Open in new window / Try shogun cloud
--- Log opened Tue Jul 30 00:00:53 2013
@iglesiasgtime to get home, see you later!00:08
-!- iglesiasg [~iglesias@2001:6b0:1:1da0:9ce1:9054:335:9d36] has quit [Quit: Ex-Chat]00:08
@sonney2kiglesiasg, I have a crashing test now!00:15
-!- foulwall` [~user@116.114.133.251] has quit [Ping timeout: 264 seconds]00:21
-!- zxtx [~zv@rrcs-76-79-81-162.west.biz.rr.com] has joined #shogun00:25
shogun-buildbotbuild #1114 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1114  blamelist: Evgeniy Andreev <gsomix@gmail.com>, Soeren Sonnenburg <sonne@debian.org>00:28
-!- travis-ci [~travis-ci@ec2-72-44-42-216.compute-1.amazonaws.com] has joined #shogun00:34
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/962200500:34
-!- travis-ci [~travis-ci@ec2-72-44-42-216.compute-1.amazonaws.com] has left #shogun []00:34
-!- travis-ci [~travis-ci@ec2-54-226-172-147.compute-1.amazonaws.com] has joined #shogun00:52
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/962282200:52
-!- travis-ci [~travis-ci@ec2-54-226-172-147.compute-1.amazonaws.com] has left #shogun []00:52
shogun-buildbotbuild #1426 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1426  blamelist: Evgeniy Andreev <gsomix@gmail.com>, Soeren Sonnenburg <sonne@debian.org>00:59
van51sonney2k: sent a PR01:14
van51sonney2k: I'll leave it running tonight to  compare01:14
van51sonney2k: I' ll also add another PR tomorrow morning about quadratic features in the HashedDocDotFeatures class01:14
-!- van51 [~van51@ppp-94-66-56-238.home.otenet.gr] has left #shogun ["QUIT :Leaving."]01:15
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]02:22
-!- hushell [~hushell@8-92.ptpg.oregonstate.edu] has quit [Quit: WeeChat 0.3.7]02:26
-!- foulwall [~user@110.17.4.161] has joined #shogun03:57
-!- FSCV [~FSCV@65.19.131.247] has joined #shogun04:52
shogun-buildbotbuild #473 of nightly_default is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/47304:52
-!- zxtx [~zv@rrcs-76-79-81-162.west.biz.rr.com] has quit [Ping timeout: 245 seconds]05:09
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has joined #shogun06:13
-!- mode/#shogun [+o iglesiasg] by ChanServ06:13
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun06:24
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun06:24
shogun-notifier-shogun: Fernando Iglesias :develop * 1d9e358 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/1d9e35861a9063586690bf34531ae0079fc07a4606:24
shogun-notifier-shogun: Update LMNN to compute outer products on demand instead of keeping them in memory.06:24
shogun-notifier-shogun: Update its unit tests too.06:24
shogun-notifier-shogun: Fernando Iglesias :develop * 7b7b8a0 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/7b7b8a0e4724c5819158663b372daf62cc5cf61b06:24
shogun-notifier-shogun: Merge pull request #1329 from iglesias/feature/lmnn06:24
shogun-notifier-shogun:06:24
shogun-notifier-shogun: Update LMNN to compute outer products on demand instead of keeping them ...06:24
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Ping timeout: 264 seconds]06:47
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun06:47
-!- travis-ci [~travis-ci@ec2-72-44-42-216.compute-1.amazonaws.com] has joined #shogun06:50
travis-ci[travis-ci] it's Fernando Iglesias'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/963283606:50
-!- travis-ci [~travis-ci@ec2-72-44-42-216.compute-1.amazonaws.com] has left #shogun []06:50
shogun-buildbotbuild #1115 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1115  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>06:54
@iglesiasghi gsomix!06:56
gsomixiglesiasg, hi06:57
gsomixand good morning06:57
@iglesiasggsomix: yeah good morning, sorry :)06:57
@iglesiasggsomix: I think cygwin has some trouble with the quotes in CSVFile, any idea?06:57
@iglesiasggsomix: http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/111506:58
gsomixiglesiasg, yep. fixing.06:58
@iglesiasggsomix: cool, thank you!06:59
@iglesiasglet's see if we can get cygwin happy (looks like an impossible task)06:59
shogun-buildbotbuild #1427 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1427  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>07:09
gsomixcompiling... =____=07:11
shogun-buildbotbuild #1116 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1116  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>07:22
gsomixiglesiasg, hey07:40
gsomixiglesiasg, can you merge it? added new unit-test, now all things works right. https://github.com/shogun-toolbox/shogun/pull/133107:41
@iglesiasggsomix: nice!07:42
@iglesiasggsomix: are you in a rush to get it merged? Otherwise, let's wait until travis build is done07:42
gsomixiglesiasg, ok, let's give him a job. :)07:44
shogun-buildbotbuild #1428 of deb3 - modular_interfaces is complete: Failure [failed compile octave_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1428  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>07:44
@iglesiasggsomix: it is looking nice indeed, I just made a very minor comment07:45
@iglesiasggsomix: mmm weird https://travis-ci.org/shogun-toolbox/shogun/jobs/963444207:47
@iglesiasgI thought c++11 was already enabled in travis07:47
gsomixiglesiasg, aww, ok07:48
gsomixI need check one thing07:48
@iglesiasggsomix: maybe it was enabled only with clang07:48
@iglesiasggsomix: yep, it seems to be that since: Checking for C++11 Atomic Integer ... no07:49
@iglesiasggsomix: what compiler do you use locally? If it is g++, did you have to enable C++11 yourself somehow?07:49
gsomixiglesiasg, g++, nope. I think I can avoid #include <cinttypes>.07:50
@iglesiasggsomix: all right07:51
@iglesiasggsomix: otherwise, maybe there is some #ifdef something we can use07:51
@iglesiasggsomix: so if Shogun is not compiled with C++11 support, at least we don't get the crash07:51
gsomixiglesiasg, do you know why SHOGUN needs to recompile all after commit/rebase/push?07:51
@iglesiasggsomix: after rebase yes if any header gets updated07:52
@iglesiasggsomix: but after commit/push I don't know07:52
gsomixany header? strange. it wants recompile all headers after rebase. :\07:53
gsomixok07:53
gsomixneed time to compile07:53
* gsomix went for tea.07:54
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun07:55
@iglesiasggsomix: yes I mean, in the moment you change a single header (note that this includes classes with templates), everything is re-compiled07:57
gsomixlisitsyn, sup, nigga?07:57
lisitsyngsomix: yes08:00
lisitsynor okay :D08:00
lisitsynI don't know what to answer08:00
@iglesiasghaha08:00
lisitsynsonney2k: how can we send an evaluation for georg?08:00
lisitsyniglesiasg: apparently I am your mentor too hahaha08:01
@iglesiasglisitsyn: he is not able to get internet connection at the end?08:01
lisitsynI am terrible08:01
lisitsyniglesiasg: yes he is off08:01
@iglesiasglisitsyn: let's hope there is no problem to make it on his behalf then08:01
lisitsyniglesiasg: yeah we will resolve that for sure08:02
lisitsyniglesiasg: I submitted my evaluation for your project08:04
lisitsynand it looks like 'submitted'08:04
@iglesiasglisitsyn: cool, thank you so much then!08:04
lisitsynso I think that can be ok even with one mentor08:04
lisitsynwell it was mentioned only 'main' mentor is required08:04
lisitsynI am not the main guy but whatever08:05
lisitsyn:D08:05
@iglesiasglisitsyn: so you did through your melange, right?08:05
lisitsyniglesiasg: furthermore that decides whether I can go to SF with sonney2k and HeikoS08:05
lisitsyn:D08:05
lisitsyniglesiasg: yes08:05
@iglesiasglisitsyn: hehe yeah now you can get your flight there!08:05
@iglesiasglisitsyn: I guess it should be fine too. Have you seen any similar query in the mentors mailing list? Do you think we should ask just in case?08:06
lisitsyniglesiasg: ahh it seems there is only one08:06
lisitsynI mean no matter how many mentors you have08:06
lisitsynhaha that's crazy because they ask some things like 'have you been a gsoc student'08:07
gsomixiglesiasg, ok, next try08:08
lisitsyniglesiasg: to increase madness I can submit evaluation for any student08:08
lisitsyn:D08:08
@iglesiasghaha08:08
@wikingmornin'08:17
gsomixiglesiasg, green :)08:20
gsomixwiking, yo08:21
@iglesiasgwiking: hey!08:24
@iglesiasglisitsyn, wiking: did you guys read from yesterday's logs about the doc sprint?08:24
lisitsyniglesiasg: I didn't08:25
@iglesiasglisitsyn: http://shogun-toolbox.org/page/contact/irclog/2013-07-29/ from about 22:1908:26
@wikingagain doc camp?08:29
@wikinglast year we didn't get in08:29
@wiking:08:29
@wiking:P08:29
lisitsyniglesiasg: I think we should try to come up with concrete idea08:29
lisitsynwiking: ^08:29
lisitsynlike 'killing hamsters with SVM'08:29
@iglesiasgeeeh all right :D08:29
@wiking:>08:29
lisitsynI guess we can write only about SVMs08:29
@wikingnow that we have GP ? :)08:30
lisitsynahhhh GP08:30
@wikingcomeon :>08:30
lisitsynbut I am clueless idiot08:30
@wikinganywho08:30
@wikinganyhow08:30
lisitsynno clue about GP at all08:30
@wikingi can hop on the blue train of doc camp08:30
@wikingto have 4 ppl08:30
lisitsynwe have 4 I think08:30
@wikingsince i have time now :P08:30
lisitsynwiking: haha yeah08:30
@wikinglisitsyn: i saw from the log: sonney2k heiko and iglesiasg08:30
lisitsynyes and me08:31
@wikingso they needed +108:31
@wikingah ok08:31
lisitsynwiking: 5 is better08:31
lisitsyn?08:31
lisitsyn;)08:31
@wikingthen u dont need me :)))08:31
@wikingbut yeah i mean no worries08:31
@wikingi can come08:31
@wikingand happy to work on it08:31
@wikingbuuuut08:31
@wikingcmake has prio08:31
lisitsynwiking: well I can have any kind of trouble08:31
@wikingso now i'll make a coffee08:31
@wikingand try to finish that today08:31
@wikingand then do a rebase08:31
@wikingand see what happens08:31
@iglesiasgwiking: full cmake by today?08:31
@iglesiasgthat sounds like a dream!08:32
@wikingiglesiasg: well it's really fucking working08:32
@iglesiasgwow, awesome08:32
@wikingso i dont know why u r surprised08:32
@iglesiasg:D08:32
@wikingjust really need to finish with the rest of modulars08:32
lisitsynwiking: argh sorry I postpone it all the time :D08:32
@wikingalthough i havent even touched static08:32
@wikingi guess that's needed for matlab as well08:32
@iglesiasg all right, looking forward to it then08:32
@iglesiasgindeed08:32
lisitsynwiking: well we can drop static hahahh08:33
lisitsynthat's my dream08:33
lisitsynI have a dream08:33
@wikinganyhow i want to make first a nice and smooth ccmake menu08:33
@wikingso now i'll work on that first08:33
@wikingand then after that i'll see about the interfaces08:33
lisitsyngood there is no cccmake08:33
lisitsynand ccccmake08:33
@iglesiasg:)08:33
@wiking:>08:33
@iglesiasglisitsyn: I guess we cannot drop static until we get something else for Matlab...08:34
lisitsyniglesiasg: of coursssssse08:37
lisitsyniglesiasg: wiking: we need something really concrete08:39
lisitsyn'SVMs for terribly big data about some shit'08:39
lisitsyn'how to sort this shit out with SVMs'08:39
lisitsynwhatever08:39
lisitsyn(brand new) (with pictures)08:40
lisitsynwith such a labels from 90s08:40
@iglesiasglisitsyn: maybe something related to MKL since sonney2k is the expert?08:42
lisitsynnooo I think we should go for something less 'smart' iglesiasg08:43
lisitsynlike the thing for hackers you showed08:43
lisitsynMKL is just a research field08:43
@wikingoh woah i didn't know about cpack08:50
gsomixiglesiasg, lisitsyn, wiking  hey guys, can you push magic "MERGE IT AND MAKE ME HAPPY" button on my PR? :)08:51
@iglesiasggsomix: sure08:51
gsomixcool08:52
gsomixthen I can take shower.08:52
shogun-notifier-shogun: Evgeniy Andreev :develop * aaa9597 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/aaa95976211fe3b7a12d5851c204885758838d3d08:52
shogun-notifier-shogun: fixed format for fprintf in CSVFile08:52
shogun-notifier-shogun: Fernando Iglesias :develop * e569cf6 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/e569cf60c5c036a76d3f85c31517bf6d4cce3d8008:52
shogun-notifier-shogun: Merge pull request #1331 from gsomix/feature/parser08:52
shogun-notifier-shogun:08:52
shogun-notifier-shogun: Fixed format for fprintf in CSVFile08:52
lisitsynthe world08:53
lisitsynthe time has come to08:53
lisitsynpush da button08:53
lisitsyngalvanize! tadada08:53
@iglesiasg:)08:53
lisitsyniglesiasg: know that song?08:53
@iglesiasglisitsyn: yeah, sure08:53
lisitsyniglesiasg: I hope they are not swedish08:53
lisitsyn;)08:53
@iglesiasgI think chemical brothers are from the UK08:54
gsomixlisitsyn, hm, now I know what to listen today. :]08:54
@iglesiasglet me check08:54
lisitsyniglesiasg: each time we talk about music it appears the band is from sweden :D08:54
@iglesiasglisitsyn: yeah, it happens a lot :)08:54
gsomixgeneral speaking training08:56
gsomixguys, what music do you prefer? or what do you listened to the last week/month?08:56
@iglesiasggsomix: I am listening these days to the tomorrowland sessions from this weekend08:57
lisitsynI like music from 70s porno movies08:57
@wikingbtw r modular is now fixed?08:57
@iglesiasglisitsyn: haha what kind of music is that? lol08:57
lisitsyniglesiasg: oh it sounds like it is from porno08:58
@iglesiasggsomix: I watched this movie the other day. Think the soundtrack is cool: http://www.youtube.com/watch?v=j97WR_slydY08:58
shogun-buildbotbuild #1609 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1609  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>08:58
@iglesiasggsomix: there is something funny with CSVFileTest.read_matrix unit_test08:59
gsomixiglesiasg, mmm?08:59
@iglesiasggsomix: http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1609/steps/test/logs/stdio09:00
@iglesiasggsomix: it crashed in my machine too, as well as ParserTest.tokenization_reals09:00
gsomixwut? O_O09:00
gsomixiglesiasg, ok, I'll check.09:01
@iglesiasggsomix: I am compiling latest code with your new commit just in case that changed something09:01
gsomixshower is canceled, huh09:02
@iglesiasggsomix: but CSVFileTest.read_matrix crashes in my machine when doing make in tests/unit, not when running it alone though. But with valgrind detects some use of uninitialised memory09:02
@iglesiasggsomix: hehe I guess the fix can wait a bit too :)09:02
-!- van51 [~van51@ppp-94-66-56-238.home.otenet.gr] has joined #shogun09:14
van51good morning09:14
-!- travis-ci [~travis-ci@ec2-54-226-172-147.compute-1.amazonaws.com] has joined #shogun09:17
travis-ci[travis-ci] it's Fernando Iglesias'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/963576009:17
-!- travis-ci [~travis-ci@ec2-54-226-172-147.compute-1.amazonaws.com] has left #shogun []09:17
@iglesiasggsomix: have a look also to that other test please when you get a moment09:23
gsomixiglesiasg, M09:24
gsomix*m&09:24
gsomix*m?09:24
gsomixawww09:24
gsomix:/09:24
@iglesiasggsomix: are you ok? :P09:24
@iglesiasggsomix: https://travis-ci.org/shogun-toolbox/shogun/jobs/963576109:24
gsomixiglesiasg, I hope.09:24
@iglesiasggsomix: SGObject.clone_equals_CSVFile, maybe there is something unitiliased? Just guessing09:25
gsomixiglesiasg, yep. already fixed.09:25
gsomixiglesiasg, btw can you send my yours valgrind log?09:26
gsomix*to me09:26
gsomixvery strange that tests works fine at my machine :\09:28
@iglesiasggsomix: sendind you valgrind trace ...09:40
@iglesiasggsomix: so valgrind does not complain at all in your machine?09:40
gsomixiglesiasg, only at unitiliased values09:41
shogun-buildbotbuild #1117 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1117  blamelist: Evgeniy Andreev <gsomix@gmail.com>09:42
@iglesiasggsomix: http://pastebin.com/z8615XKp09:42
gsomixiglesiasg, thanks. it's fixed.09:46
@iglesiasggsomix: PR then?09:46
@iglesiasggsomix: we will wait this time to see everything is fine with travis :)09:46
gsomixiglesiasg, yeah.09:47
shogun-buildbotbuild #1429 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1429  blamelist: Evgeniy Andreev <gsomix@gmail.com>10:11
gsomixiglesiasg, I figured out where is problem.10:37
@iglesiasggsomix: where is it then?10:37
gsomixiglesiasg, ref/unref sgobjects in my code. I'm doing it totally wrong.10:38
gsomixfixing10:38
@wikingsonney2k: here?10:49
@wikinganybody here using java modular?10:49
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has quit [Quit: Leaving]11:25
-!- thoralf [~thoralf@enki.zib.de] has joined #shogun11:46
thoralfHeyho11:47
-!- gsomix_ [~gsomix@95.67.160.18] has joined #shogun11:50
-!- HeikoS [~heiko@nat-184-201.internal.eduroam.ucl.ac.uk] has joined #shogun11:52
-!- mode/#shogun [+o HeikoS] by ChanServ11:52
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]11:52
-!- gsomix [~gsomix@95.67.137.114] has quit [Ping timeout: 264 seconds]11:54
@wikingHeikoS: yoyo11:57
@HeikoSwiking: hello mister11:57
@HeikoShow are things?11:57
@wikingworking on cmake11:57
@wikingas soon as i have it ready i'll add those automatic tests11:58
@HeikoSwiking: cool!11:58
@HeikoSwiking: what do you think about tests on non-trivial instances?11:58
@HeikoSlike for example DenseFeatures11:58
@HeikoSwhich then has a feature matrix, a subset and a preprocessor11:58
@HeikoShow to implement this best?11:59
@wikingHeikoS: aaah extra tests for those?11:59
@wikingah no idea yet11:59
@wiking:)11:59
@HeikoSthoralf had the idea of automagically matching types and stuff11:59
@wikingi'm still trying to get ruby_modular working :)))11:59
@HeikoSI had the idea to just add a generate_std_instance() to every class11:59
@wikingi have a SGBase.i:322: Error: Unable to find 'std_vector.i'11:59
@HeikoSwiking: ah I see11:59
@HeikoSyeah, so lets do this later :)11:59
thoralfHey HeikoS11:59
@HeikoSthoralf: hey11:59
@HeikoSthoralf: lets discuss with wiking at some point all our ideas12:00
@HeikoSwiking: but the automagic tests already revealed 100s of bugs12:00
@HeikoSvery useful12:00
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has joined #shogun12:00
-!- mode/#shogun [+o iglesiasg] by ChanServ12:00
@HeikoSif we have non-trivial tests, we can finally rely on serialisation etc12:00
thoralfHeikoS: Of course.  Let's make an appointment.12:00
@wikingHeikoS: what would u like to have in non-trivial tests again?12:00
@iglesiasghello hello12:00
@HeikoSwiking: currenlty, we do the automated tests on classes that have been generated with std constructors12:01
@HeikoSwiking: which means that everything is NULL, 0, -1 or similar12:01
@HeikoSwiking: but it would be good to also have concrete data in them12:02
@HeikoSand then test equals/clone/save12:02
@HeikoSwill detect more problems12:02
@HeikoSand makes all these methods very stable12:02
@wikingHeikoS: well then we have to use the non default ctors ;)12:02
@HeikoSwiking: but how to do that for all classes automatically?12:02
@wikingHeikoS: aaall classes?12:03
@HeikoSwiking: there are also sometimes constraints on what data is allowed12:03
@wikingHeikoS: problem is that the non std ctor varies a lot12:03
@HeikoSwiking: so I had the idea to force people to implement a method that creates at least one non-trivial instance12:03
@HeikoSand then we can use that recursively to create tests12:03
@iglesiasggsomix_: ping12:04
gsomix_iglesiasg, pong12:04
@wikingHeikoS: mmmm lemme think about this a bit12:04
@HeikoSwiking: ok12:04
@iglesiasggsomix_: news about CSVFile tests?12:05
@HeikoSwiking: CSGObject* CSGObject::create_non_trivial_instance()=0;12:05
@HeikoSwiking: which is implemented by the person who creates the class12:05
@HeikoSsince he knows what to do12:05
gsomix_iglesiasg, working. have many issues with ref/unref. :(12:06
thoralfHeikoS: Ah, you're discussing right now.12:06
@HeikoSwiking: thorals suggested that this should not be in the class definition itself but somewhere else to keep code clean from tests12:06
@HeikoSthoralf: no, just giving wiking the ideas we had12:06
@iglesiasggsomix_: all right, get them tiger! :P12:07
thoralfHeikoS: What about using github-wiki for this?12:07
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun12:07
shogun-notifier-shogun: Roman Votyakov :develop * aabab97 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/aabab97ae6b80860266ea4d1792d3625a4838b7812:07
shogun-notifier-shogun: add numerical integration stuff12:07
shogun-notifier-shogun: Heiko Strathmann :develop * 4db24b7 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/4db24b7d88fafab5b827581e81f60f9d2a65254712:07
shogun-notifier-shogun: Merge pull request #1324 from votjakovr/feature/numerical_integration12:07
shogun-notifier-shogun:12:07
shogun-notifier-shogun: Add numerical integration stuff12:07
@HeikoSthoralf: for what?12:08
thoralfHeikoS: We could write a sketch of our ideas.  Maybe i helps us to overlook drawbacks of every idea.12:08
thoralfHeikoS: For documenting the requirements and our ideas.12:08
@HeikoSthoralf: yeah lets start an issue, there is a label on general things to discuss12:08
shogun-buildbotbuild #1610 of deb1 - libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/161012:17
gsomix_HeikoS, iglesiasg hey, can you help me? when I create SGObject by using new, should I do SG_REF()?12:19
@iglesiasggsomix_: no, there is no need12:20
thoralfHeikoS: Let's go https://github.com/shogun-toolbox/shogun/issues/133212:20
@iglesiasggsomix_: it shouldn't hurt either. You will have to do SG_UNREF afterwards anyway12:21
@HeikoSgsomix_: what exactly is the problem?12:25
@HeikoSgsomix_: you should SG_REF it only if you want to keep the reference and use it later12:25
@HeikoSgsomix_: for example when you store it as a class member12:25
gsomix_HeikoS, yeah. it's all that I need.12:26
@HeikoSgsomix_: then REF when you create and unref in destructor12:26
@HeikoSgsomix_: also unref potential old objects12:26
@HeikoSbefore you create new ones12:26
@HeikoSat the same pointer12:26
gsomix_aha12:26
gsomix_HeikoS, iglesiasg thanks!12:27
shogun-buildbotbuild #1118 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1118  blamelist: Roman Votyakov <votjakovr@gmail.com>12:27
-!- travis-ci [~travis-ci@ec2-174-129-100-76.compute-1.amazonaws.com] has joined #shogun12:33
travis-ci[travis-ci] it's Heiko Strathmann'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/964140612:33
-!- travis-ci [~travis-ci@ec2-174-129-100-76.compute-1.amazonaws.com] has left #shogun []12:33
* gsomix_ away12:41
-!- FSCV [~FSCV@65.19.131.247] has quit [Quit: This computer has gone to sleep]12:54
shogun-buildbotbuild #1119 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1119  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>12:57
shogun-buildbotbuild #1430 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1430  blamelist: Roman Votyakov <votjakovr@gmail.com>13:02
-!- van51 [~van51@ppp-94-66-56-238.home.otenet.gr] has quit [Quit: Leaving.]13:02
@sonney2kHeikoS, wiking hey there13:11
@HeikoSsonney2k: hi!13:12
@sonney2kHeikoS, so what about doc camp?13:17
@HeikoSsonney2k: hey13:17
@HeikoSI checked a bit what they do there13:17
@sonney2kHeikoS, you, iglesias & me and wiking? lisitsyn?13:17
@HeikoSI am not sure whether its applicable for us13:17
@sonney2kHeikoS, they write a book :)13:18
@sonney2kHeikoS, ?13:18
@HeikoSsonney2k: its larger projects, large groups of users etc13:18
@HeikoSsonney2k: so not sure13:18
@HeikoSwould be cool though13:18
@sonney2kHeikoS, yes sure large as in KDE big13:18
@sonney2ks/big/size13:18
@HeikoSsonney2k: yep13:18
@sonney2kbut apart from that it fits13:18
@HeikoSsonney2k: and what would we write about?13:19
@sonney2kHeikoS, maybe the big data hype could pay off13:19
@HeikoSsonney2k: haha yeah maybe13:19
@HeikoSok then13:19
@sonney2kHeikoS, everything we have in shogun13:19
@HeikoSwe can recycle last years application13:19
@sonney2kas in kernels, distances, kernel / distance machines13:19
@sonney2ketc etc13:19
@sonney2kit is doable IMHO13:19
@HeikoSsonney2k: I would rather not focus on the algorithms I think13:20
@HeikoShaving seen what was done last year13:20
@HeikoSmore like how development works, how to install, how to use, some basic algorightms ...13:21
@HeikoSnot too mathematical13:21
@HeikoSand not a list of things13:21
@HeikoSbut more an introduction to the concepts13:21
@HeikoSwiking: how to use the new random number generator?13:24
@HeikoSwiking: where is the instance ?13:24
shogun-buildbotbuild #1431 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1431  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>13:32
@sonney2kHeikoS, btw please send your evaluations today you are the last one missing :)13:33
@sonney2kHeikoS, no problem with that13:33
@HeikoSsonney2k: will do13:34
@HeikoSsonney2k: I gotta go now, see you later today13:34
@wikingHeikoS: sg_rand13:35
@wikingHeikoS: it's a global13:36
@wikingsonney2k: yo13:36
@wikingsonney2k: u used java_modular right?13:36
@wikingsonney2k: what was the jnilib name ?13:37
@sonney2kwiking, look at lambday's blog - it is on shogun planet13:37
@sonney2khe wrote some howto13:37
@wikingoh libmodshogun.so13:38
@wikingwoah13:38
@wikingi'll have to fix this one13:38
@wiking;)13:38
-!- gsomix_ is now known as gsomix13:38
* gsomix here13:38
gsomixsonney2k, hello13:40
@sonney2kwiking, ?13:43
@sonney2kgsomix, hey13:43
gsomixsonney2k, fixing bugs whole day. :\13:45
gsomixsonney2k, I found strange behavior of fgets in unit-test.13:45
gsomixoooor it's strange behavior of my test13:48
gsomix=___=13:48
-!- HeikoS [~heiko@nat-184-201.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]13:49
@wikingsonney2k: cmake created the following lib for me: libjava_modular.jnilib13:53
@sonney2kwiking, ahh ok that doesn't work then :)13:56
@wikingsonney2k: but it should be ok i just need to rename it to libmodshogun.so during installing13:58
@wikingand that should be find13:58
@wiking*fine13:58
@sonney2kwiking, ahh ok14:03
@wikingok so lua, ruby, python and java modular are compiled fine with cmake14:05
@wikingso now what we are actually missing is csharp, r, octave and matlab_static14:05
@sonney2kwiking, piece of cake then when you've come that far14:08
@wikingwell yeh14:09
@wikingalthough i should do some more checks14:09
@wikingand then i should still work on the compiler flags14:09
@wikingas i really haven't even touched that part14:09
@sonney2kwiking, yeah well it is some hefty project14:10
@sonney2kit might take another (few) months to complete14:11
@wikingnoway14:12
lisitsynnahh compiler flags is easy part14:12
@wikingsonney2k: http://www.cmake.org/Wiki/CTest:Submission_Issues14:40
@wikingsonney2k: we could collect gtest outputs from users14:45
@wiking;)14:45
@wikingautomagically14:45
shogun-notifier-shogun: Soeren Sonnenburg :develop * 8588515 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/85885154699dd618d961b566b389845e95ca1d7f14:54
shogun-notifier-shogun: add SGMatrixList more memory unittests and fix SG_MALLOC(SGMatrix...) issue14:54
@sonney2kwiking, errm lets say buildbots are sufficient :D14:55
@sonney2kiglesiasg, ^ that should do it!14:55
@sonney2kblue time!14:55
@wikingcsharp modular done...15:09
shogun-buildbotbuild #1120 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1120  blamelist: Soeren Sonnenburg <sonne@debian.org>15:19
-!- travis-ci [~travis-ci@ec2-174-129-100-76.compute-1.amazonaws.com] has joined #shogun15:28
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/964652315:28
-!- travis-ci [~travis-ci@ec2-174-129-100-76.compute-1.amazonaws.com] has left #shogun []15:28
@iglesiasgsonney2k: great, thank you!15:29
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has quit [Quit: Leaving]15:36
shogun-notifier-shogun: Viktor Gal :feature/CMake * 54232cc / / (17 files): https://github.com/shogun-toolbox/shogun/commit/54232cc1ec2c429c1d7efb8e983aa0efdaffaf9015:41
shogun-notifier-shogun: add lua, csharp, java, octave modular support15:41
@wikinglets see what travis does with this15:43
@wikingalthough example and test suit is still missing from cmake15:44
@wikinglisitsyn: transfer_multitask_clustered_logistic_regression.py15:46
@wikingdo u know y no work? :)15:46
-!- FSCV [~FSCV@108.60.131.13] has joined #shogun15:52
shogun-buildbotbuild #1432 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1432  blamelist: Soeren Sonnenburg <sonne@debian.org>15:58
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun16:18
-!- travis-ci [~travis-ci@ec2-23-22-48-223.compute-1.amazonaws.com] has joined #shogun16:23
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/964847116:23
-!- travis-ci [~travis-ci@ec2-23-22-48-223.compute-1.amazonaws.com] has left #shogun []16:23
shogun-notifier-shogun: Viktor Gal :feature/CMake * 0f30d1f / CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/0f30d1f42197279742feb7e3c575d52e23bd1bb316:26
shogun-notifier-shogun: Restructure interface choosing in cmake16:26
@wikingcan anybody tell me what are the required packages on deb/ubuntu for octave modular?16:42
@wiking(i dont have an ubuntu/deb distro at hand atm)16:42
thoralfwiking: Beside octave I only installed octave_dev (afaik)16:50
@wikingthoralf: on ubuntu?16:51
thoralfwiking: Yes.16:51
@wikingwhat's the exact name of that package ;)16:51
thoralfwiking: octave_dev16:51
thoralf:)16:51
@wikingliboctave-dev16:51
@wikingthere's no such pkg on ubuntu according to packages.ubuntu.com16:51
thoralfOh, yes.16:51
thoralfYou're right.16:52
@wikinggsomix: your latest pr crashes circbuffer example ;)16:52
@wikinggsomix: library_circularbuffer16:53
@wikinggsomix: https://travis-ci.org/shogun-toolbox/shogun/jobs/965000916:53
-!- naywhayare [~ryan@spoon.lugatgt.org] has quit [Quit: leaving]17:01
@wikingsonney2k: http://www.cmake.org/Wiki/CMake:Packaging_With_CPack17:07
@wikingi think we want this17:07
gsomixwiking, :(17:10
pickle27lisitsyn: hey!17:18
lisitsynpickle27: hey there17:18
@wikinglisitsyn: did u know about this http://www.cmake.org/Wiki/CMake:Packaging_With_CPack17:18
pickle27lisitsyn: do you have a second? For the time being I'm going to pass on a R_modular example but I still want a R_static one17:19
lisitsynpickle27: yes sure17:19
pickle27lisitsyn: my 0ctave_static works but I get an error on my r_static I was hoping you could give me a hand17:19
lisitsynwiking: no, what's it?17:19
lisitsynwiking: I see it packages something17:20
lisitsyn:D17:20
lisitsynto what packages?17:20
-!- van51 [~van51@ppp-94-66-56-238.home.otenet.gr] has joined #shogun17:20
@wikinglisitsyn: well as we talked that i'd be great to have packages of the nightly builts (tar.gz/bz2)17:20
@wikingthis would be the best thing to use17:20
@wikingand as i can see it has debian package suppport as well17:21
@wikingand many other crazy package types17:21
@wikinghttp://www.cmake.org/Wiki/CMake:CPackPackageGenerators17:21
@wikinglike rpm: http://www.cmake.org/Wiki/CMake:CPackPackageGenerators#RPM_.28Unix_Only.2917:21
pickle27lisitsyn: this is the error I get17:22
pickle27Error in sg("apply_converter") :17:22
pickle27  [ERROR] assertion m_lhs_counter>=0 && m_lhs_counter<m_nlhs failed in void CRInterface::set_arg_increment(SEXP) file RInterface.h line 16617:22
@wikinglisitsyn: so once we have cmake ready we can right away generate various packages as well17:22
pickle27wiking: Im excited for cmake, do we have an eta?17:23
@wikingpickle27: mmmm today...? :)17:23
pickle27wiking: no way! amazing!17:23
@wikingalthough i have to do a rebase with the develop branch so that'll be some extra17:23
@wikingmaybe the latest tomorrow17:23
lisitsynwiking: yeah cool thing17:24
pickle27wiking: I want to start messing around with making a Go wrapper but I was waiting for cmake17:24
lisitsynpickle27: hmm not sure what it is17:24
lisitsynsome ehm counter17:25
pickle27lisitsyn: maybe I forgot to add something in my apply converter code17:26
shogun-notifier-shogun: Viktor Gal :feature/CMake * a152427 / .travis.yml,CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/a1524270311894eef824136c31b571ea4e6be9ae17:29
shogun-notifier-shogun: add missing endif in cmake17:29
shogun-notifier-shogun: add preliminary support for cpack17:29
shogun-notifier-shogun: add octave target to travis builds17:29
-!- travis-ci [~travis-ci@ec2-23-22-48-223.compute-1.amazonaws.com] has joined #shogun17:30
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/965015417:30
-!- travis-ci [~travis-ci@ec2-23-22-48-223.compute-1.amazonaws.com] has left #shogun []17:30
pickle27lisitsyn: what is the counter supposed to do?17:35
lisitsynpickle27: IDK :D17:39
pickle27lisitsyn: okay I'll dig into it!17:39
-!- HeikoS [~heiko@nat-164-120.internal.eduroam.ucl.ac.uk] has joined #shogun17:41
-!- mode/#shogun [+o HeikoS] by ChanServ17:41
shogun-notifier-shogun: Roman Votyakov :develop * bdd0f58 / src/ (3 files): https://github.com/shogun-toolbox/shogun/commit/bdd0f5835a029ed1a5bf39549ea01ccd272e3b1f17:48
shogun-notifier-shogun: fix documentation for numerical integration stuff17:48
shogun-notifier-shogun: Heiko Strathmann :develop * 963ddc9 / src/ (3 files): https://github.com/shogun-toolbox/shogun/commit/963ddc9d9b8590777fb56f5b75da3ba3986c200f17:48
shogun-notifier-shogun: Merge pull request #1333 from votjakovr/feature/numerical_integration17:48
shogun-notifier-shogun:17:48
shogun-notifier-shogun: Fix documentation for numerical integration stuff17:48
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun18:00
shogun-buildbotbuild #1121 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1121  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>18:02
pickle27lisitsyn: fixed it!18:28
pickle27now that I found the bug I'm not sure how octave worked...18:28
-!- foulwall [~user@110.17.4.161] has quit [Ping timeout: 245 seconds]18:30
shogun-buildbotbuild #1433 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1433  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>18:38
@HeikoSlisitsyn: around?18:40
@HeikoSI need c++ help :)18:40
lisitsynpickle27: cool!18:40
lisitsynHeikoS: yes18:40
@HeikoSor anyone else?18:40
@wikingtell18:40
@HeikoSlisitsyn: ok, I warn you, basic stuff ;)18:41
@HeikoSso I have a base class and a class that inherits from the base18:41
@HeikoSthis call works18:41
@HeikoSfloat64_t result=((CProbabilityDistribution*)gauss)->log_pdf(x);18:41
@HeikoSand this call gives me a compile error18:41
@HeikoSfloat64_t result=gauss->log_pdf(x);18:41
@HeikoSwhy?18:41
lisitsynwhat is the type of gauss?18:41
pickle27lisitsyn: needed to set the number of return arguments18:41
lisitsynpickle27: ahh18:42
@HeikoSgauss is of type CGaussian which inherits from CProbabiltyD...18:42
lisitsynHeikoS: then it should work :D18:42
pickle27lisitsyn: I believe the same fix needs to be applied to the embed method18:42
@HeikoSI know, but it doesnt, what are the caveats one can do while inheriting things?18:42
lisitsynpickle27: oh please do it then18:42
@HeikoSI checked all I know but  still ...18:43
lisitsynHeikoS: log_pdf is virtual, righT?18:43
@HeikoSlisitsyn: yes18:43
@HeikoSvirtual float64_t log_pdf(SGVector<float64_t> single_sample) const;18:43
pickle27lisitsyn: I'm just going to port an example of using the embed converter to one of the static interfaces as we currently don't have any18:43
gsomixmaybe dynamic_cast is needed?18:43
@HeikoSgsomix: would work, since the static one works18:44
@HeikoSbut why does the compiler complain?18:44
@HeikoSerror: no matching function for call to 'shogun::CGaussianDistribution::log_pdf(shogun::SGVector<double>&)'18:45
@HeikoSthats the error18:45
@HeikoSfloat64_t result=gauss->log_pdf(x);18:45
@HeikoSand thats the call18:45
lisitsynHeikoS: no I see no reason actually18:45
@HeikoSwhere x is SGVector18:45
@HeikoSinheritance is public18:45
@HeikoSstatic cast works, thats the weirdest18:45
@HeikoSand the method is virtual18:46
lisitsynHeikoS: so if you do this with static cast18:46
lisitsynyou get no error?18:46
@HeikoSyes18:46
@HeikoS(it also runs)18:46
shogun-buildbotbuild #1122 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1122  blamelist: Roman Votyakov <votjakovr@gmail.com>18:46
lisitsynHeikoS: the only caveat I can imagine is different declarations of that method18:47
@HeikoSit is only defined in base class18:48
@HeikoSnowhere else18:48
lisitsynHeikoS: is it public'ly inherited?18:52
lisitsynlike18:52
lisitsynclass CGaussian: public CProbabilityDistribution?18:52
@HeikoSclass CGaussianDistribution: public CProbabilityDistribution18:52
@HeikoSyes18:52
lisitsynHeikoS: is it implemented in CGaussianDistribution?18:52
@HeikoSno18:53
@HeikoSin base class18:53
gsomixis method also public? just for check.18:56
@HeikoSgsomix:  yes18:56
votjakovrHeikoS: hi! The difference between our implementation of logit likelihood and GPML is less than 1e-3. Is it Ok?19:01
@HeikoSvotjakovr: uh, thats not so precise19:01
@HeikoSvotjakovr: can you test a more complex case?19:01
@HeikoSif it stays that low, its ok19:01
@HeikoSbut better check19:02
@HeikoSso yes for now, but please check a bit after you sent the PR19:02
votjakovrHeikoS: we have more accurate integral evaluation than GPML has19:02
@HeikoSvotjakovr: I know19:03
votjakovrHeikoS: absolute tolerance is something about 1e-1019:03
lisitsynHeikoS: I have no ideas left19:03
@HeikoSvotjakovr: I think it is fine, but pls have a look at an example with a bit more data19:03
@HeikoSlisitsyn: weird stuff ....19:03
lisitsyn:/19:03
lisitsynHeikoS: yeah19:03
lisitsynHeikoS: but it would make sense to carefully check definitions19:03
@HeikoSlisitsyn: did 10 times already, maybe tomorrow again19:04
lisitsynHeikoS: then something else yeah19:04
lisitsynI guess you'd notice something19:04
-!- travis-ci [~travis-ci@ec2-23-22-48-223.compute-1.amazonaws.com] has joined #shogun19:06
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/965329519:06
-!- travis-ci [~travis-ci@ec2-23-22-48-223.compute-1.amazonaws.com] has left #shogun []19:06
shogun-buildbotbuild #1434 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1434  blamelist: Roman Votyakov <votjakovr@gmail.com>19:08
pickle27lisitsyn: yeah we need it for embedding too, I didn't make an example but embed simply wraps apply19:08
pickle27so its pretty clear we need it19:08
pickle27I can make a PR after my current one is merged19:09
-!- vgorbati [c3ee5cb1@gateway/web/freenode/ip.195.238.92.177] has quit [Quit: Page closed]19:17
votjakovrHeikoS: i've just sent a PR, if you have a time, please look at it19:18
shogun-notifier-shogun: Roman Votyakov :develop * e5365e6 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/e5365e63d185ad49f4ad26ae87d7d6fff2e687dd19:21
shogun-notifier-shogun: implemented evaluation of predictions for logit likelihood19:21
shogun-notifier-shogun: Heiko Strathmann :develop * f0188bb / / (2 files): https://github.com/shogun-toolbox/shogun/commit/f0188bb05b86aa610b9791260d9c64d4c7d4343519:21
shogun-notifier-shogun: Merge pull request #1335 from votjakovr/feature/gp_binary_classification19:21
shogun-notifier-shogun:19:21
shogun-notifier-shogun: Implemented evaluation of predictions for logit likelihood19:21
@HeikoSvotjakovr: so the logit classification stuff works now? :)19:22
votjakovrHeikoS: yep, but i think it needs an example :)19:22
@HeikoSvotjakovr: very nice!19:22
@HeikoSvotjakovr: could you write a simple and basic example for GP classification in c++ and in python_modular?19:23
@HeikoSvotjakovr: just very very basic api usage and how to obtain predictions/confidences19:23
@HeikoSvotjakovr: to make it easy for users to play with that19:23
@HeikoSvotjakovr: big example will follow19:23
@HeikoSvotjakovr: maybe in the example just use both likelihood we currently have and laplacian, but comment out all but one19:24
@HeikoSvotjakovr: should be quickly be done since the stuff is already tested19:25
votjakovrHeikoS: sure, i'll do that :)19:25
votjakovrHeikoS: btw do GPs work well with python_modular now? I mean there was some problems19:27
@HeikoSvotjakovr: I fixed that :)19:27
@HeikoSvotjakovr: there is a new GaussianProcess.i file now, check it out19:28
votjakovrHeikoS: oh, thanks!19:28
@HeikoSeverything is added in there19:28
@HeikoSvotjakovr: the example can really be simple api demonstration ones, no need for fancy ML in there19:28
@HeikoSmaybe a few comments on what is possible19:29
votjakovrHeikoS: btw shouldn't we again report to Carl about the problems?19:33
@HeikoSvotjakovr: yep I agree, I think this time we are right since it seems to be an octave bug ;)19:34
@HeikoSvotjakovr: could you send me your fix diff?19:34
votjakovrHeikoS: hmm, i've just replaced 32th line with: s2zero = 1; if nargin>3, if length(s2)>0, s2zero = 0; end, end         % s2==0 ?19:36
@HeikoSyep that is fine19:37
votjakovrHeikoS: in likT.m19:37
shogun-buildbotbuild #1123 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1123  blamelist: Roman Votyakov <votjakovr@gmail.com>19:37
-!- lambday [67157e4f@gateway/web/freenode/ip.103.21.126.79] has joined #shogun19:39
lambdayHeikoS: hi19:39
@HeikoSlambday:  hi!19:40
lambdayHeikoS: this week is really being tough on me :'(19:40
@HeikoSlambday: nice work with the wrapper!19:40
lambdayHeikoS: I am having many many instu related stuffs to be taken care of :'(19:40
lambdayHeikoS: thanks :(19:40
lambdayinsti*19:40
@HeikoSlambday: dont worry too much, schedule is looking very good19:41
lambdayHeikoS: is it okay if I start working a bit late, say, thursday or friday this week?19:41
lambdayHeikoS: my prof has threaten me to strip off my stipend if I don't work on his project :(19:41
@HeikoSlambday: yes thats fine, please do that19:42
@HeikoSlambday: it would be very bad if your stipend got canceled19:42
@HeikoSyour degree is more important that GSoC19:42
lambdayHeikoS: since Lapack works, I don't it will take me more than a few days to add Lanczos stuffs19:42
@HeikoSand also you probably can do this stuff easily19:42
lambdayHeikoS: yes19:43
@HeikoSlambday: yeah, I think that will go fast19:43
@HeikoSlambday: and then the only thing thats missing is the probing sampler19:43
lambdayHeikoS: thanks man! :D19:43
@HeikoSthats a lot of work, but we also got a lot of time19:43
@HeikoSlambday: dont worry, you have been doing quite a lot of stuff already, so relax and make your prof happy :)19:43
lambdayHeikoS: probably I can use that PR's code that iglesias pointed out19:44
@HeikoSlambday: maybe, but I am not sure19:44
@HeikoSlambday: it might be a good idea to also check out some external libs, Ill also have a look19:44
lambdaybut I am not sure.. so I'll finish other stuffs by this week and think about next step from next week onwards19:44
lambdayHeikoS: that would be great! they use colpack, may be we can check out that too...19:45
@HeikoSlambday: but they compute the colouring in a not so efficient way by multiplying matrices with itself19:46
lambdayHeikoS: recently eigen3 people had some PRs regarding matrix powers... I am not too sure but wondering if we can take help from that19:46
-!- naywhayare [~ryan@spoon.lugatgt.org] has joined #shogun19:47
@HeikoSlambday: cool, that might be a good idea19:47
-!- travis-ci [~travis-ci@ec2-23-22-48-223.compute-1.amazonaws.com] has joined #shogun19:47
travis-ci[travis-ci] it's Heiko Strathmann'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/965407119:47
-!- travis-ci [~travis-ci@ec2-23-22-48-223.compute-1.amazonaws.com] has left #shogun []19:47
@HeikoSlisitsyn: python modular fails19:50
@HeikoSlisitsyn: its one of yours I think, transfer_multitask_clustered_logistic_regression19:50
lambdayHeikoS: today my prof gave me a long lecture about how I am not doing anything significant for the lab while I've working for GSoC and actually doing something! I am not sure how I should take this :D he seems quite pissed off :(19:51
lambdayHeikoS: oh and regarding the computation framework, there is a part in one of my lab's project where there are parallel computation of individual computable components are required19:52
lisitsynHeikoS: yeah I know sorry19:52
lambdayHeikoS: I said, that I probably can give something in that.. they exactly need to the openmpi implementation stuffs..19:52
lambdayHeikoS: may be after GSoC we can discuss?19:53
@HeikoSlambday: yeah19:53
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.]19:53
@HeikoSlambday: I would be very exxcited to do some mpi stuff19:53
@HeikoSbut thats all later19:53
@HeikoSfor now, make your prof happy and finish log-dets :)19:53
@HeikoSbut very cool stuff19:54
lambdayHeikoS: yes... me too19:54
@HeikoSI in fact had a meeting today, talking about large-scale implementations of the log-determinants19:54
lambdayHeikoS: seems tha computation framework is going to be used by many many people! :D that makes me really happy :D19:54
@HeikoSand we will probably here build a large system for that...19:54
@HeikoSlambday: yeah totally19:54
lambdayHeikoS: ahan! :D19:55
@HeikoSah shit, I have to go now, theatre19:55
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 245 seconds]19:55
@HeikoSI am going to be late ;)19:55
@HeikoSsee you!19:55
lambdayHeikoS: no problem man! I too have to go because I gotta prepare for a presentation for tomorrow :(19:55
lambdayHeikoS: have a nice time! see you :)19:55
@HeikoSlambday: good luck with that!19:55
@HeikoSsee you :)19:55
lambdaybye :)19:55
-!- HeikoS [~heiko@nat-164-120.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]19:56
-!- lambday [67157e4f@gateway/web/freenode/ip.103.21.126.79] has quit []19:57
shogun-buildbotbuild #1435 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1435  blamelist: Roman Votyakov <votjakovr@gmail.com>20:08
shogun-buildbotbuild #1124 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1124  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>20:14
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)]20:16
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has joined #shogun20:19
-!- mode/#shogun [+o iglesiasg] by ChanServ20:20
gsomixiglesiasg, finally I learned how to work with SGObjects.20:22
gsomixgood evening20:23
@iglesiasggsomix: cool!20:23
@iglesiasggsomix: do you feel like writing about it? maybe we can put it in the website so it is useful for everybody20:23
@iglesiasggsomix: a very few lines would be enough20:23
gsomixiglesiasg, hm, good idea.20:24
@iglesiasggsomix: or maybe just a mail in the mailing list like what you did for git flow20:25
gsomixhuh, now it is hard time at my university. many school leavers tries win admission to SSAU.20:31
gsomixa lot of buzz in internets20:31
@iglesiasggsomix: school leavers? What did they do?20:33
@iglesiasgstop studying or so?20:33
gsomixiglesiasg, I mean many people after high school ties enroll at university. now it's time.20:37
gsomix*tries20:37
@iglesiasggsomix: aaah ok, got it20:37
@iglesiasggsomix: I have a question btw20:37
@iglesiasggsomix: https://github.com/gsomix/shogun/blob/b933aca7c513d8073bb7af830317fb4ff7ccf85f/examples/undocumented/libshogun/library_circularbuffer.cpp20:37
@iglesiasggsomix: is start given some value by default? I thought local variables are not given any default value and this may cause problems in this example20:38
@iglesiasggsomix: valgrind does not complain here about anything, so I am just wrong probably, sorry.20:41
gsomixiglesiasg, hm, next_token_idx gets reference to 'start' variable and initializes it20:43
gsomixhttp://arxiv.org/ftp/arxiv/papers/1110/1110.2832.pdf --- best abstract20:44
gsomixiglesiasg, but I can init it with value by default. not problem :)20:45
@iglesiasggsomix: no, no need sorry20:46
@iglesiasggsomix: I asked you before checking the function signature and missed the &20:46
-!- lisitsyn [~lisitsyn@213.87.131.211] has joined #shogun20:52
-!- van51 [~van51@ppp-94-66-56-238.home.otenet.gr] has quit [Read error: Connection reset by peer]20:59
shogun-notifier-shogun: Evgeniy Andreev :develop * b933aca / / (12 files): https://github.com/shogun-toolbox/shogun/commit/b933aca7c513d8073bb7af830317fb4ff7ccf85f21:02
shogun-notifier-shogun: fixed ref/unref for readers21:02
shogun-notifier-shogun: Fernando Iglesias :develop * 02860e2 / / (12 files): https://github.com/shogun-toolbox/shogun/commit/02860e2cc0cdc906f3ce1be076c3a546d4b9ac3b21:02
shogun-notifier-shogun: Merge pull request #1334 from gsomix/feature/parser21:02
shogun-notifier-shogun:21:02
shogun-notifier-shogun: Bugfixes21:02
gsomixiglesiasg, thanks!21:02
@iglesiasggsomix: thanks to you for the fix :)21:03
gsomixit's just my job. ?)21:04
lisitsyniglesiasg: lets do stuff!21:05
@iglesiasglisitsyn: what do you want to do?21:13
lisitsyniglesiasg: no idea :D21:13
pickle27lisitsyn: have you had a chance to look at my PR?21:18
shogun-buildbotbuild #1125 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1125  blamelist: Evgeniy Andreev <gsomix@gmail.com>21:18
lisitsynpickle27: I am now trying to get rid of my bandwidth limit and then I'll check21:18
pickle27lisitsyn: okay np!21:19
* gsomix weeps21:19
shogun-buildbotbuild #1126 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1126  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>21:19
pickle27okay I'll be back in a bit!21:20
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving]21:20
-!- zxtx [~zv@rrcs-76-79-81-162.west.biz.rr.com] has joined #shogun21:22
gsomixsonney2k, hey21:29
@sonney2kgsomix, wassup?21:31
gsomixsonney2k, I will not available tomorrow at morning and day. want to visit lecture of Nobel prize winner 200021:33
gsomixsonney2k, I fixed today some bugs related to memory and ref/unref.21:33
@sonney2kgsomix, I hope you wrote some unit tests for that?21:35
gsomixsonney2k, hm, I wrote additional unit-test for reading/writing integers in csv. but for ref/unref test are the same21:36
gsomix*tests21:37
gsomixsonney2k, I just learned how we works with SGObjects in SHOGUN, no more.21:37
gsomixsonney2k, but I found bad case for CircularBuffer today. so I plan to refactor and simplify it for further support. I think it takes me one evening (tomorrow?)21:39
gsomixnow I have CircularBuffer at paper and it works fine. :D21:39
gsomixnext I can finish libsvm and quoting.21:40
gsomixsonney2k, all ok?21:42
@wikingsonney2k: ping21:49
@wikinglisitsyn: check this shit, it's really cool: http://www.cmake.org/Wiki/CMake:Component_Install_With_CPack21:51
lisitsynwiking: checking21:52
@wikinglisitsyn: its really cool. with the current structure of cmake i can easily create different components, i.e. packages like deb, rpm for the modular interfaces21:53
shogun-buildbotbuild #1437 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1437  blamelist: Evgeniy Andreev <gsomix@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>21:53
lisitsynwiking: well easily?21:54
lisitsyn;)21:54
lisitsynI am usually quite stucked with that shit21:54
@wikinglisitsyn: yeah easily since i just have to add one extra line into INSTALL(...) lines (COMPONENT python_modular....)21:55
@wikingand after that i we just have to define what kind of cpack generator we want to use21:55
@wikinglike deb, rpm, tar.gz, tar.bz2 etc.21:55
@wikinggood shit ;)21:56
@wikingand after this we just have a new build target: make package21:56
@wikingand that'll generate the packages ;P21:56
lisitsynwiking: yeah that's cool indeed21:56
@iglesiasgwiking: magic!21:59
@wikingyeah it's all fucking magic with cmake :)21:59
@wikingalthough one big target is still missing from the current setup22:00
@wikingmake tests22:00
@wiking:D22:00
@wikingthe rest is solved22:00
@wikingalthough a rebase will give me some headache22:01
@wikingok back to finish up make tests :)22:03
shogun-notifier-shogun: Viktor Gal :feature/CMake * ccdd072 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/ccdd0726c6e8db177c3c1a36e9860e8044f444de22:11
shogun-notifier-shogun: Add FindRubyNArray and fix octave build on travis22:11
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun22:16
@sonney2kwiking, yeah that won't work for debian packages but sure for pure data containers it would22:47
@sonney2kgsomix, well eah22:48
@sonney2kyeah22:48
@wikingsonney2k: ?23:01
@wikingsonney2k: why wouldn't it work?23:01
@wikingsonney2k: i mean i get it that we have a compilation size problem to get the debian packages into the debian upstream but hack we still can provide the deb packages ourselves... same with rpm23:07
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving]23:11
-!- travis-ci [~travis-ci@ec2-23-22-148-84.compute-1.amazonaws.com] has joined #shogun23:13
travis-ci[travis-ci] it's Heiko Strathmann'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/965791323:13
-!- travis-ci [~travis-ci@ec2-23-22-148-84.compute-1.amazonaws.com] has left #shogun []23:13
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has quit [Quit: Leaving]23:15
-!- FSCV [~FSCV@108.60.131.13] has quit [Quit: Leaving]23:25
@sonney2kwiking, the biggest issue w/ creating a .deb / rpm is to have a proper license file, dependencies complete and use the native build system (of e.g. dh_python2 in debian) - so you manually need to write some control, rules and, copyright file23:27
shogun-buildbotbuild #1436 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1436  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>23:28
@wikingsonney2k: dude have u read the wiki?23:29
@wikingsonney2k: cpack generates most of the files for u23:29
@wikingand uses native tools23:29
@wikinghence i cannot generate deb packages with cmake on my osx23:29
-!- hushell [~hushell@c-24-21-169-136.hsd1.or.comcast.net] has joined #shogun23:30
@sonney2kargh somebody broke gaussianprocessregression23:38
@sonney2k...in modular interfaces23:38
@wikingsonney2k: i.e. i've just created a shogun.dmg with it and it used the native PackageMaker of OSX23:42
@wikingso it's pretty cool23:42
@sonney2kwiking, sounds like!23:42
@sonney2kwiking, that certainly would solve the complaints we were hearing at the WS23:43
shogun-notifier-shogun: Soeren Sonnenburg :develop * d6b980e / src/interfaces/modular/GaussianProcess.i: https://github.com/shogun-toolbox/shogun/commit/d6b980eee9055cc9875c104236920789133f4dd323:44
shogun-notifier-shogun: fix gp regression for modular interfaces23:44
@wikingsonney2k: http://www.cmake.org/Wiki/CMake:CPackPackageGenerators#DEB_.28UNIX_only.2923:50
@wikingif u r interested23:50
@sonney2kwiking, I don't think it will work with e.g python but libshogun sure23:53
@wikingsonney2k: why not?23:54
@sonney2kwiking, look at the current debian package23:56
@sonney2kit is not that trivial23:56
@sonney2kwiking, https://bollin.googlecode.com/svn/shogun/trunk/debian/23:57
@sonney2kalright the build should go green now and it is time to sleep for me23:57
shogun-buildbotbuild #1127 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1127  blamelist: Soeren Sonnenburg <sonne@debian.org>23:57
-!- travis-ci [~travis-ci@ec2-23-22-148-84.compute-1.amazonaws.com] has joined #shogun23:58
travis-ci[travis-ci] it's Fernando Iglesias'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/966159823:58
-!- travis-ci [~travis-ci@ec2-23-22-148-84.compute-1.amazonaws.com] has left #shogun []23:58
--- Log closed Wed Jul 31 00:00:54 2013