> hmm?
2008-08-22T18:44:36  <zenhase> you mean architecture?
2008-08-22T18:47:44  <ThomasWaldmann> cpu, ram, disk, os, python version
2008-08-22T18:48:50  <zenhase> my laptop = core 2 duo 2ghz, 1gb, 80gb laptopdisk, osx, 2.5.1
2008-08-22T18:49:21  <zenhase> my desktop = athlon64 3000+, 1gb, 500gb (crypted), freebsd 7, 2.5.2
2008-08-22T18:50:12  <ThomasWaldmann> pentium-m 1GHz 1GB flash disk ubuntu 8.04 -> very noticeable
2008-08-22T18:50:18  <zenhase> both run in about 60 seconds (120 in the first run because of bytecompiling i suppose)
2008-08-22T18:50:29  <zenhase> and strangest is dreimarks system
2008-08-22T18:50:40  <zenhase> with 900 seconds or so
2008-08-22T18:50:50  <zenhase> on a fast dualcore athlon64 in a ramdisk
2008-08-22T18:51:18  <zenhase> and 1.8-current runs about the same on my machines
2008-08-22T18:52:33  <ThomasWaldmann> i just retry with current code
2008-08-22T18:53:03  <ThomasWaldmann> how are we proceeding with the wsgi docs?
2008-08-22T18:55:30  <zenhase> i caught up on some non-gsoc issues i have been pushing after the deadline and also my wrist made itself noticable again, so i did not make a lot of progress there, but i planned on doing them this weekend
2008-08-22T18:55:56  <ThomasWaldmann> test_Page 6.2s 1.8, 11.5s 1.8-wsgi
2008-08-22T18:56:24  <zenhase> yeah, that is quite noticable :o
2008-08-22T18:56:51  <zenhase> with dreimarks result i am really worried what could have gone wrong there, but i have no explanation
2008-08-22T18:56:52  <ThomasWaldmann> please post output of time py.test test_page.py of both repos > pastebin
2008-08-22T18:56:58  <zenhase> ok
2008-08-22T19:01:29  <zenhase> http://paste.pocoo.org/show/83030/
2008-08-22T19:01:54  <zenhase> i pulled my tree fresh again
2008-08-22T19:02:13  <zenhase> (and skipped the first run on both, so the unpacking and bytecompiling is missing)
2008-08-22T19:02:25  <zenhase> this is on my desktop now, the athlon
2008-08-22T19:03:01  * zenhase is running both full tests again
2008-08-22T19:03:31  <ThomasWaldmann> test_caching is 3.8 : 24 !
2008-08-22T19:04:06  <zenhase> can have modern cpus that of an effect?
2008-08-22T19:04:22  <zenhase> i mean, my laptop has 3mb of l2 cache per core
2008-08-22T19:04:59  <zenhase> i am not that informed about the effect of cpus on languages like python
2008-08-22T19:05:42  <ThomasWaldmann> maybe you are doing cpu intensive stuff. you machine has faster cpu, so you see a smaller difference
2008-08-22T19:06:11  <zenhase> hmm, may be a possibility
2008-08-22T19:06:19  <zenhase> but then, what about dreimark?
2008-08-22T19:06:22  <ThomasWaldmann> dreimark: what's your cpu?
2008-08-22T19:06:25  <zenhase> doesn't really make sense
2008-08-22T19:06:39  <zenhase> [13:13] <dreimark> eigentlich ziemlich flott 64bit amd doppelkern 2.2GHz
2008-08-22T19:06:52  <zenhase> he told me during the meeting on wednesday
2008-08-22T19:06:55  <ThomasWaldmann> strange
2008-08-22T19:06:59  <zenhase> sure is
2008-08-22T19:07:11  <zenhase> and i guess he will have a suitable amount of ram too
2008-08-22T19:11:35  <zenhase> i will try to compare profiling of both current vs. wsgi on my desktop
2008-08-22T19:12:20  <zenhase> but the numbers in the profile where so low, that i can't really read anything particular strange from them
2008-08-22T19:16:05  <ThomasWaldmann> i even have same py.test version
2008-08-22T19:16:48  <xorAxAx> zenhase: i would call the profiler in py.test
2008-08-22T19:16:57  <xorAxAx> i.e. cp -a `which py.test` mypytest
2008-08-22T19:17:01  <xorAxAx> vim mypytest
2008-08-22T19:17:37  <zenhase> xorAxAx: hmm good idea
2008-08-22T19:17:54  <zenhase> but i will have to try that later
2008-08-22T19:18:06  <zenhase> i have to meet someone in 45min and still need to shower
2008-08-22T19:18:27  <xorAxAx> yeah
2008-08-22T19:18:28  <zenhase> i will be back later this evening
2008-08-22T19:18:42  <zenhase> btw, which profiler is preferable?
2008-08-22T19:18:46  <zenhase> cProfile or hotshot?
2008-08-22T19:19:05  <zenhase> i did not quite knew which one to pick and started out with cProfile
2008-08-22T19:24:45  <xorAxAx> cprofile
2008-08-22T19:39:22  <ThomasWaldmann> zenhase: it doesn't look cpu bound. i ran top in parallel and it mostly shows 90% idle.
2008-08-22T19:39:51  <xorAxAx> with a hot cache?
2008-08-22T19:40:25  <ThomasWaldmann> if it ever finishes, I'll rerun it :)
2008-08-22T19:44:09  <ThomasWaldmann> test_wikidicts tages ages
2008-08-22T19:48:08  <ThomasWaldmann> 290s for test_wikidicts
2008-08-22T20:15:41  <ThomasWaldmann> zenhase: do you have a complete underlay.tar extracted when running tests?
2008-08-22T20:21:01  <ThomasWaldmann> test_wikidicts.py:121 TestGroupDicts().testRenameGroupPage ok (42.44)
2008-08-22T20:21:09  <ThomasWaldmann> 42.44secs
2008-08-22T20:33:43  <ThomasWaldmann> zenhase: a create_page() call is taking 20s
2008-08-22T20:34:24  <ThomasWaldmann> renamePage 10s
2008-08-22T20:35:06  <ThomasWaldmann> nuke_page 10s
2008-08-22T20:35:27  <ThomasWaldmann> wtf!?
2008-08-22T20:44:14  <grzywacz> :D
2008-08-22T21:23:31  <xorAxAx> students, dont forget http://groups.google.com/group/google-summer-of-code-announce/web/how-to-provide-google-with-sample-code
2008-08-22T21:23:35  <xorAxAx> "Any student who does not provide a code sample will be expected to return the final program payment to us."
2008-08-22T21:31:09  <ThomasWaldmann> zenhase: init_test_request takes 70ms
2008-08-22T21:31:42  <ThomasWaldmann> (so while not super fast, it is not the main culprit)
2008-08-22T21:35:05  <dennda> As a requirement for the Google Summer of CodeTM, all students who receive passing final evaluations will need to provide a copy of the code they produced as part of participating in the program.
2008-08-22T21:35:19  <dennda> When will the outcome of the evaluations be announced?
2008-08-22T21:40:38  <ThomasWaldmann> after docs are finished :D
2008-08-22T21:41:39  <dennda> Clever response :-)
2008-08-22T22:06:36  <ThomasWaldmann> zenhase: be careful, I know your IP! :P
2008-08-22T22:10:13  <ThomasWaldmann>  tests finished: 618 passed, 6 failed, 74 skipped in 95.40 seconds (from wsgi repo code)
2008-08-22T22:16:43  <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4097:b6b6979b4813 1.8-wsgi-fkrupicka/MoinMoin/web/request.py: fix test slowness due to failing reverse DNS lookups
2008-08-22T22:18:30  <dennda> ThomasWaldmann: 127.0.0.1? Hack him! :-)
2008-08-22T22:19:04  <ThomasWaldmann> no, 10.10.10.10 :)
2008-08-22T23:01:12  <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4219:07862b0663fd 1.8-wsgi-fkrupicka/ (1063 files in 200 dirs): merge moin/1.8 repo
2008-08-22T23:30:42  <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4026:d3c2d8d89959 1.8/MoinMoin/action/fckdialog.py: fix failing sourcecode tests, including removing CRLF lineends
2008-08-22T23:30:48  <ThomasWaldmann> xorAxAx: ^^