--- Log opened Thu Sep 12 00:00:55 2013 | ||
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 00:16 | |
thoralf_ | hey | 00:16 |
---|---|---|
-!- hushell [~hushell@8-92.ptpg.oregonstate.edu] has joined #shogun | 00:26 | |
@iglesiasg | hello hello | 00:26 |
-!- HeikoS [~heiko@213.190.120.70] has joined #shogun | 00:27 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 00:27 | |
@HeikoS | . | 00:28 |
@iglesiasg | HeikoS, so how is the city? | 00:31 |
@HeikoS | iglesiasg: man, its raining | 00:31 |
@HeikoS | just got completely soaked | 00:31 |
@iglesiasg | haha | 00:31 |
@HeikoS | to the underpants | 00:31 |
@iglesiasg | welcome to Scandinavia :D | 00:32 |
@HeikoS | haha | 00:32 |
@HeikoS | apart from that, its nice | 00:32 |
@HeikoS | just had lobster soup and shrimpks | 00:32 |
@HeikoS | shrimps | 00:32 |
* iglesiasg does not eat bugs puagh! | 00:32 | |
@iglesiasg | hehe | 00:33 |
@HeikoS | was very tasty | 00:33 |
@HeikoS | They also had whale, but was quite expensive | 00:34 |
@iglesiasg | aham I see | 00:34 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Read error: Connection reset by peer] | 01:02 | |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun | 01:03 | |
-!- hushell [~hushell@8-92.ptpg.oregonstate.edu] has quit [Ping timeout: 256 seconds] | 01:13 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 01:14 | |
-!- iglesiasg_ [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 01:14 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 01:43 | |
shogun-notifier- | shogun: Viktor Gal :develop * f98a814 / examples/undocumented/python_modular/distance_mahalanobis_modular.py: https://github.com/shogun-toolbox/shogun/commit/f98a814bfe135a87c6b6fcd365089a6e64256814 | 01:43 |
shogun-notifier- | shogun: fix distance_mahalanobis_modular.py example | 01:43 |
-!- HeikoS [~heiko@213.190.120.70] has quit [Quit: Leaving.] | 01:52 | |
-!- Tuxa [3e2ff647@gateway/web/freenode/ip.62.47.246.71] has joined #shogun | 02:08 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 02:09 | |
iglesiasg_ | good night to everybody! | 02:09 |
-!- Tuxa [3e2ff647@gateway/web/freenode/ip.62.47.246.71] has quit [Client Quit] | 02:09 | |
-!- iglesiasg_ [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 02:09 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 02:10 | |
@wiking | thoralf_: here or just a ghost of u? | 02:12 |
@wiking | :) | 02:12 |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 02:19 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 02:20 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 02:29 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 02:30 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 02:39 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 02:40 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 02:49 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 02:50 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 02:59 | |
shogun-buildbot | build #64 of FC19 - modular_interfaces is complete: Failure [failed git] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20modular_interfaces/builds/64 blamelist: Viktor Gal <viktor.gal@maeth.com> | 03:00 |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 03:00 | |
-!- travis-ci [~travis-ci@ec2-50-17-6-85.compute-1.amazonaws.com] has joined #shogun | 03:06 | |
travis-ci | [travis-ci] it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11259762 | 03:06 |
-!- travis-ci [~travis-ci@ec2-50-17-6-85.compute-1.amazonaws.com] has left #shogun [] | 03:06 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 03:09 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 03:10 | |
shogun-buildbot | build #1782 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1782 | 03:18 |
-!- sonne|osx_ [~sonne@f053046093.adsl.alicedsl.de] has joined #shogun | 03:20 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 03:20 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 03:20 | |
-!- sonne|osx [~sonne@f053034179.adsl.alicedsl.de] has quit [Ping timeout: 268 seconds] | 03:22 | |
-!- sonne|osx_ is now known as sonne|osx | 03:22 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 03:30 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 03:30 | |
shogun-buildbot | build #1438 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1438 blamelist: lambday <heavensdevil6909@gmail.com>, Soeren Sonnenburg <sonne@debian.org>, Thoralf Klein <thoralf.klein@zib.de>, hushell <hushell@hushell-U510.(none)>, Kevin <kevinhughes27@gmail.com>, | 03:37 |
shogun-buildbot | Viktor Gal <viktor.gal@maeth.com>, Evgeniy Andreev <gsomix@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, van51 <vangelis_51@hotmail.com> | 03:37 |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 03:40 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 03:40 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 03:50 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 03:50 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 04:00 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 04:00 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 04:10 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 04:10 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 04:20 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 04:20 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 04:30 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 04:30 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 04:40 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 04:41 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 04:43 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 04:50 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 04:51 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 05:00 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:01 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 05:10 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:11 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:12 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:12 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:16 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:17 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:21 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:22 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:27 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:27 | |
-!- hushell [~hushell@c-98-232-178-161.hsd1.or.comcast.net] has joined #shogun | 05:31 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:32 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:32 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:37 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:37 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:42 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:42 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:47 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:47 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:52 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:52 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:54 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:54 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Client Quit] | 05:59 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has joined #shogun | 05:59 | |
-!- thoralf_ [~thoralf@24-134-83-14-dynip.superkabel.de] has quit [Ping timeout: 240 seconds] | 06:06 | |
-!- sonne|osx [~sonne@f053046093.adsl.alicedsl.de] has quit [Quit: sonne|osx] | 07:13 | |
-!- lisitsyn1 [~lisitsyn@fb2-lo1.global63.net] has joined #shogun | 07:28 | |
-!- besser82 [~besser82@77-22-24-208-dynip.superkabel.de] has joined #shogun | 07:37 | |
-!- besser82 [~besser82@77-22-24-208-dynip.superkabel.de] has quit [Changing host] | 07:37 | |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 07:37 | |
-!- lisitsyn1 [~lisitsyn@fb2-lo1.global63.net] has quit [Quit: Leaving.] | 07:50 | |
-!- sonne|osx [~sonne@89.204.137.111] has joined #shogun | 08:00 | |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Read error: Connection reset by peer] | 08:21 | |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun | 08:23 | |
-!- sonne|osx [~sonne@89.204.137.111] has quit [Quit: sonne|osx] | 08:25 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 08:25 | |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 9ab664e / examples/undocumented/python_modular/ (9 files): https://github.com/shogun-toolbox/shogun/commit/9ab664eee666db3c7e1984d13cd296314326960a | 08:25 |
shogun-notifier- | shogun: convert further examples to use CSVFile | 08:25 |
shogun-buildbot | build #1439 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1439 blamelist: Soeren Sonnenburg <sonne@debian.org> | 08:46 |
shogun-buildbot | build #65 of FC19 - modular_interfaces is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20modular_interfaces/builds/65 blamelist: Soeren Sonnenburg <sonne@debian.org> | 08:48 |
-!- lisitsyn1 [~lisitsyn@fb2-lo1.global63.net] has joined #shogun | 09:15 | |
shogun-buildbot | build #1783 of deb3 - modular_interfaces is complete: Failure [failed test python_modular test ruby_modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1783 blamelist: Soeren Sonnenburg <sonne@debian.org> | 09:18 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 618878d / examples/undocumented/python_modular/ (5 files): https://github.com/shogun-toolbox/shogun/commit/618878d37385961fc3fdf3fc2262faf43fb21242 | 09:32 |
shogun-notifier- | shogun: fix a number of drunken horse syndromes | 09:32 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 09:38 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 09:38 | |
shogun-buildbot | build #66 of FC19 - modular_interfaces is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20modular_interfaces/builds/66 blamelist: Soeren Sonnenburg <sonne@debian.org> | 09:46 |
shogun-buildbot | build #1440 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1440 blamelist: Soeren Sonnenburg <sonne@debian.org> | 09:56 |
@wiking | i've got the bug | 10:00 |
lisitsyn1 | haha drunken horse | 10:02 |
lisitsyn1 | good catch | 10:02 |
@wiking | tricky bug: | 10:05 |
@wiking | https://github.com/shogun-toolbox/shogun/issues/1576 | 10:06 |
-!- lisitsyn1 [~lisitsyn@fb2-lo1.global63.net] has left #shogun [] | 10:06 | |
-!- hoijui [~hoijui@dslb-088-075-033-009.pools.arcor-ip.net] has joined #shogun | 10:07 | |
-!- HeikoS [~heiko@213.190.120.70] has joined #shogun | 10:11 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 10:11 | |
@wiking | HeikoS: sup man? lot of bugs awaiting :D | 10:12 |
@HeikoS | wiking: hi! yes I know, but travelling currently | 10:12 |
@HeikoS | how are things going? | 10:13 |
@wiking | slowly | 10:13 |
@HeikoS | wiking: I still need to ask you stupid questions on how to link shogun against .a files that I got on this cluster machine :D | 10:13 |
@wiking | finding more and more bugs :) | 10:13 |
@HeikoS | ah | 10:13 |
@HeikoS | yeah I have been there ;) | 10:13 |
@HeikoS | its just full of those | 10:13 |
@wiking | HeikoS: btw https://github.com/shogun-toolbox/shogun/issues/1574 | 10:14 |
shogun-buildbot | build #1784 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1784 blamelist: Soeren Sonnenburg <sonne@debian.org> | 10:14 |
@HeikoS | why does that cause problems? | 10:14 |
@wiking | anything againsg having like CT_INIT, -1 | 10:14 |
@wiking | ? | 10:14 |
@wiking | HeikoS: well because as i've written there | 10:14 |
@wiking | we'll have no idea | 10:14 |
@wiking | what that equation will lead to | 10:15 |
@HeikoS | wiking: yep thats fine, but I made sure not to use the std ctor and its not SGO* | 10:15 |
@wiking | after that you cannot test that value for anythig | 10:15 |
@HeikoS | ok but you are right | 10:15 |
@wiking | HeikoS: well when u create LibLinear | 10:15 |
@HeikoS | will change | 10:15 |
@wiking | you'll have like 2 instances | 10:15 |
@wiking | that'll hold this value | 10:15 |
@wiking | :P | 10:16 |
@HeikoS | ah really? | 10:16 |
@wiking | yep | 10:16 |
@HeikoS | ok, I will try to change soon | 10:16 |
@HeikoS | maybe some time tonight or so | 10:16 |
@wiking | so change can go? | 10:16 |
@HeikoS | wiking: there might be some problems | 10:16 |
@HeikoS | with me checking for the weird values | 10:16 |
@HeikoS | but I will search for that | 10:16 |
@HeikoS | ok man I have to go now, workshop starts | 10:16 |
@wiking | okok | 10:16 |
@wiking | cyaz | 10:16 |
@HeikoS | byez | 10:17 |
-!- HeikoS [~heiko@213.190.120.70] has quit [Quit: Leaving.] | 10:20 | |
-!- hoijui [~hoijui@dslb-088-075-033-009.pools.arcor-ip.net] has quit [Quit: Leaving] | 10:23 | |
hushell | wiking: hi | 10:38 |
@wiking | hushell: yo | 10:38 |
hushell | wiking: how do I deal with ERROR ('module' object has no attribute 'parameter_list')? | 10:38 |
@wiking | where? | 10:38 |
hushell | I added python interfaces for my code | 10:39 |
hushell | wiking: Test #157: integration-python_modular-tester-structure_factor_graph_model | 10:39 |
shogun-notifier- | shogun: hushell :develop * 05996bf / CMakeLists.txt,cmake/FindMosek.cmake: https://github.com/shogun-toolbox/shogun/commit/05996bf2caf8eac58cc0107abebf9205a2676fd6 | 10:39 |
shogun-notifier- | shogun: FindMosek.cmake | 10:39 |
shogun-notifier- | shogun: Viktor Gal :develop * 847bb71 / CMakeLists.txt,cmake/FindMosek.cmake: https://github.com/shogun-toolbox/shogun/commit/847bb7109061cfb49e648c6e8e3c8be24ea04fcc | 10:39 |
shogun-notifier- | shogun: Merge pull request #1539 from hushell/mosek | 10:39 |
shogun-notifier- | shogun: | 10:39 |
shogun-notifier- | shogun: FindMosek.cmake | 10:39 |
hushell | wiking: I don't know if I followed the rule to add modular interfaces | 10:40 |
hushell | wiking: so basically I only touched Structure.i and Structure_include.i in modular directory | 10:41 |
@wiking | hushell: can u show me a diff? | 10:41 |
hushell | wiking: okay, a second | 10:43 |
hushell | wiking: here it is: https://github.com/hushell/shogun/compare/develop | 10:44 |
hushell | wiking: BTW, thanks for merging the mosek PR! | 10:45 |
@wiking | hushell: yeah just testing | 10:45 |
@wiking | thnx for the patch | 10:45 |
@wiking | hushell: why did u put stuff in modshogun_ignores.i ? | 10:47 |
@wiking | i mean i'm just wondering | 10:47 |
hushell | wiking: where can I check the code for integration_test? I am a bit lost in tests/integration/python_modular | 10:48 |
@wiking | ah i'm lost with that as well | 10:48 |
@wiking | :P | 10:48 |
hushell | wiking: haha, hard to deal with generated code | 10:50 |
sonne|work | wiking, hushell: ctest ! | 10:53 |
hushell | sonne|work: any idea where I was wrong? | 10:56 |
sonne|work | hushell: ?? | 10:57 |
sonne|work | hushell: what do you want to do? | 10:57 |
hushell | sonne|work: I just failed an integration test after I added a python example of my factor graph model | 10:58 |
sonne|work | hushell: which? | 10:58 |
hushell | https://github.com/hushell/shogun/compare/develop this is the diff | 10:59 |
hushell | the example passed but the integration no | 10:59 |
sonne|work | hushell: ??? | 10:59 |
sonne|work | hushell: you only showed some python script | 11:00 |
hushell | okay let me show the error message | 11:00 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 3075452 / examples/undocumented/python_modular/ (2 files): https://github.com/shogun-toolbox/shogun/commit/3075452625b734fa7a18d96cacf3c4fece140f47 | 11:00 |
shogun-notifier- | shogun: further fixes | 11:00 |
hushell | sonne|work: Test #157: integration-python_modular-tester-structure_factor_graph_model ................................***Failed 0.17 sec | 11:01 |
hushell | structure_factor_graph_model.py ERROR ('module' object has no attribute 'parameter_list') | 11:01 |
sonne|work | hushell: so now you need to make it testable - for that you have to create the parameter_list | 11:01 |
sonne|work | hushell: look at any other .py example | 11:01 |
sonne|work | hushell: the parameter list contains settings that your function is called with | 11:01 |
sonne|work | hushell: the output of your function when run with certain settings can then be stored when you run generator.py | 11:02 |
hushell | sonne|work: I see. I was thinking that's not useful | 11:02 |
sonne|work | hushell: then tester.py YOUREXAMPLE.py will check if the output is still the same | 11:02 |
sonne|work | it is basically a test for free | 11:03 |
hushell | sonne|work: thanks for the detailed explanation! | 11:03 |
sonne|work | hushell: but you should return sth in you function that can be checked against | 11:03 |
sonne|work | e.g. the trained model or so | 11:03 |
hushell | sonne|work: I am reading iglesiasg 's example ;) | 11:04 |
sonne|work | hushell: and please in the final example don't print anything (just comment the lines) | 11:04 |
lisitsyn | wiking: what would you say about latent svm in opencv? | 11:05 |
@wiking | lisitsyn: it's just a classifier | 11:05 |
@wiking | lisitsyn: it's not being able to train | 11:05 |
@wiking | or at least the last time i've checked | 11:05 |
shogun-notifier- | shogun: Viktor Gal :develop * bc579d6 / src/shogun/mathematics/Mosek.cpp: https://github.com/shogun-toolbox/shogun/commit/bc579d6ad24551c890d68c4769fcbc89772b6913 | 11:06 |
shogun-notifier- | shogun: Check for Mosek major version | 11:06 |
shogun-notifier- | shogun: towards fixing issue #1287 | 11:06 |
@wiking | sonne|work: https://github.com/shogun-toolbox/shogun/issues/1576 | 11:06 |
@wiking | ideas about this one? | 11:06 |
lisitsyn | wiking: ohh I see | 11:06 |
hushell | wiking, lisitsyn: these code could be candidates to be included: http://www.centrale-ponts.fr/tutorials/cvpr2013/ | 11:06 |
shogun-buildbot | build #1441 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1441 blamelist: hushell <hushell@hushell-U510.(none)> | 11:07 |
lisitsyn | hushell: what's license? | 11:07 |
@wiking | hehehe i'll run that data on our train :))0 | 11:07 |
@wiking | coool hushell thnx for the link | 11:07 |
hushell | I asked Pawan, he said he would be happy to see his code in Shogun | 11:08 |
@wiking | hushell: partly it's in | 11:09 |
@wiking | hushell: which code of his exactly? | 11:09 |
sonne|work | wiking: no idea | 11:09 |
lisitsyn | hushell: ohh you should git config your git | 11:09 |
hushell | wiking: I'd like to help if you need after gsoc | 11:09 |
@wiking | sonne|work: i mean clearly now the inputparser is broken | 11:09 |
lisitsyn | that's not really cool "<hushell@hushell-U510.(none)>" | 11:09 |
sonne|work | lisitsyn: do you still have some contact to gsomix? | 11:09 |
lisitsyn | sonne|work: phone ;) | 11:10 |
@wiking | hushell: self-paced is not in... | 11:10 |
lisitsyn | sonne|work: ah well I can break into his room during the night too | 11:10 |
hushell | lisitsyn: I left it default :( I'll do it | 11:10 |
@wiking | hushell: but i'll benchmark now his stuff with shogun's | 11:10 |
@wiking | hushell: do u have contact with pawan? | 11:10 |
hushell | wiking: I met him in cvpr | 11:10 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 11:11 | |
hushell | wiking: in that tutorial | 11:11 |
@wiking | hushell i see ... i'm just wondering if he'd be interested in adding shogun reference there | 11:11 |
hushell | wiking: I knew you were interested in his paper, so I mentioned maybe we will port his code | 11:11 |
hushell | wiking: we can send him an email to get the permit | 11:12 |
@wiking | hushell: well we have a bit faster/better code in shogun than actually self-paced one | 11:12 |
@wiking | hushell: but at least now i have this dataset to prove it ;P | 11:13 |
hushell | wiking: he has many variants of latent svm, max margin min entropy for example | 11:13 |
hushell | and his latest loss based lsvm | 11:13 |
hushell | wiking: what dataset? | 11:14 |
sonne|work | wiking: we have to convert this whole system to SGVector and friends. only then the errors will go away but it is a lot of work | 11:14 |
shogun-buildbot | build #1706 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1706 blamelist: hushell <hushell@hushell-U510.(none)> | 11:15 |
shogun-buildbot | build #67 of FC19 - modular_interfaces is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20modular_interfaces/builds/67 blamelist: Soeren Sonnenburg <sonne@debian.org>, hushell <hushell@hushell-U510.(none)> | 11:20 |
shogun-buildbot | build #1442 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1442 blamelist: Soeren Sonnenburg <sonne@debian.org> | 11:22 |
@wiking | hushell: on that website | 11:23 |
@wiking | sonne|work: the problem is that now we have like null pointer dereferences... | 11:23 |
@wiking | sonne|work: we cannot do a release like this | 11:23 |
sonne|work | I don't have enough spare time to fix it | 11:25 |
sonne|work | so no idea what to do | 11:25 |
@wiking | sonne|work: ok the question is now | 11:25 |
@wiking | should that be actually a referenced SGVector? | 11:25 |
@wiking | ex_ring[i].fv | 11:26 |
@wiking | ? | 11:26 |
sonne|work | wiking: yes but it will create gazillions of issues | 11:26 |
sonne|work | wiking: thoralf is heavily using this ... | 11:26 |
@wiking | sonne|work: yeah i mean i've tried adding theere | 11:26 |
@wiking | ex_ring[i].fv = SGVector<T>(); | 11:26 |
@wiking | that was fucking nightmare | 11:26 |
@wiking | :) | 11:26 |
@wiking | i thought T is a typename | 11:26 |
@wiking | but apparently it's not only that... | 11:27 |
sonne|work | wiking: yeah I experienced the same in my last attempt | 11:31 |
sonne|work | wiking: I think to get it to work we would need many tests for VW and the streaming features | 11:32 |
sonne|work | wiking: even before refactoring I mean | 11:32 |
-!- van51 [~van51@athedsl-410351.home.otenet.gr] has joined #shogun | 11:32 | |
@wiking | sonne|work: oh waiiiit fuck but now i seee :) | 11:33 |
@wiking | class Example | 11:33 |
@wiking | { | 11:33 |
@wiking | public: /// Label float64_t label; /// Feature vector of type T SGVector<T> fv; | 11:33 |
@wiking | }; | 11:33 |
sonne|work | van51: hey there! | 11:33 |
sonne|work | van51: IMHO you did it all totally OK. | 11:33 |
@wiking | should that be T fv? | 11:33 |
van51 | sonne|work: hello! | 11:33 |
van51 | sonne|work: are you referring to the email? | 11:34 |
sonne|work | van51: yes | 11:34 |
van51 | sonne|work: ok, good to know :D | 11:34 |
sonne|work | van51: sure one gets collisions but on the other hand we don't want even higher dim vectors! | 11:34 |
sonne|work | van51: and the way you did it (seeding different dims / quadratic differently) should minimize collisions | 11:35 |
van51 | sonne|work: ok sounds good to me :) | 11:37 |
van51 | sonne|work: thanks for the clarification | 11:38 |
shogun-buildbot | build #1707 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1707 blamelist: Soeren Sonnenburg <sonne@debian.org> | 11:38 |
@wiking | sonne|work: ok actually the only problem would be StreamingDenseFeatures.h as there the class parameters are actually typenames | 11:38 |
@wiking | or the other way round | 11:39 |
sonne|work | wiking: if you can fix it do it... it would be great to have this proper | 11:39 |
@wiking | ppl should have never used InputParser<class> | 11:40 |
thoralf | sonne|work: I heard my name? ;) | 11:40 |
@wiking | sonne|work: i'd like to i'm just wondering wtf is happening here | 11:40 |
@wiking | i mean sometimes they use typename sometimes class as template parameter for InputParser | 11:40 |
sonne|work | wiking: well all needs to be converted to use gsomix' line reader | 11:40 |
@wiking | which is shiit | 11:40 |
@wiking | sonne|work: ? | 11:40 |
thoralf | Hey wiking - still on your pointer problem? ;) | 11:40 |
sonne|work | wiking: well we have like 10 line readers in shogun | 11:41 |
@wiking | thoralf: yeah now at least i know what's the problem | 11:41 |
thoralf | wiking: Oh, tell me. | 11:41 |
@wiking | thoralf: https://github.com/shogun-toolbox/shogun/issues/1576 | 11:41 |
sonne|work | thoralf: well I guess you know that https://github.com/shogun-toolbox/shogun/issues/1576 ? | 11:41 |
hushell | wiking: oh, yeah that data is definitely a good one for testing ideas | 11:41 |
@wiking | thoralf: the good thing is that ex_ring is calloced... that's why we dont get a huge crash | 11:41 |
@wiking | sonne|work: ok so just fill me in | 11:42 |
@wiking | sonne|work: what needs to converted to what | 11:42 |
@wiking | sonne|work: maybe i can fix it | 11:42 |
thoralf | wiking: Did you try what I suggested yesterday? Just cloning the tmp vector? | 11:42 |
@wiking | sonne|work: but this part of shogun i barely have ideas what's happening | 11:42 |
sonne|work | wiking: well the big picture is: | 11:42 |
@wiking | thoralf: which vector... i mean there's nothing to clone anywhere... as you see there's nothing at all allocated there | 11:42 |
sonne|work | you have streaming features | 11:42 |
sonne|work | wiking: that in contrast to our dense etc features are streamed from whereever | 11:43 |
@wiking | yep | 11:43 |
@wiking | and they are buffered somehow now | 11:43 |
sonne|work | wiking: so you only have a function get_next_vector etc | 11:43 |
@wiking | yep yep | 11:43 |
@wiking | that i know | 11:43 |
thoralf | wiking: (1) SGVector<T> SGVector<T>::clone() const | 11:43 |
sonne|work | wiking: since we want to do this efficiently | 11:43 |
@wiking | thoralf: null::clone() | 11:43 |
@wiking | ? | 11:43 |
sonne|work | wiking: the reading is done in an extra thread writing to a buffer (trying to keep it full) | 11:43 |
@wiking | thoralf: i cannot clone a null obj | 11:43 |
thoralf | wiking: Clone tmp vector in get_next_example() and pass the clone | 11:43 |
lisitsyn | wiking: I tried to convert it to sgvectors once | 11:44 |
lisitsyn | :D | 11:44 |
lisitsyn | good luck! | 11:44 |
@wiking | sonne|work: yeah i see that much as well | 11:44 |
sonne|work | wiking: now that buffer needs conversion to store SGVectors | 11:44 |
thoralf | SGVector<char> tmp; <-- This is null? | 11:44 |
sonne|work | instead of raw pointers | 11:44 |
thoralf | features/streaming/StreamingHashedDocDotFeatures.cp, line 153? | 11:44 |
@wiking | thoralf: no man everything in ParseBuffer is null | 11:44 |
lisitsyn | thoralf: no that's not null | 11:44 |
sonne|work | lisitsyn: 3 people and 2 gave up already *yay*! | 11:44 |
@wiking | sonne|work: but why do we want it to have SGVectors | 11:45 |
@wiking | this i dont get | 11:45 |
@wiking | i mean there's this Example<> | 11:45 |
sonne|work | wiking: in addition you would need to convert all the VW stuff underneath to work with that | 11:45 |
thoralf | wiking: Okay, tell me how I can reproduce it and I'll try a quick fix. | 11:45 |
@wiking | thoralf: run that example | 11:45 |
sonne|work | wiking: yes Example comes from VW IIRC | 11:45 |
lisitsyn | sonne|work: I died there so only 8 lives left | 11:45 |
@wiking | and you'll get that straight away | 11:45 |
sonne|work | lisitsyn: only thoralf was clever enough to not get burned | 11:45 |
sonne|work | thoralf: you have to revive the party | 11:46 |
thoralf | sonne|work: I left my right leg. | 11:46 |
sonne|work | thoralf: some magic potion please | 11:46 |
sonne|work | thoralf: cure us! | 11:46 |
@wiking | sonne|work: i mean currently we have something like this | 11:46 |
@wiking | sonne|work: in example: SGVector< SGSparseVector<float64_t> > fv; | 11:46 |
-!- sonne|work is now known as zombie|sonnewrk | 11:46 | |
@wiking | i mean this bound to be fucked | 11:46 |
zombie|sonnewrk | wiking: yeah totally not supported | 11:47 |
@wiking | it should be just SGSparseVector<float64_t> fv | 11:47 |
@wiking | or something like that | 11:47 |
zombie|sonnewrk | wiking: we didn't have these types back then | 11:47 |
zombie|sonnewrk | but yes | 11:47 |
@wiking | of course this depends on whether the template parameter is actually a class or a type | 11:47 |
zombie|sonnewrk | wiking: SGVectors should only have numeric types - full stop. | 11:48 |
thoralf | wiking: There's a difference? ;) | 11:48 |
zombie|sonnewrk | that is what they are intended for | 11:48 |
@wiking | zombie|sonnewrk: yeah | 11:48 |
zombie|sonnewrk | thoralf: ever did horse + cow = ? | 11:48 |
thoralf | zombie|sonnewrk: Even complex64 is a border case. :) | 11:48 |
@wiking | zombie|sonnewrk: but what's the use of SGVector<type> when it expects SGSparseVector<float64_t> | 11:48 |
@wiking | ? | 11:48 |
zombie|sonnewrk | thoralf: yes ... | 11:48 |
zombie|sonnewrk | wiking: no idea what this is... | 11:49 |
zombie|sonnewrk | wiking: I guess multiple sparse vectors are meant | 11:49 |
@wiking | zombie|sonnewrk: well the problem is that let's say StreamingSparseFeatures.h | 11:50 |
@wiking | it runs like this now | 11:50 |
@wiking | /// The parser object, which reads from input and returns parsed example objects. CInputParser< SGSparseVectorEntry<T> > parser; | 11:50 |
@wiking | i mean it expects SGSparseVectorEntry<T> for the get_next_example right? | 11:51 |
shogun-buildbot | build #1786 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1786 blamelist: Soeren Sonnenburg <sonne@debian.org> | 11:51 |
@wiking | and then if one checks CInputParser<T>::get_next_example | 11:51 |
@wiking | you'll see this | 11:51 |
@wiking | template <class T> int32_t CInputParser<T>::get_next_example(T* &fv, int32_t &length, float64_t &label) | 11:51 |
@wiking | fv = ex->fv.vector; | 11:51 |
@wiking | length = ex->fv.vlen; | 11:51 |
zombie|sonnewrk | wiking: it should return full vectors nor entries | 11:51 |
@wiking | and this is bound to be fucked | 11:52 |
@wiking | as T = SGSparseVectorEntry<T> | 11:52 |
zombie|sonnewrk | van51: how is the example going? | 11:52 |
@wiking | i.e SGSparseVectorEntry<T> fv | 11:52 |
@wiking | ah no | 11:52 |
@wiking | it actually could still work | 11:52 |
van51 | zombie|sonnewrk: pretty good. I just have to add the code for the back and forth communication and it'll be good to go | 11:53 |
@wiking | no not reaally | 11:53 |
@wiking | SGSparseVectorEntry<T> fv = a T array | 11:53 |
@wiking | which makes no fucking sense obviously | 11:53 |
shogun-buildbot | build #68 of FC19 - modular_interfaces is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20modular_interfaces/builds/68 blamelist: Viktor Gal <vigsterkr@gmail.com> | 11:53 |
@wiking | what it actually should do is | 11:54 |
@wiking | that ex->fv should be the same type as fv | 11:54 |
@wiking | and then you can do a fv = ex->fv | 11:54 |
@wiking | the problem there will be that sometimes the template param is a type (like float or soemthing) and sometimes it's a fucking class | 11:55 |
@wiking | even more a templated class | 11:55 |
@wiking | hehe ok | 11:55 |
shogun-buildbot | build #1708 of bsd1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1708 | 11:56 |
@wiking | what if we just convert StreamingDenseFeature to call InputParser< SGVector<type> > | 11:56 |
lisitsyn | wiking: template<template<T>> | 11:56 |
lisitsyn | ?? | 11:56 |
lisitsyn | :D | 11:56 |
@wiking | and then it's kind of solved | 11:56 |
@wiking | StreamingHashedDocDotFeatures.h: | 11:56 |
@wiking | is the other problem | 11:56 |
@wiking | that it uses CInputParser<char> parser | 11:57 |
@wiking | does it really want char one-by-one? | 11:57 |
@wiking | or rather an SGVector<char> | 11:58 |
@wiking | ??? | 11:58 |
@wiking | van51: u know this | 11:59 |
@wiking | u wrote the whole class | 11:59 |
zombie|sonnewrk | wiking: that is the least problem van51 could easily change that | 11:59 |
@wiking | i mean wtf is this | 11:59 |
@wiking | SGSparseVector<float64_t> current_vector | 11:59 |
@wiking | but the input parser is <char>-ed | 12:00 |
@wiking | ?! | 12:00 |
van51 | wiking: it need a SGVector<char>, yeah | 12:00 |
@wiking | ah ok | 12:00 |
@wiking | van51: ah ok cool | 12:00 |
@wiking | then maaaybe we can get this fixed | 12:00 |
@wiking | ok firing a feature branch | 12:00 |
@wiking | let's fucking see | 12:00 |
van51 | wiking: but I also did something weird with SparseVectorEntry in StreamingHashedSparseFeatures | 12:00 |
@wiking | yeah i mean... | 12:01 |
@wiking | :) | 12:01 |
van51 | wiking: it was the only way I could get get_example to work the way I wanted it | 12:01 |
thoralf | wiking: I'm trying to reproduce the bug and see what I can do. If you're starting something else, will my work be obsolete? | 12:01 |
@wiking | thi sis all weird | 12:01 |
@wiking | thoralf: this needs to be refactored | 12:01 |
@wiking | that's the deal | 12:01 |
van51 | wiking: but if you want me to make any changes to make this whole easier. let me know | 12:01 |
@wiking | we'll see | 12:02 |
shogun-buildbot | build #1785 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1785 blamelist: hushell <hushell@hushell-U510.(none)> | 12:02 |
thoralf | wiking: Okay, stopping now. Cannot even compile with clang | 12:02 |
thoralf | /home/tklein/Projekte/2013-03-06--msc-hashing-svm/shogun-github/src/shogun/base/class_list.cpp:18:1: error: unknown type name 'REPLACE' | 12:02 |
shogun-buildbot | build #1443 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1443 blamelist: Viktor Gal <vigsterkr@gmail.com> | 12:04 |
@wiking | zombie|sonnewrk: how do i disable to try to call a class with typenames (PT_*) a template? | 12:05 |
@wiking | ah ok | 12:06 |
@wiking | i guess #define IGNORE_IN_CLASSLIST | 12:06 |
zombie|sonnewrk | wiking: any class with C prefix will be attempted to be serialized all others not. if you add this ignore define then it will also not be in class list | 12:07 |
@wiking | it still fucking does | 12:08 |
thoralf | undefined reference to `shogun::SGVector<shogun::SGSparseVectorEntry<short> > | 12:08 |
thoralf | Yeah. | 12:08 |
thoralf | With clang. | 12:08 |
thoralf | wiking: How did you get this compile? | 12:09 |
@wiking | thoralf: i did not | 12:09 |
thoralf | wiking: How can you get the input parser error from the issue? | 12:10 |
@wiking | thoralf: no error... only if u do clang | 12:11 |
@wiking | thoralf: otherwise just gdb it | 12:11 |
thoralf | wiking: I don't get it. https://github.com/shogun-toolbox/shogun/issues/1576 suggests that you compiled with clang and got unit tests running... resulting in the described bug. | 12:12 |
thoralf | wiking: How do I get to this error? Which compiler, which cmake flags, which binary to call? | 12:12 |
thoralf | I tried 2 compilers now, but not getting to your point. | 12:13 |
@wiking | thoralf: cmake -DENABLE_UBSAN=ON -DENABLE_TESTING=ON -DENABLE_CCACHE=OFF -DCMAKE_BUILD_TYPE=Debug .. | 12:14 |
@wiking | u need clang 34 | 12:14 |
@wiking | 3.4 | 12:14 |
thoralf | Okay, only 3.2 here. | 12:14 |
@wiking | but if u gdb that example | 12:14 |
@wiking | or unittest | 12:14 |
@wiking | you'll see that at that fv will be actually a big NULL | 12:15 |
thoralf | wiking: Where did you set your break point? | 12:18 |
@wiking | thoralf: set it to InputParser.h:533 | 12:20 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Read error: Connection reset by peer] | 12:25 | |
-!- hushell [~hushell@c-98-232-178-161.hsd1.or.comcast.net] has quit [Quit: WeeChat 0.3.7] | 12:28 | |
shogun-buildbot | build #1787 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1787 blamelist: Viktor Gal <vigsterkr@gmail.com> | 12:29 |
thoralf | wiking: Oh, damn. I now see where the problem is. | 12:40 |
thoralf | wiking: Well, hope you're already refactoring? ;) | 12:40 |
@wiking | trying | 12:42 |
@wiking | but the problem is that imho InputParser is being used in a wrong way | 12:43 |
thoralf | Yeah. | 12:43 |
thoralf | Abusing templates to not change API. | 12:44 |
@wiking | zombie|sonnewrk: am i right that circular buffer and parsebuffer is kind of like a redundant code? | 12:53 |
shogun-buildbot | build #1709 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1709 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:57 |
thoralf | wiking: sure | 12:58 |
shogun-buildbot | build #80 of FC19 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20libshogun/builds/80 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:59 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun | 12:59 | |
thoralf | wiking: I finally need to say: You're problem is different from what I found | 12:59 |
shogun-buildbot | build #69 of FC19 - modular_interfaces is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20modular_interfaces/builds/69 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:59 |
thoralf | but "mine" also exists in the code I've been looking at yesterday/today :(( | 12:59 |
@wiking | hehehe great | 13:01 |
@wiking | more and more bugs :) | 13:01 |
thoralf | But refactoring will occupy you several days... it's too close to the tests and examples. | 13:02 |
shogun-buildbot | build #47 of precise - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/47 blamelist: Viktor Gal <viktor.gal@maeth.com> | 13:06 |
shogun-buildbot | build #1444 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1444 blamelist: Viktor Gal <viktor.gal@maeth.com> | 13:10 |
@wiking | hehe ok | 13:19 |
@wiking | now it failed | 13:19 |
@wiking | it's probably because of that bug | 13:19 |
shogun-notifier- | shogun: Viktor Gal :feature/FixStreaming * 057c667 / src/shogun/io/streaming/InputParser.h,src/shogun/io/streaming/ParseBuffer.h: https://github.com/shogun-toolbox/shogun/commit/057c667214ff6093c07d3dc12a88a76e6c06bae8 | 13:26 |
shogun-notifier- | shogun: InputParser and ParseBuffer requires pthread | 13:26 |
shogun-notifier- | shogun: check if it's available | 13:26 |
@wiking | first things first ;D | 13:27 |
shogun-buildbot | build #1788 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1788 blamelist: Viktor Gal <viktor.gal@maeth.com> | 13:28 |
thoralf | Why didn't you just remove pthreads? ;) | 13:30 |
thoralf | Makes less sense in the parse buffer. | 13:31 |
@wiking | why? | 13:31 |
@wiking | it actually does all the threaded read | 13:31 |
@wiking | in the background... | 13:31 |
van51 | zombie|sonnewrk: foulwall: I've made a first attempt, here is the PR | 13:54 |
van51 | zombie|sonnewrk: foulwall: let me know if something needs to be changed | 13:55 |
zombie|sonnewrk | van51: nice | 13:55 |
zombie|sonnewrk | van51: data/lang_detect/default.svm.gz | 13:58 |
zombie|sonnewrk | van51: did you add that file ^ | 13:58 |
van51 | zombie|sonnewrk: in shogun-demo, I had to edit .gitmodules to be able to get the data submodule | 13:58 |
van51 | zombie|sonnewrk: I was about to get that | 13:58 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Read error: No route to host] | 13:58 | |
zombie|sonnewrk | van51: please do that and then git add the data dir from shogun-demo to require the new data revision | 13:58 |
van51 | zombie|sonnewrk: but it didn't fetch the lang_detection folder along with the dataset I had pushed in that | 13:59 |
zombie|sonnewrk | van51: why that? | 13:59 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun | 13:59 | |
zombie|sonnewrk | van51: the way submodules work is that you check them out separately modify them to your likes | 14:00 |
zombie|sonnewrk | van51: and in the shogun and shogun-demo where the submodules are used | 14:00 |
zombie|sonnewrk | van51: you have to git pull the right version | 14:00 |
zombie|sonnewrk | van51: then git add data | 14:00 |
zombie|sonnewrk | and commit / push | 14:00 |
zombie|sonnewrk | to mark the version you want from there | 14:00 |
lisitsyn | have you guyz heard freebsd turned away from gcc? | 14:01 |
thoralf | lisitsyn: Why? | 14:04 |
@wiking | thoralf: gpl3 | 14:04 |
@wiking | it's not really bsd lincense ready | 14:04 |
lisitsyn | well and in the end clang is much better | 14:06 |
lisitsyn | but thats gpl issue indeed | 14:06 |
@wiking | :> | 14:07 |
@wiking | yeah man gcc 4.8 and it's error messages | 14:07 |
@wiking | is like fucking 1980s | 14:07 |
@wiking | same with the warnings | 14:08 |
lisitsyn | wiking: openmp was a flaw | 14:08 |
lisitsyn | but it is here | 14:08 |
lisitsyn | so | 14:08 |
lisitsyn | no idea why to use it still | 14:08 |
@wiking | hehehe | 14:08 |
lisitsyn | :) | 14:08 |
@wiking | i mean this is the other things | 14:09 |
@wiking | -rwx------ 1 buildslave buildslave 218M Sep 11 12:14 libshogun.so.14.0 | 14:09 |
@wiking | fucking instrumented debug library is smaller | 14:09 |
@wiking | than the -O3 gcc library | 14:09 |
@wiking | .. | 14:09 |
thoralf | wiking: And compared with O2? | 14:10 |
@wiking | note that one above is generated with -O0 + thread sanitizaer... | 14:10 |
thoralf | O3 is not fair. | 14:10 |
@wiking | so gcc is > /dev/null :) | 14:10 |
thoralf | wiking: I think gcc is good - as clang is. It's always good to have concurrent implementations. | 14:11 |
@wiking | thoralf: well it's good but it now currently way behind clang features | 14:11 |
lisitsyn | clang has api to reach AST | 14:12 |
@wiking | thoralf: like sanitizing components and static analysis... | 14:12 |
lisitsyn | that's enough | 14:12 |
@wiking | and that's another thing | 14:12 |
@wiking | libclang mand | 14:12 |
lisitsyn | to kill gcc | 14:12 |
@wiking | man | 14:12 |
lisitsyn | with gcc you write C++ parser each time you want something ide-like | 14:12 |
@wiking | libclang just made insane fast code completion possible | 14:12 |
@wiking | ctags mang | 14:13 |
@wiking | *an | 14:13 |
@wiking | *man | 14:13 |
@wiking | :DDD | 14:13 |
@wiking | ctags is the most fun to see :))) | 14:13 |
lisitsyn | wiking: yeah 'exuberant ctags' | 14:13 |
@wiking | :D | 14:13 |
lisitsyn | well it is ok but lets keep it in 2000s | 14:13 |
@wiking | yeah | 14:13 |
@wiking | but then again | 14:14 |
@wiking | VC++ | 14:14 |
@wiking | !!!! | 14:14 |
@wiking | that's like 1970s | 14:14 |
lisitsyn | visual studio? oh | 14:14 |
lisitsyn | I don't know how to use it though | 14:14 |
@wiking | i dont udnerstand why those guys just dont start switching to llvm | 14:14 |
lisitsyn | which guys? | 14:14 |
@wiking | m$ | 14:14 |
lisitsyn | come on | 14:14 |
@wiking | instead on maintaining a total shit compiler | 14:14 |
lisitsyn | they are not of these type of guys | 14:15 |
@wiking | that does not support any fucking standard | 14:15 |
lisitsyn | they are do it yourself guys | 14:15 |
@wiking | yeah DIY | 14:15 |
@wiking | eeeeverything | 14:15 |
lisitsyn | I am not surprised at all | 14:15 |
@wiking | we can see where that leads to | 14:15 |
@wiking | :) | 14:15 |
@wiking | IE | 14:15 |
@wiking | :> | 14:15 |
lisitsyn | they have a product for everything, that's their policy I guess | 14:15 |
@wiking | internet explorer 10 :D | 14:15 |
@wiking | fully web 2.0 compilant of course :) | 14:16 |
@wiking | html5 ready and everything | 14:16 |
@wiking | anyhow | 14:16 |
@wiking | fuckdizshit | 14:16 |
@wiking | lisitsyn: when's the meeting | 14:16 |
@wiking | lisitsyn: where's the doodle | 14:16 |
lisitsyn | wiking: okay setting it | 14:16 |
@wiking | lisitsyn: if u want to reduce LoC | 14:16 |
@wiking | throw out the inputparser/buffer | 14:16 |
@wiking | :) | 14:16 |
@wiking | as really now we are like | 14:17 |
@wiking | having parser for every specialized shit | 14:17 |
lisitsyn | wiking: there are a lot of things to throw out | 14:17 |
@wiking | but i guess 3.1 will be the 'screw u guys, we threw out everything' release :) | 14:17 |
thoralf | shogun/lib/SGCachedVector.h <-- die! | 14:18 |
@wiking | wtf is that | 14:18 |
@wiking | :D | 14:18 |
thoralf | A fixme :) | 14:18 |
lisitsyn | wiking: I want small 'apply' core | 14:18 |
@wiking | yep | 14:19 |
@wiking | that is needed | 14:19 |
lisitsyn | wiking: may be real-time as you mentioned that once ;) | 14:19 |
@wiking | lisitsyn: yeah i mean just like in case of video codecs | 14:20 |
@wiking | decoder only | 14:20 |
lisitsyn | yes yes | 14:20 |
@wiking | or encoder only | 14:20 |
@wiking | or de/encoder together | 14:20 |
lisitsyn | that must be separate .so | 14:20 |
@wiking | y | 14:20 |
lisitsyn | wiking: I believe ideally 'apply' one should be MIT/BSD | 14:20 |
@wiking | that'd be better ye | 14:20 |
lisitsyn | so people can be not afraid of gpl stuff | 14:20 |
@wiking | yep yep | 14:21 |
@wiking | good idea totally second that | 14:21 |
lisitsyn | http://doodle.com/4ctd4cwr588nwe52 | 14:23 |
thoralf | wiking: ever tried clang-modernize? | 14:25 |
van51 | zombie|sonnewrk: I've pulled the latest version alright but how do I push my commit for the submodule now? | 14:27 |
zombie|sonnewrk | van51: what do you mean? | 14:27 |
van51 | zombie|sonnewrk: I've added in data/ a saved svm and commited in the submodule | 14:28 |
zombie|sonnewrk | van51: so you added it to shogun-data? then send a PR for that | 14:30 |
van51 | zombie|sonnewrk: https://github.com/shogun-toolbox/shogun-data/pull/15 | 14:33 |
shogun-notifier- | shogun-data: van51 :master * f8d4cac / lang_detection/default.svm.gz: https://github.com/shogun-toolbox/shogun-data/commit/f8d4cac365f03189b1f5b294b6e3002d79b9cd7d | 14:34 |
shogun-notifier- | shogun-data: Added default svm for language detection | 14:34 |
shogun-notifier- | shogun-data: Soeren Sonnenburg :master * a98cfe7 / lang_detection/default.svm.gz: https://github.com/shogun-toolbox/shogun-data/commit/a98cfe7bf22330abe0a4b4cf8af49e5682f35d26 | 14:34 |
shogun-notifier- | shogun-data: Merge pull request #15 from van51/master | 14:34 |
shogun-notifier- | shogun-data: | 14:34 |
shogun-notifier- | shogun-data: Added default svm for language detection | 14:34 |
zombie|sonnewrk | van51: now go to shogun-demo | 14:34 |
zombie|sonnewrk | and git add data | 14:34 |
zombie|sonnewrk | and PR again | 14:34 |
zombie|sonnewrk | or if you have not updated data there | 14:34 |
zombie|sonnewrk | go to the submodule | 14:34 |
zombie|sonnewrk | and git pull from there | 14:34 |
van51 | zombie|sonnewrk: ok done :) | 14:35 |
zombie|sonnewrk | then commit and send a PR | 14:36 |
foulwall | Hi zombie|sonnewrk , how can I add a HAVE_CURL flag like HAVE_HDF5, in CMakeList FIND_PACKAGE? | 14:36 |
van51 | zombie|sonnewrk: I've updated the PR as well | 14:36 |
zombie|sonnewrk | wiking: can you help foulwall with that? | 14:37 |
foulwall | wiking: I did this | 14:37 |
foulwall | FIND_PACKAGE(CURL) | 14:37 |
foulwall | IF (CURL_FOUND) | 14:37 |
foulwall | SET(HAVE_CURL 1) | 14:37 |
foulwall | INCLUDE_DIRECTORIES(${CURL_INCLUDE_DIR}) | 14:37 |
foulwall | LIST(APPEND DEFINES HAVE_CURL) | 14:37 |
foulwall | SET(CURL_LIBRARIES "${CURL_LIBRARY}") | 14:37 |
foulwall | SET(POSTLINKFLAGS ${POSTLINKFLAGS} ${CURL_LIBRARIES}) | 14:37 |
foulwall | ENDIF(CURL_FOUND) | 14:37 |
@wiking | ah u need curl? | 14:38 |
@wiking | foulwall: doesn't work? | 14:38 |
@wiking | aaaalthough | 14:38 |
@wiking | INCLUDE_DIRECTORIES(${CURL_INCLUDE_DIR}) | 14:38 |
@wiking | not the best | 14:38 |
@wiking | nonono | 14:39 |
@wiking | foulwall: you have a pr? | 14:39 |
@wiking | or not yet? | 14:39 |
foulwall | wiking: https://github.com/shogun-toolbox/shogun/pull/1566 | 14:39 |
@wiking | i'll comment there | 14:40 |
@wiking | it's easier | 14:40 |
foulwall | wiking: ok:) thanks | 14:40 |
zombie|sonnewrk | van51: I gave some minor comments. The rest is sth foulwall should comment | 14:40 |
van51 | zombie|sonnewrk: yeah I saw them. I'm making the changes now | 14:41 |
thoralf | PR time? wiking: https://github.com/shogun-toolbox/shogun/pull/1570 | 14:41 |
foulwall | ah zombie|sonnewrk I'll check | 14:42 |
@wiking | foulwall: there are the comments | 14:43 |
@wiking | foulwall: with those changes if FindCURL finds curl then it should work | 14:44 |
foulwall | Ok wiking , I'll have a try | 14:45 |
foulwall | van51: I'll make the demo tonight:) | 14:45 |
van51 | foulwall: cool :) let me know when it's ready! | 14:46 |
besser82 | zombie|sonnewrk, wiking: What purpose is the new optional dependency on ARPRECK? | 14:50 |
zombie|sonnewrk | besser82: ask lisitsyn! | 14:50 |
besser82 | kk, will do ;) | 14:51 |
besser82 | lisitsyn: What purpose is the new optional dependency on ARPRECK? | 14:51 |
lisitsyn | besser82: arpack you mean? | 14:53 |
lisitsyn | besser82: dimensionality reduction module depends on it | 14:53 |
besser82 | lisitsyn: no, I'm talking about ARPREC ;) | 14:54 |
besser82 | -- The following OPTIONAL packages have not been found: | 14:54 |
besser82 | * CPLEX | 14:54 |
besser82 | * NLopt | 14:54 |
besser82 | * ColPack | 14:54 |
besser82 | * ARPREC | 14:54 |
besser82 | lisitsyn: ^^^ | 14:54 |
besser82 | lisitsyn: NLopt and ColPack are clear to me, but no ARPREC :) | 14:55 |
@wiking | besser82: git grep ARPREC | 14:55 |
@wiking | besser82: src/shogun/mathematics/JacobiEllipticFunctions.h | 14:55 |
besser82 | wiking, lisitsyn: kk, thanks ;) So it's another nifty enhancement ;) | 14:56 |
zombie|sonnewrk | besser82: now I know that is sth Heiko and lambday are doing | 14:57 |
besser82 | zombie|sonnewrk: kk, so I'll ask them what this is about in detail ;) Thanks! | 15:01 |
foulwall | wiking: It really works. thanks:) | 15:02 |
lisitsyn | besser82: arprec is lambday's gsoc project | 15:03 |
lisitsyn | ah | 15:03 |
lisitsyn | already answered | 15:03 |
lisitsyn | :) | 15:03 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 60ae76c / src/shogun/features/SparseFeatures.cpp,src/shogun/features/StringFeatures.cpp: https://github.com/shogun-toolbox/shogun/commit/60ae76c153f67f684d5732aa2d3484dd8739745c | 15:03 |
shogun-notifier- | shogun: fix not implemented error in sparse/string feature constructors - call loaders at the right place | 15:03 |
shogun-buildbot | build #2109 of deb1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2109 blamelist: Soeren Sonnenburg <sonne@debian.org> | 15:07 |
-!- van51 [~van51@athedsl-410351.home.otenet.gr] has quit [Quit: Leaving.] | 15:08 | |
foulwall | Hi wiking , how can I start a unit test? I tried make tests, but It seems just run over the examples. | 15:37 |
foulwall | Oh I have found the README.cmake :) | 15:39 |
foulwall | s | 15:40 |
@wiking | foulwall: if u add a new unit test dont forget to re-run cmake.! | 15:51 |
thoralf | foulwall: If you found out something useful, feel free to add it to the README. | 15:53 |
thoralf | We should rename it to WRITEME.cmake. | 15:53 |
-!- pickle27 [~Kevin@199.119.128.114] has joined #shogun | 16:33 | |
-!- foulwall [~zhengyang@114.255.40.22] has quit [Ping timeout: 256 seconds] | 16:34 | |
-!- zhengyangl [~zhengyang@114.255.40.22] has joined #shogun | 17:18 | |
-!- zhengyangl is now known as foulwall | 17:18 | |
-!- foulwall [~zhengyang@114.255.40.22] has quit [Ping timeout: 268 seconds] | 17:31 | |
-!- foulwall [~zhengyang@114.255.40.22] has joined #shogun | 17:31 | |
-!- van51 [~van51@athedsl-410351.home.otenet.gr] has joined #shogun | 17:42 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 18:03 | |
-!- pickle27 [~Kevin@199.119.128.114] has quit [Quit: Leaving] | 18:26 | |
-!- van51 [~van51@athedsl-410351.home.otenet.gr] has quit [Quit: Leaving.] | 19:04 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Read error: No route to host] | 19:53 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 19:54 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 246 seconds] | 20:00 | |
-!- van51 [~van51@athedsl-410351.home.otenet.gr] has joined #shogun | 20:07 | |
-!- sonne|osx [~sonne@89.204.135.73] has joined #shogun | 20:18 | |
-!- lisitsyn1 [~lisitsyn@fb2-lo1.global63.net] has joined #shogun | 20:20 | |
sonne|osx | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 20:21 |
shogun-buildbot | build forced [ETA 34m11s] | 20:21 |
shogun-buildbot | I'll give a shout when the build finishes | 20:21 |
-!- sonne|osx [~sonne@89.204.135.73] has quit [Client Quit] | 20:22 | |
lisitsyn1 | ha! voyager is out of solar system | 20:31 |
@wiking | x`ha who knows | 20:40 |
@wiking | :) | 20:40 |
@wiking | it's not 100% certain | 20:40 |
@wiking | lisitsyn1: how would u do the cut of the methods? | 20:40 |
@wiking | #ifdef? | 20:40 |
@wiking | train/apply | 20:40 |
lisitsyn1 | wiking: what's that about? | 20:44 |
@wiking | well first of all it's not agreed by everybody that voyager is really out of solar system | 20:44 |
@wiking | the other question was about | 20:44 |
@wiking | how would u cut the library? | 20:44 |
lisitsyn1 | wiking: ah yeah that's just from news | 20:44 |
lisitsyn1 | wiking: ahh like 'apply' library? | 20:44 |
@wiking | yeah | 20:44 |
@wiking | would u cut it with ifdef-s? | 20:45 |
lisitsyn1 | no, definitely not ifdef | 20:45 |
lisitsyn1 | separate target | 20:45 |
@wiking | or you would create the different h/cpp s | 20:45 |
@wiking | yeah i know that different target | 20:45 |
@wiking | but still how would you do the cut | 20:45 |
@wiking | in the current source | 20:45 |
lisitsyn1 | wiking: I'd go for some different h/cpps | 20:46 |
@wiking | aha | 20:46 |
@wiking | it'll be a hard task | 20:46 |
@wiking | :) | 20:46 |
lisitsyn1 | wiking: it needs some redesign I think | 20:46 |
@wiking | yeah | 20:46 |
@wiking | full redesign | 20:46 |
@wiking | imho that would the good opportunity | 20:46 |
@wiking | to introduce the D-ptrs | 20:47 |
lisitsyn1 | wiking: not only D-ptrs but shared_ptrs too | 20:47 |
@wiking | yeah | 20:47 |
lisitsyn1 | wiking: but actually hey | 20:47 |
@wiking | SG_REF > /dev/null | 20:47 |
lisitsyn1 | why do we need shared_ptrs | 20:47 |
lisitsyn1 | we can use real instances I mean | 20:47 |
@wiking | well holding refs to the objs would be cool for swig | 20:48 |
@wiking | no? | 20:48 |
@wiking | ok no other way then rewrite the fucking streaming shit | 20:49 |
lisitsyn1 | wiking: I don't know how does it work w/o pointers | 20:49 |
@wiking | as currently that's a complete leak/deref by zero | 20:49 |
@wiking | black whole | 20:49 |
@wiking | i mean if we would do stats about mailing list | 20:49 |
@wiking | it would be 1) mkl | 20:49 |
@wiking | 2) model selection | 20:50 |
@wiking | 3) streaming | 20:50 |
@wiking | :) | 20:50 |
@wiking | heheh it'd be actually funny to run some shogun task on the mailing list | 20:50 |
@wiking | :))) | 20:50 |
@wiking | as a demo | 20:50 |
@wiking | anybody out there who would join the task? | 20:50 |
lisitsyn1 | wiking: hah I am out you know | 20:50 |
@wiking | lisitsyn1: yeah when r u back? | 20:50 |
@wiking | this year | 20:51 |
@wiking | or you are full now till the end of this year? | 20:51 |
lisitsyn1 | wiking: well | 20:51 |
lisitsyn1 | wiking: I am trying to get rid of second job I accidentally took | 20:51 |
lisitsyn1 | :D | 20:51 |
@wiking | heheh | 20:51 |
@wiking | same for me | 20:51 |
lisitsyn1 | wiking: then I'll try to recover and start shoguning again | 20:51 |
@wiking | hehe ok | 20:51 |
@wiking | btw | 20:52 |
@wiking | cv::mat -> sgmatrix | 20:52 |
lisitsyn1 | wiking: oh yeah | 20:52 |
@wiking | something shall be done about this | 20:52 |
@wiking | i mean | 20:52 |
@wiking | actually | 20:52 |
@wiking | cv::mat <-> sgmatrix | 20:52 |
lisitsyn1 | yes there must be bridge | 20:52 |
lisitsyn1 | between them | 20:52 |
@wiking | lisitsyn1: i mean they started to use eigen no? | 20:53 |
lisitsyn1 | wiking: haven't looked into internals tbh | 20:53 |
@wiking | http://stackoverflow.com/questions/14783329/opencv-cvmat-and-eigenmatrix | 20:53 |
lisitsyn1 | wiking: sometimes I hate cv::Mat I have to admit | 20:54 |
@wiking | yeah | 20:54 |
@wiking | true that | 20:54 |
@wiking | still | 20:54 |
@wiking | we need a function at least | 20:55 |
@wiking | with #ifdef | 20:55 |
lisitsyn1 | wiking: I need it right now | 20:55 |
lisitsyn1 | if I want to use shogun | 20:55 |
@wiking | yeah usually me too | 20:56 |
lisitsyn1 | wiking: but once I send them my code I have to think what to do about it | 20:56 |
lisitsyn1 | because shogun is gpl and they'll definitely be afraid of it | 20:56 |
@wiking | hehehe | 20:57 |
lisitsyn1 | that's the thing that bothers me and makes me think we need less restrictive 'apply' core | 20:57 |
-!- van51 [~van51@athedsl-410351.home.otenet.gr] has quit [Quit: Leaving.] | 20:58 | |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Disconnected by services] | 21:00 | |
-!- lisitsyn1 is now known as lisitsyn | 21:00 | |
-!- mode/#shogun [+o lisitsyn] by ChanServ | 21:00 | |
@lisitsyn | ha | 21:00 |
@lisitsyn | special ops | 21:00 |
-!- lisitsyn1 [~lisitsin@mxs.kg.ru] has joined #shogun | 21:00 | |
-!- lisitsyn1 was kicked from #shogun by lisitsyn [lisitsyn1] | 21:01 | |
@lisitsyn | go away | 21:01 |
@wiking | :D | 21:06 |
@wiking | lisitsyn: doable | 21:06 |
@wiking | lisitsyn: in time :) | 21:06 |
@wiking | lisitsyn: shogun 3.5 | 21:06 |
@lisitsyn | wiking: yeah | 21:06 |
@lisitsyn | wiking: and I like to push that idea with static identifiers | 21:07 |
@lisitsyn | instead of diy getters/setters | 21:07 |
@wiking | static identifiers? | 21:08 |
@lisitsyn | wiking: get(width) | 21:08 |
@lisitsyn | with width being some static Keyword width; | 21:08 |
@wiking | ok once again | 21:08 |
@wiking | wtF? :) | 21:08 |
@lisitsyn | wiking: haha | 21:09 |
@wiking | A culinary project for a professional holiday - Programmers Day, celebrated on the 256th day of the year. | 21:09 |
@wiking | This year Programmers Day is Friday the 13th of September! | 21:09 |
@wiking | lol | 21:09 |
@wiking | tomorrow is 13th of friday | 21:09 |
@wiking | :) | 21:09 |
@wiking | didn't know | 21:09 |
* wiking wonders whether it's worth to fix streamingfile | 21:09 | |
@lisitsyn | wiking: https://github.com/shogun-toolbox/shogun/issues/1265 | 21:09 |
@wiking | or just throw the shit out | 21:09 |
@wiking | and use linereader | 21:09 |
@wiking | lisitsyn: ah ok i see | 21:10 |
@wiking | so you kind of want to have a get/set function | 21:10 |
@lisitsyn | wiking: yeah once defined in base class | 21:10 |
@wiking | which would be a SGParamater like stuff | 21:10 |
@lisitsyn | yeah I'd like to rewrite it | 21:10 |
@wiking | yeah yeah i got it | 21:10 |
@wiking | good idea | 21:11 |
@wiking | the whole thing would be more consistent then | 21:11 |
@wiking | i mean the API | 21:11 |
@wiking | yeah i'm totally up for that | 21:11 |
@lisitsyn | wiking: and we can easily add new parameters | 21:11 |
@wiking | but then as said | 21:11 |
@wiking | merge it with Parameter somehow | 21:11 |
@lisitsyn | wiking: that's first in my list when I get some time | 21:12 |
@lisitsyn | well I can try sooner but I am afraid to give up early | 21:12 |
@lisitsyn | its getting harder to focus on sth with too many things :D | 21:12 |
@wiking | heheh yeah | 21:12 |
@wiking | shogun is getting fucking messy | 21:12 |
@wiking | i mean now really | 21:12 |
@wiking | we have this streaming shit | 21:12 |
@wiking | that has it's own reader/buffering mechanizm | 21:13 |
@wiking | and on top of this | 21:13 |
@wiking | we now has new reader/buffering mechanizm for input readers | 21:13 |
@lisitsyn | wiking: I believe the main value of the code is that we know how to do things | 21:13 |
@lisitsyn | and how not to do things | 21:13 |
@wiking | yeah | 21:13 |
@lisitsyn | code is of no value for me actually | 21:13 |
@wiking | now we managed to blow up quite some the codebase | 21:13 |
@wiking | and it was like | 21:13 |
@wiking | 'we want aaaal the features possible' | 21:13 |
@wiking | and now like, ok great but we cannot fucking manage this anymore | 21:14 |
@lisitsyn | yeah we now know we don't need all the features | 21:14 |
@lisitsyn | :D | 21:14 |
@wiking | and that we have like redudant code | 21:14 |
@wiking | all over the place | 21:14 |
@lisitsyn | wiking: well we need to become modular again | 21:15 |
@lisitsyn | wiking: like we have transfer learning code | 21:16 |
@lisitsyn | it just have to be a separate library | 21:16 |
@lisitsyn | swig requires us to be modular but may be there is a way | 21:16 |
@lisitsyn | IDK | 21:16 |
@lisitsyn | wiking: it would be much nicer to have multiple .so's | 21:17 |
@lisitsyn | with some being just optional | 21:17 |
@lisitsyn | but swig strikes back yeah.. | 21:17 |
@wiking | yep yep | 21:18 |
@lisitsyn | wiking: there is a way well.. | 21:18 |
@lisitsyn | wiking: hide real classes | 21:18 |
@lisitsyn | but that's string-driven - we have seen that somewhere | 21:19 |
@lisitsyn | may be static interfaces :D | 21:19 |
-!- zxtx [~zv@rrcs-76-79-81-162.west.biz.rr.com] has joined #shogun | 21:29 | |
-!- gsomix [~gsomix@185.6.239.51] has joined #shogun | 21:30 | |
gsomix | good evening | 21:30 |
@wiking | gsomix: helloooooo | 21:32 |
gsomix | sonney2k, I need the advice. I'm implementing now splitting SGVector for packing into protobuf messages. | 21:32 |
gsomix | protobuf works with file descriptors and c++ fstreams. | 21:33 |
gsomix | we prefer works with C's FILE*. | 21:34 |
gsomix | for now I see simple solution. reading messages in byte arrays by fread | 21:34 |
@wiking | gsomix: FILE * fdopen(int fildes, const char *mode); | 21:34 |
@wiking | gsomix: the first argu is the file descriptor | 21:35 |
gsomix | wiking, I'm not sure. is this function in standart? | 21:35 |
@wiking | The fdopen() function associates a stream with the existing file descrip- tor, fildes. The mode of the stream must be compatible with the mode of the file descriptor. When the stream is closed via fclose(3), fildes is closed also. | 21:35 |
@wiking | gsomix: man fdopen | 21:36 |
@wiking | #include <stdio.h> | 21:36 |
@wiking | LIBRARY Standard C Library (libc, -lc) | 21:36 |
@wiking | it cannot get more standard ;) | 21:36 |
-!- travis-ci [~travis-ci@ec2-54-242-252-41.compute-1.amazonaws.com] has joined #shogun | 21:36 | |
travis-ci | [travis-ci] it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11275384 | 21:36 |
-!- travis-ci [~travis-ci@ec2-54-242-252-41.compute-1.amazonaws.com] has left #shogun [] | 21:36 | |
gsomix | wiking, I'm just not see it there: http://www.cplusplus.com/reference/cstdio/ | 21:36 |
gsomix | hm. if this is standard everything is ok. | 21:37 |
@wiking | this should be really stanard | 21:37 |
@wiking | *standard | 21:37 |
@wiking | even microsoft has it ;) | 21:37 |
@wiking | http://msdn.microsoft.com/en-us/library/ms235351.aspx | 21:38 |
@wiking | hahahah | 21:38 |
@wiking | "This POSIX function is deprecated. Use the ISO C++ conformant _fdopen instead" | 21:38 |
@wiking | (but that's just m$) dont worry | 21:38 |
gsomix | wiking, btw, I also need some help with cmake. look, I added protobuf detection, but SET(USE_PROTOBUF 1) doesn't work. maybe I forgot declare sth... | 21:39 |
@wiking | yeah | 21:39 |
@wiking | do u have a PR for that? | 21:39 |
@wiking | but most probably you forgot do add USE_PROTOBUF to ./src/shogun/lib/config.h.in | 21:39 |
@wiking | as | 21:39 |
@wiking | #cmakedefine USE_PROTOBUF 1 | 21:39 |
@wiking | right? | 21:39 |
gsomix | oops | 21:39 |
gsomix | right | 21:39 |
@wiking | and dont forget as well | 21:40 |
@wiking | LIST(APPEND DEFINES USE_PROTOBUF) | 21:40 |
@wiking | ;) | 21:40 |
gsomix | yep, there is | 21:40 |
@wiking | okok | 21:41 |
@wiking | just making it sure | 21:41 |
@wiking | add that line to ./src/shogun/lib/config.h.in | 21:41 |
@wiking | and then you'll get a #define USE_PROTOBUF 1 | 21:41 |
@wiking | in shogun/lib/config.h | 21:41 |
@wiking | streaming is maaadnessss | 21:43 |
gsomix | wiking, great, that's work. | 21:44 |
@lisitsyn | what's work? | 21:44 |
@lisitsyn | :D | 21:44 |
gsomix | awww | 21:45 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 21:45 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 21:45 | |
gsomix | *this works | 21:45 |
@lisitsyn | la iglesiasg de developeres in da house | 21:46 |
@iglesiasg | lisitsyn, haha hey there | 21:46 |
@iglesiasg | lisitsyn, el iglesiasg, la is for feminine hehe | 21:46 |
@wiking | :> | 21:47 |
@lisitsyn | iglesiasg: nah I just speak some non-sense don't try to understand it | 21:48 |
@lisitsyn | :D | 21:48 |
@wiking | hah fucking helloooooo | 21:52 |
@wiking | helloooooo | 21:52 |
@wiking | !@#!@#!@!@$!@#!@# | 21:52 |
@wiking | man i'm going to get now really upset | 21:52 |
@wiking | i mean Heiko detected a leak | 21:52 |
@wiking | and thus he fixed it by not allocating memory | 21:53 |
@wiking | but hello it's going to leak if you force the parser not TO free memory | 21:53 |
@wiking | after destroy | 21:53 |
@wiking | all the fucking streaming features has this | 21:53 |
@wiking | src/shogun/features/streaming/StreamingDenseFeatures.cpp: parser.set_free_vectors_on_destruct(false); | 21:53 |
@wiking | src/shogun/features/streaming/StreamingHashedDenseFeatures.cpp: parser.set_free_vectors_on_destruct(false); | 21:53 |
@wiking | src/shogun/features/streaming/StreamingHashedDocDotFeatures.cpp: parser.set_free_vectors_on_destruct(false); | 21:53 |
@wiking | src/shogun/features/streaming/StreamingHashedSparseFeatures.cpp: parser.set_free_vectors_on_destruct(false); | 21:53 |
@wiking | src/shogun/features/streaming/StreamingStringFeatures.cpp: parser.set_free_vectors_on_destruct(false); | 21:53 |
@wiking | obviously they'll 'leak' as it's ment to be to free by the USER | 21:53 |
@wiking | !@#!@%@#$%!$ | 21:53 |
gsomix | wiking, aww, I'm stupid, man. I don't need fd to FILE* conversion, I need FILE* to fd... fileno, right? | 21:55 |
@wiking | fd | 21:57 |
@wiking | filedescr | 21:57 |
@wiking | but yeah | 21:57 |
@wiking | fileno will give you that | 21:58 |
gsomix | thanks! | 22:00 |
shogun-buildbot | build #1789 of deb3 - modular_interfaces is complete: Failure [failed compile csharp_modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1789 | 22:08 |
@iglesiasg | wiking, lisitsyn: can you lend me a hand a moment guys? | 22:08 |
-!- pickle27 [~Kevin@199.119.128.114] has joined #shogun | 22:09 | |
@wiking | iglesiasg: what'janeed | 22:10 |
@iglesiasg | wiking, I have added this new class CLMNNStatistics in LMNN, https://github.com/iglesias/shogun/commit/04974102eced8664378d7f99d5a2c5b0252d6450 | 22:10 |
gsomix | Friday ,13 in my time zone! | 22:11 |
@iglesiasg | wiking, very simple stuff. But when I compile in Release mode, clone unit tests, another for serialize, and another one fail | 22:11 |
@iglesiasg | wiking, I have just realize that they work fine in Debug mode though | 22:11 |
@wiking | gsomix: \o/ | 22:12 |
@wiking | mmm | 22:12 |
@iglesiasg | wiking, does that make sense to you? | 22:13 |
@wiking | iglesiasg: i think then you'll need to debug the Release | 22:13 |
@wiking | well not really, but then again | 22:13 |
@wiking | release is optimized quite heavily | 22:13 |
@wiking | so maybe the compiler does something that it shouldn't :( | 22:13 |
@iglesiasg | that's unfortunate, for some reason my valgrind always breaks in release :( | 22:13 |
@iglesiasg | I will try in the fatbot | 22:13 |
@wiking | yeah valgrind is not ment | 22:13 |
@wiking | meant | 22:13 |
@wiking | to use with optimized code | 22:14 |
@wiking | iglesiasg: "Use of -O2 and above is not recommended as Memcheck occasionally reports uninitialised-value errors which don't really exist." | 22:14 |
@wiking | and release is -o3 | 22:14 |
@iglesiasg | I see | 22:14 |
@wiking | so that for sure wont really work ( | 22:14 |
@wiking | :< | 22:15 |
@iglesiasg | wiking, in Release when the test failed I was getting this warning that the there is probably something wrong with the class like get_name, or initialization | 22:16 |
@iglesiasg | I checked many times though | 22:16 |
@wiking | mmmm | 22:20 |
@wiking | iglesiasg: worth to just watch what's the output of clone | 22:20 |
@wiking | as there's a lot of things we test there | 22:20 |
@wiking | like get_name | 22:20 |
@wiking | i think after shogun release 3.0 i can call myself a test engineer :D | 22:21 |
@wiking | fucking hell i hate that | 22:21 |
@wiking | :P | 22:21 |
@wiking | shogun-buildbot: force build --branch=develop 'deb1 - libshogun' | 22:22 |
shogun-buildbot | build forced [ETA 51m02s] | 22:22 |
shogun-buildbot | I'll give a shout when the build finishes | 22:22 |
@wiking | 51 mins | 22:22 |
@wiking | wtF? | 22:22 |
shogun-buildbot | build #2110 of deb1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2110 | 22:25 |
-!- sonne|osx [~sonne@89.204.135.73] has joined #shogun | 22:29 | |
sonne|osx | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 22:30 |
-!- travis-ci [~travis-ci@ec2-54-242-252-41.compute-1.amazonaws.com] has joined #shogun | 22:30 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/11278899 | 22:30 |
-!- travis-ci [~travis-ci@ec2-54-242-252-41.compute-1.amazonaws.com] has left #shogun [] | 22:30 | |
@wiking | sonne|osx: it's coming... | 22:30 |
sonne|osx | wiking: what? | 22:30 |
@wiking | the modular | 22:30 |
@wiking | as libshogun just finished | 22:30 |
@wiking | 22:25 < shogun-buildbot> build #2110 of deb1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2110 | 22:30 |
@wiking | mmm although it has not triggered the others :S | 22:31 |
@wiking | strange maybe force build does not do this | 22:31 |
sonne|osx | wiking: makes sense | 22:31 |
@wiking | mmm | 22:32 |
sonne|osx | shogun-buildbot: force build branch=develop 'deb3 - modular_interfaces' | 22:32 |
shogun-buildbot | no such builder 'branch=develop' | 22:32 |
@wiking | wwhat does not make sense | 22:32 |
@wiking | is streaming buffering :D | 22:32 |
@wiking | i'm reading now a line | 22:32 |
@wiking | where one takes out a free memory space (buffer) writes into it | 22:32 |
@wiking | and then again | 22:32 |
sonne|osx | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 22:32 |
shogun-buildbot | build forced [ETA 34m11s] | 22:32 |
shogun-buildbot | I'll give a shout when the build finishes | 22:32 |
@wiking | as if it wasn't written in it | 22:32 |
@wiking | copies it into | 22:33 |
@wiking | :) | 22:33 |
@wiking | again | 22:33 |
@wiking | :D | 22:33 |
sonne|osx | wiking: what? | 22:33 |
@wiking | yeah exactly | 22:33 |
sonne|osx | hey gsomix! | 22:34 |
gsomix | sonne|osx, hey ho | 22:34 |
@wiking | https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/io/streaming/InputParser.h#L532 | 22:34 |
@wiking | sonne|osx: ^ | 22:34 |
@wiking | this line gets a memory space from the ring buffer | 22:34 |
sonne|osx | gsomix: I have an issue - how can I read a file line by line into a string list | 22:34 |
sonne|osx | gsomix: I mean does any of the files do this? | 22:34 |
@wiking | https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/io/streaming/InputParser.h#L538 | 22:34 |
@wiking | this writes actually data in that memory | 22:35 |
sonne|osx | wiking: you should use gsomix' stuff for that | 22:35 |
sonne|osx | wiking: he has a line reader and a ring buffer | 22:35 |
gsomix | sonne|osx, CSVFile | 22:35 |
@wiking | sonne|osx: yeah i know | 22:35 |
@wiking | sonne|osx: it's not threaded though | 22:35 |
@wiking | sonne|osx: https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/io/streaming/InputParser.h#L555 | 22:35 |
@wiking | and this writes back the stuff into the buffer that's already in the buffer | 22:35 |
sonne|osx | wiking: well you would have to start the read in an extra thread of course... | 22:35 |
@iglesiasg | wiking, this is really weird, the tests are now fine in Release too | 22:36 |
sonne|osx | argh man this needs to be rewritten :/ | 22:36 |
@wiking | iglesiasg: :< | 22:36 |
@wiking | sonne|osx: yeah i was thinking about using linebuffer | 22:36 |
sonne|osx | gsomix: so CSVFile would read things line by line right? | 22:36 |
sonne|osx | gsomix: I mean strings | 22:36 |
@iglesiasg | wiking, I don't really get it. But if there is no error.. I think I am going to issue a PR and see what travis thinks | 22:36 |
gsomix | sonne|osx, CSVFile can read string list. | 22:36 |
sonne|osx | gsomix: so when I do StringFeatures(CSVFile(fname), DNA) it should work | 22:37 |
@wiking | iglesiasg: yeah travis today seems to be a bit sick | 22:37 |
@iglesiasg | wiking, do you agree with that or do you think it could be dangerous if any bug slips through? | 22:37 |
sonne|osx | gsomix: then one line == one string | 22:37 |
gsomix | yep | 22:37 |
@wiking | iglesiasg: run all the sanitizers through it :))) | 22:37 |
@iglesiasg | wiking, what sanitizers? :) | 22:37 |
sonne|osx | gsomix: ok then I hope the error I fixed fixes the bot | 22:37 |
sonne|osx | gsomix: I am trying to get rid of ascii file that is why I am askingg | 22:38 |
sonne|osx | gsomix: alright then how is the status with protobuf? | 22:38 |
@wiking | iglesiasg: http://clang.llvm.org/docs/UsersManual.html#controlling-code-generation | 22:38 |
sonne|osx | gsomix: my estimate is that this takes you 1-2 weeks so lets not delay it | 22:39 |
@wiking | iglesiasg: you can enable it via ENABLE_TSAN=ON | 22:39 |
@wiking | iglesiasg: of course u need a quite recent clang | 22:39 |
@iglesiasg | wiking, 3.0.6 here right now, I guess I could get a newer one | 22:40 |
@wiking | iglesiasg: yeah aim for 3.4 | 22:40 |
@wiking | ;) | 22:40 |
sonne|osx | gsomix: ? | 22:41 |
gsomix | sonne|osx, fine. documentation is read, protobuf in cmake is enabled. now I'm writing splitting SGVector into set of protobuf messages. | 22:42 |
gsomix | sonne|osx, got it. | 22:42 |
sonne|osx | gsomix: please send a PR already tonight so I can give feedback | 22:42 |
sonne|osx | gsomix: I've done a couple of protobuf projects already so if we get the base right everything else will be trivial | 22:43 |
sonne|osx | gsomix: but sounds good! | 22:43 |
sonne|osx | wiking: agh nice cyg1 is back | 22:44 |
sonne|osx | wiking: when home I hope I can fix at least the build | 22:44 |
sonne|osx | wiking: which reminds me - did you enable the tests / example checks for the old static interfaces? | 22:46 |
@wiking | aha | 22:46 |
@wiking | no | 22:46 |
@wiking | *haha | 22:46 |
sonne|osx | wiking: shall I file a bug or will you remember? | 22:48 |
sonne|osx | alright | 22:49 |
sonne|osx | leaving train | 22:49 |
sonne|osx | brb | 22:49 |
-!- sonne|osx [~sonne@89.204.135.73] has quit [Quit: sonne|osx] | 22:49 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 22:51 | |
shogun-notifier- | shogun: hushell :develop * a413670 / / (9 files): https://github.com/shogun-toolbox/shogun/commit/a4136707269b80e08549cc93be9dbc3625026596 | 22:51 |
shogun-notifier- | shogun: python modular of FactorGraphModel | 22:51 |
shogun-notifier- | shogun: Fernando Iglesias :develop * 168d6db / / (9 files): https://github.com/shogun-toolbox/shogun/commit/168d6db250855cc80f53fe21fd027ab8d52e204b | 22:51 |
shogun-notifier- | shogun: Merge pull request #1577 from hushell/develop | 22:51 |
shogun-notifier- | shogun: | 22:51 |
shogun-notifier- | shogun: python modular of FactorGraphModel | 22:51 |
shogun-buildbot | build #81 of FC19 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20libshogun/builds/81 | 22:59 |
shogun-buildbot | build #1710 of bsd1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1710 | 23:02 |
shogun-buildbot | build #48 of precise - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/48 | 23:07 |
shogun-buildbot | build #1445 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1445 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 23:11 |
shogun-buildbot | build #70 of FC19 - modular_interfaces is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/FC19%20-%20modular_interfaces/builds/70 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>, hushell <hushell@hushell-U510.(none)> | 23:14 |
* sonney2k Re | 23:26 | |
shogun-buildbot | build #1446 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1446 blamelist: hushell <hushell@hushell-U510.(none)> | 23:28 |
@sonney2k | shogun-buildbot, force build --branch=develop 'cyg1 - libshogun' | 23:29 |
shogun-buildbot | build forced [ETA 41m58s] | 23:29 |
shogun-buildbot | I'll give a shout when the build finishes | 23:29 |
shogun-notifier- | shogun-demo: van51 :master * dc93f60 / / (5 files): https://github.com/shogun-toolbox/shogun-demo/commit/dc93f609a3783705af15b7ae98a46e08d9241d37 | 23:32 |
shogun-notifier- | shogun-demo: Added lang_detect backend for the demo | 23:32 |
shogun-notifier- | shogun-demo: Soeren Sonnenburg :master * 408e1ed / / (5 files): https://github.com/shogun-toolbox/shogun-demo/commit/408e1edcddb171a20e892103aa663e67416d933b | 23:32 |
shogun-notifier- | shogun-demo: Merge pull request #22 from van51/master | 23:32 |
shogun-notifier- | shogun-demo: | 23:32 |
shogun-notifier- | shogun-demo: Added lang_detect backend for the demo | 23:32 |
-!- sonne|osx [~sonne@f053046093.adsl.alicedsl.de] has joined #shogun | 23:41 | |
@sonney2k | man there is no way that cygwin could be slower | 23:54 |
--- Log closed Fri Sep 13 00:00:56 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!