Open in new window / Try shogun cloud
--- Log opened Tue Apr 16 00:00:28 2013
-!- FSCV [] has joined #shogun00:22
-!- FSCV [] has quit [Quit: Leaving]00:30
-!- medeeiip [~medeeiip@] has quit [Quit: Leaving]01:20
shogun-buildbotbuild #363 of nightly_default is complete: Failure [failed test]  Build details are at
-!- iglesiasg [] has quit [Quit: leaving]04:31
-!- Yanglittle [deb20af8@gateway/web/freenode/ip.] has joined #shogun04:48
YanglittleHey, is there any body?04:48
Yanglittle- -04:52
-!- Yanglittle [deb20af8@gateway/web/freenode/ip.] has quit [Quit: Page closed]04:56
-!- kaushalb09 [] has joined #shogun05:21
-!- sumit_ [73f91219@gateway/web/freenode/ip.] has joined #shogun06:01
-!- shogun-notifier- [] has joined #shogun06:18
shogun-notifier-shogun: iglesias :develop * 6c54e86 / / (15 files):
shogun-notifier-shogun: Rename HMSVMLabels to SequenceLabels.06:18
shogun-notifier-shogun: Because these labels can be used in other StructuredModels apart06:18
shogun-notifier-shogun: from the HMSVMModel.06:18
shogun-notifier-shogun: Soeren Sonnenburg :develop * 3fcafc4 / / (15 files):
shogun-notifier-shogun: Merge pull request #973 from iglesias/develop06:18
shogun-notifier-shogun: Rename HMSVMLabels to SequenceLabels.06:18
-!- zuhao [] has joined #shogun06:19
-!- kaushalb09 [] has quit [Quit: Leaving]06:21
-!- sumit_ [73f91219@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]06:31
-!- hoijui [] has joined #shogun06:38
-!- hoijui [] has quit [Ping timeout: 245 seconds]07:49
-!- erlenda [] has joined #shogun08:12
-!- lisitsyn [~blackburn@] has quit [Quit: Leaving.]08:24
-!- hoijui [] has joined #shogun08:31
-!- Yanglittle [deb20af2@gateway/web/freenode/ip.] has joined #shogun08:51
Yanglittlehey, the L1 norm mkl is too slow.08:53
-!- zuhao [] has quit [Quit: zuhao]09:12
-!- shogun-notifier- [] has quit [Quit: transmission timeout]09:18
-!- lisitsyn [] has joined #shogun09:33
-!- mode/#shogun [+o lisitsyn] by ChanServ09:33
-!- zuhao [] has joined #shogun09:46
-!- hoijui [] has quit [Ping timeout: 248 seconds]09:51
-!- hoijui [] has joined #shogun10:08
-!- Yanglittle [deb20af2@gateway/web/freenode/ip.] has quit [Quit: Page closed]10:09
-!- hoijui [] has quit [Ping timeout: 264 seconds]10:15
-!- hoijui [] has joined #shogun10:20
-!- iglesiasg [] has joined #shogun10:46
-!- mode/#shogun [+o iglesiasg] by ChanServ10:46
-!- erlenda [] has quit [Remote host closed the connection]11:04
-!- HeikoS [~heiko@] has joined #shogun11:27
-!- shogun-notifier- [] has joined #shogun11:30
shogun-notifier-shogun: lambday :develop * fd9d964 / tests/unit/mathematics/
shogun-notifier-shogun: fixed memory leak in Statistics.log_det_test_3 unit-test11:30
shogun-notifier-shogun: lambday :develop * 7031798 / tests/unit/mathematics/
shogun-notifier-shogun: fixed memory leak in Statistics.log_det_test_3 unit-test11:30
shogun-notifier-shogun: Heiko Strathmann :develop * e1718e0 / tests/unit/mathematics/
shogun-notifier-shogun: Merge pull request #975 from lambday/develop11:30
shogun-notifier-shogun: fixed memory leak in Statistics.log_det_test_3 unit-test11:30
@lisitsynHeikoS: btw it is my bad I merged that stuff sorry11:30
@lisitsynnever trust people hah11:31
HeikoSlisitsyn:  no worries11:31
HeikoSI mean he said he has run the tests11:31
HeikoSmaybe unit tests only (they seem to work)11:31
HeikoSwe should have a howto on that11:31
HeikoSis he replying to fix?11:31
@lisitsynnot yet replied11:31
HeikoSlisitsyn: say, about this eigen3 sparse matrix class11:31
HeikoShow do I make a class invisible for swig?11:32
HeikoSjust not put an interface?11:32
@lisitsynHeikoS: yes if it is not included it won't be used by swig11:32
HeikoSlisitsyn: okay11:32
HeikoSwhere is a good place to put the eigen3 helper class?11:32
@lisitsynHeikoS: actually mathematics/eigen3.h may be a good place11:32
HeikoSlisitsyn: okay!11:33
@lisitsynHeikoS: I hope we don't include it in headers11:33
HeikoSlisitsyn: what do you mean?11:33
HeikoSinclude eigen3=?11:33
@lisitsyninclude <shogun/mathematics/eigen3.h> should happen only in .cpp11:33
HeikoSwhy should we? class members should be G11:33
HeikoSSG type11:34
HeikoShah typing is sometimes not so easy11:34
@lisitsynwhat are you using? :)11:34
HeikoSlisitsyn: LOL11:36
HeikoSthere is another guy for the gaussian project11:36
HeikoSwho has the same name11:36
HeikoSas the first one :)11:37
@lisitsynHeikoS: no way11:39
@lisitsynHeikoS: really?11:39
HeikoSyes, so first name11:39
@lisitsynwait - just the same?11:39
HeikoSfunny isn't it?11:39
@lisitsynSoumyajit De11:39
@lisitsynthat guy?11:39
@lisitsynand there is an another one?11:40
HeikoSlisitsyn: he is acutally called Rahul11:40
HeikoSand the other one also is11:40
HeikoSat least thats how they sigh their mails11:40
@lisitsynHeikoS: I've lost a track what is happening :D11:40
-!- hoijui [] has quit [Ping timeout: 246 seconds]11:41
sonne|workHeikoS: did you read the workshop CfP?11:52
HeikoSsonne|work: yes, I will send some feedback later11:52
-!- gsomix [~androirc@] has joined #shogun11:59
HeikoSgsomix: hi!12:00
-!- erlenda [] has joined #shogun12:10
HeikoSerlenda: ha!12:15
HeikoSany particular reason why you are here? :)12:15
@lisitsynHeikoS: I asked all mentors to hang here12:16
HeikoSlisitsyn: I see12:16
HeikoSwas just wondering whether there are any questions12:16
erlendayeah, so I just started hanging12:16
HeikoSerlenda: nice :)12:16
@lisitsynerlenda: HeikoS: how do you guys planning to mentor it?12:16
@lisitsynthree mentors!12:16
HeikoSlisitsyn: I had the following in mind: me: helping out with shogun stuff, erlend: helping out with C++-implentation questions/design, dan: helping out with general big picture of what to use/not use12:17
@lisitsynI see12:18
HeikoSwe probably will skype about it soon12:18
HeikoSto get a better plan12:18
@lisitsynerlenda: so you have some background in C++?12:18
HeikoSlisitsyn: he wrote the KRYLSTAT that we want to copy :)12:19
@lisitsynI see12:19
erlendaWritten some C++. Some experience in templating and stuff12:19
erlendaBut I am no guru12:19
@lisitsynnice to know12:19
HeikoSlisitsyn: you are the guru :)12:20
@lisitsynHeikoS: of C++?12:20
HeikoSyes, and + other things12:20
@lisitsynwell, I know some tricks but I know nothing about machine learning though :)12:20
HeikoSerlenda: I will propose a google hangout with daniel soon where we can discuss a few things, also would be nice just to say hello face to face. Next week or so12:21
HeikoSlisitsyn: so I always can ask you thats enough :)12:22
erlenda@HeikoS: Sounds good12:22
@iglesiasglisitsyn: nothing??! maybe you don't have a Ph.D. in ML but I think you know way more than nothing ;)12:22
@lisitsyniglesiasg: comparing to say HeikoS? :)12:22
@iglesiasgnot good to compare, we are talking about absolute knowledge :)12:22
@lisitsynI do not do any research - that's my disaster :)12:23
@lisitsyniglesiasg: did I told you I got that dummy callbacks thing to work?12:24
@iglesiasglisitsyn: yes12:24
@lisitsynI got pretty tired after that and powered off :D12:25
@iglesiasghehe understandable12:25
@lisitsynwill commit that stuff today12:25
@iglesiasgnice, I will see it then12:25
@lisitsynhave to cover it with a few tests12:25
@lisitsynHeikoS: ah btw we wanted to discuss a new dimension reduction algorithm based on MMD :)12:27
HeikoSlisitsyn: yes, we should do that at some point :)12:28
HeikoSlisitsyn: once I finish this one project here, I will have way more time12:28
HeikoSand its almost done12:29
@lisitsynHeikoS: ok just let me know12:29
HeikoSwill do!12:29
-!- erlenda [] has quit [Quit: Leaving]12:40
-!- iglesiasg [] has quit [Ping timeout: 245 seconds]12:52
-!- hoijui [] has joined #shogun12:59
-!- HeikoS [~heiko@] has quit [Quit: Leaving.]13:13
-!- iglesiasg [] has joined #shogun13:15
-!- mode/#shogun [+o iglesiasg] by ChanServ13:15
@iglesiasgany hint why a class guarded with a USE_* variable is not compiled when that variable is actually set?13:17
@iglesiasgI am including this variable by hand in the DEFINES line in the .config file13:17
@iglesiasgand I actually can see it used in the compile lines when issue make13:18
-!- iglesiasg [] has quit [Client Quit]13:19
-!- iglesiasg [] has joined #shogun13:20
-!- mode/#shogun [+o iglesiasg] by ChanServ13:20
-!- hoijui [] has quit [Ping timeout: 264 seconds]13:20
@lisitsyn iglesiasg it should :)13:26
@iglesiasglisitsyn: no idea why the class is not compiled13:26
sonne|workiglesiasg: swig?13:37
sonne|workiglesiasg: no it needs to be in lib/config.h13:37
@iglesiasgsonne|work: no, it is my PrimalMosekSOSVM that is guarded with USE_MOSEK13:37
@iglesiasgsonne|work: the USE_MOSEK needs to be in lib/config.h?13:38
@iglesiasgI have built it several times without modifying lib/config.h13:38
@iglesiasgit is true it is not the first time I find this problem though13:38
sonne|workiglesiasg: lib/config.h is autogenerated13:38
sonne|workbut defines need to be in there13:38
@iglesiasgsonne|work: how is it possible that it worked other times then?13:39
@iglesiasgI have never put it in there13:40
sonne|workiglesiasg: didn't we change configure all the time for that?13:40
@iglesiasgsonne|work: I don't understand, the configure script? when all the time?13:41
-!- deerishi [73f88294@gateway/web/freenode/ip.] has joined #shogun13:42
deerishihow can we plot the data of a object if it is a shogun feature object?,using matplotlib13:42
@iglesiasgdeerishi: getting its underlying data13:43
@iglesiasgdeerishi: what type of feature are you using?13:43
deerishii am trying to implement kernel pca on a non separable data,13:44
-!- gsomix [~androirc@] has quit [Quit: AndroIRC - Android IRC Client ( )]13:44
deerishii want to see the data after kernel pca has been done13:44
@iglesiasgdeerishi: please see examples/undocumented/python_modular/graphical/ for example13:46
@iglesiasgline 81: X = embedding.get_feature_matrix()13:46
@iglesiasgX is now a numpy array and you can do with it plotting stuff13:46
deerishi@iglesiasg: will look in that.thank u13:48
-!- hoijui [] has joined #shogun13:58
-!- abinash [~abinash@] has joined #shogun14:09
-!- shogun-notifier- [] has quit [Quit: transmission timeout]14:30
-!- lambday [3ba21749@gateway/web/freenode/ip.] has joined #shogun14:45
deerishiwhat does LoadMatrix mean?14:51
deerishii mean its function?14:54
@iglesiasgdeerishi: why not to check in the source code? :)14:56
-!- HeikoS [] has joined #shogun15:01
lambdayHeikoS: hi..15:02
HeikoSlambday: hi!15:02
HeikoSjust answered your mail15:03
lambdayHeikoS: checking...15:03
lambdayHeikoS: umm... only from cpp for avoiding including it everywhere even when not needed..15:05
HeikoSlambday: yes and to save trouble with swig15:06
HeikoSif you include it in some class that has swig interfaces, it might cause problems15:06
HeikoSso just use in cpp15:06
HeikoSbut that was just a minor comment15:06
HeikoSlambday: good stuff, we will need this often so very useful15:07
-!- abinash [~abinash@] has quit [Ping timeout: 258 seconds]15:09
lambdayHeikoS: I guess :) ... so I'm adding a cpp..15:11
lambdayHeikoS: I just saw the other mail.. this guy is from my institute only :D15:11
@iglesiasgsonne|work: adding USE_MOSEK in lib/config.h fixed it, thanks!15:16
@iglesiasgstill, weird thing that I didn't need to do that any time before. Did something related to lib/config.h change recently?15:17
-!- diwmaron [528b8ddc@gateway/web/freenode/ip.] has joined #shogun15:22
@lisitsyndiwmaron: hey15:23
diwmaronI'm interested about topic of Google Summer of code15:23
lambdayHeikoS: the custom EigenTriplet definition shouldn't also be in the header? Its already there in the newer eigen, we just need it for older eigen. would it cause trouble too?15:23
diwmaronImplement metric learning algorithms with applications to metagenomics15:23
@lisitsynlambday: I'd put it to eigen3.h15:24
diwmaronI'm not sure if I understood problem correctly15:24
HeikoSlambday: yeah, agree with Sergey15:24
@lisitsyndiwmaron: which problem?15:24
diwmaronIs there anyone who could give me some more details?15:24
lambdayHeikoS, lisitsyn: okay..15:25
@iglesiasgdiwmaron: someone else asked a few hours ago in the mailing list about this project15:25
@lisitsyndiwmaron: there is a mentor for that idea - he is not around right now but I'd recommend send him an email15:25
@iglesiasgdiwmaron: and the mentor answered to that mail15:25
@iglesiasgdiwmaron: you may find that info interesting15:25
diwmaronThanks a lot!15:25
diwmaronsee you in the future15:26
@iglesiasgdiwmaron: do you know how to read the mailing list or should I point you to it?15:26
diwmaronThere is no need. I have basic knowlege of mailing lists  :) Thanks!15:27
-!- iglesiasg [] has quit [Ping timeout: 260 seconds]15:42
-!- iglesiasg [] has joined #shogun15:43
-!- mode/#shogun [+o iglesiasg] by ChanServ15:43
-!- gsomix [~gsomix@] has joined #shogun15:57
-!- shogun-notifier- [] has joined #shogun16:00
shogun-notifier-shogun: Roman Votyakov :develop * fd4547d / src/shogun/evaluation/GradientResult.h:
shogun-notifier-shogun: Fix segfaults in libshogun GP examples16:00
shogun-notifier-shogun: Heiko Strathmann :develop * cf23154 / src/shogun/evaluation/GradientResult.h:
shogun-notifier-shogun: Merge pull request #976 from votjakovr/develop16:00
shogun-notifier-shogun: Fix segfaults in libshogun GP examples16:01
sonne|workiglesiasg: no nothing changed16:02
@iglesiasgsomething fishy then16:02
-!- gsomix [~gsomix@] has quit [Ping timeout: 245 seconds]16:06
-!- diwmaron [528b8ddc@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]16:13
@iglesiasgsonne|work: the with only the line branches: it should work?16:23
-!- deerishi [73f88294@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]16:27
sonne|workiglesiasg: totally remove the branches16:30
@iglesiasgyaay I am back to black screwing up pull requests16:32
@iglesiasgaaah shit I think I know what I did wrong16:34
wikingiglesiasg: what would be really good as well to do different batches on travis.16:34
wikingiglesiasg: with that we could test as well the different interfaces on travis16:34
@iglesiasgwiking: aham!16:35
@iglesiasgit sounds interesting16:35
wikingiglesiasg: i've had this planned out but never managed to go around it and change the yml file16:36
wikingiglesiasg: but this can be definitely done by using batches or what...16:36
wikingiglesiasg: with env variable we can do this16:38
wikingiglesiasg: so by defining an env variables like INTERFACES16:38
wikingiglesiasg: and then basically create a matrix that has the various interfaces and then the configure should be something like ./configure --interfaces=$INTERFACES16:39
wikingthe only thing should be still fixed is that the right packages are installed for the given interface16:40
@iglesiasgwiking: that could get messy, or?16:40
wikingthe reasone why we need to do a one interface per env variable is because otherwise you run out of hard limit for building time16:40
wikingiglesiasg: why?16:40
@iglesiasgto get all the dependencies16:40
wikingiglesiasg: well as we know what are the dependencies for the given inteface16:41
wikingwe just set a new env variable for the extra packages we need for the given interface16:41
wikinglet say you have a EXTRA_PACKAGES env variable16:41
wikingand that is set16:41
wikingas well with the INTERFACE variable16:42
wikingiglesiasg: and then in the before_install part you append only that variable $EXTRA_PACKAGES16:42
wikingto the apt-get install16:42
wikingit can be empty16:43
wikingi mean $EXTRA_PACKAGES16:43
wikingif it only build the core part16:43
@lisitsynwiking: oh long time no see!16:43
wikinglisitsyn: yeah back16:43
wikinglisitsyn: conference is over16:43
@lisitsynwiking: what conference?16:43
wikingisbi 201316:43
@lisitsynand how it was?16:44
wikingit was great16:45
wikingand san francisco is cool as well16:45
wikingthe city itself16:45
wikingi've promoted there shogun16:45
@lisitsynoh that's nice16:45
wiking2 guys heard and constantly use it ;)16:45
wikingand another 5 guys were really interested in it16:45
wikingthe 2 guys who use it really appreciate it16:46
wikingand they really like it... at least they told me so16:46
@iglesiasgwiking: what were they using it for?16:46
@iglesiasgor are16:46
@lisitsynwiking: so you didn't make them suffer to say that?16:46
@lisitsynno torments?16:46
wikingmostly they are using the svm part of it16:46
wikingi told them there is much more to it as well... as well as there's gonna be some really good stuff coming this summer with GSoC1316:47
wikingiglesiasg: so what do you say about this ?16:48
@iglesiasgwiking: about the travis thing?16:48
wikingiglesiasg: yes16:50
wikingi mean what we really need only16:50
wikingis for a given interface what other packages we need16:50
wikingapart from the basic ones16:50
@iglesiasgI don't understand it completely clear TBH16:50
wikingso now we build only libshogun interface16:51
wikingon travis16:51
@iglesiasgok, I didn't know it was just libshogun actually :)16:51
wikingwhat we could do is that we also build perl_modular, python_modular, r_modular, octave_modular, java_modular, csharp_modular, lua_modular, ruby_modular16:51
wikingthe only problem is16:52
wikingthat travis has a hard limit on compilation time16:52
wikingso if we build all the modular interfaces at once16:53
wikingwe'll run out of that hard limit16:53
wikingso what we shoudl do is divide up this task16:53
wikingi.e. each interface should be build separately16:53
HeikoSwiking, it would be cool to have the python working16:53
HeikoSthis is the most interesting for travis16:53
wikingso this can be done by using enviromental variables on travis16:54
wikingbecause travis automatically will try to run a separate build process for each different values of a given env variable16:54
wikingso what i suggest is to have 2 env variables16:55
wikingiglesiasg: see the example what they say there for ruby and FOO and BAR16:55
wikingso 1 env variable is INTERFACES16:56
wikingwhich always have only one value16:56
wikinge.g. INTERFACES=python_modular16:56
@lisitsynyes makes sense16:56
wikingand you have a corresponding env variable16:56
wikingcan i now16:57
wikingdirectly commit to develop16:57
@lisitsynwiking: yes sure16:57
wikingwith git-flow method?16:57
wikingok let's see how this works :)))16:57
@lisitsynwiking: I don't mind even w/o git-flow for small commits (not sure what others think)16:57
@iglesiasgwiking: and how is it done to control when one of the builds ends (say python_modular) another starts (libshogun or whatever)16:57
wikingiglesiasg: travis will take care of automatically16:58
wikingcan anybody tell me what is the dependency for python_modular16:59
wikingi mean packages on16:59
@iglesiasganything else I can think of apart from python, numpy, swig17:00
@iglesiasgnothing* hehe17:01
@lisitsynwiking: python-numpy-dev17:02
@lisitsynswig2.0 IIRC17:02
@iglesiasghaha oldie one17:04
wikingEXTRA_PACKAGES="python-numpy-dev, swig2.0"17:08
wikinganything else?17:08
@lisitsynwiking: sorry not under ubuntu right now17:09
@lisitsynwiking: python-dev maybe17:09
HeikoSlisitsyn: how to tell the buildbot to rebuild the python modular?17:11
@lisitsynshogun-buildbot force17:11
@lisitsynshogun-buildbot force restart something17:12
HeikoSshogun-buildbot: force restart deb317:12
shogun-buildbottry 'force build [--branch=BRANCH] [--revision=REVISION] [--props=PROP1=VAL1,PROP2=VAL2...]  <WHICH> <REASON>'17:12
@lisitsynshogun-buildbot:  force build 'deb3 - modular_interfaces'17:12
@lisitsynshogun-buildbot:  force build 'deb3 - modular_interfaces' 'please'17:13
HeikoSdid it work?17:13
@lisitsynno idea :D17:13
HeikoSpleeeease :)17:13
@lisitsynshogun-buildbot, force build 'deb3 - modular_interfaces' 'please'17:13
shogun-buildbotbuild forced [ETA 44m36s]17:13
shogun-buildbotI'll give a shout when the build finishes17:13
@lisitsynHeikoS: ha triggered17:13
HeikoS :)17:13
@lisitsynHeikoS: do you know there is a programming language17:13
HeikoSthe please did it!17:13
@lisitsynwhere you have to put please17:13
@lisitsyn if "PLEASE" does not appear often enough, the program is considered insufficiently polite, and the error message says this; if too often, the program could be rejected as excessively polite. 17:14
HeikoShaha :)17:14
HeikoSplease give up!17:14
@lisitsynok be back later17:15
-!- lisitsyn [] has quit [Quit: Leaving.]17:15
shogun-notifier-shogun: Viktor Gal :develop * 6b18c90 / .travis.yml:
shogun-notifier-shogun: Add support for building various interfaces on Travis17:16
shogun-notifier-shogun: Currently it only works for python_modular17:16
wikinglet see how this will end up17:16
HeikoSwiking: nice! :)17:16
wikingi wonder if it's going to be really working17:18
wikingbut either my internet17:18
wikingor travis is slow17:18
-!- foulwall [~foulwall@2001:da8:215:c252:1961:62de:1e65:4d09] has joined #shogun17:19
wikingah ok17:23
wikingi understand now17:23
wikingiglesiasg: here17:23
wikingiglesiasg: we have to change the travis.yml file in master branch17:24
wikingotherwise it'll never realise17:24
wikingthat it should start building the develop branch :D17:25
wikingwho will merge it into master? :)17:26
HeikoSlambday: check python_modular warnings:17:26
wikingHeikoS: wanna merge it?17:34
-!- deerishi [73f88294@gateway/web/freenode/ip.] has joined #shogun17:35
lambdayHeikoS: oh.. I added precision_matrix later, forgot to add the doxygen info17:37
HeikoSwiking, fernandos PR?17:37
HeikoSlambday:  no worries!17:37
-!- foulwall [~foulwall@2001:da8:215:c252:1961:62de:1e65:4d09] has quit [Ping timeout: 245 seconds]17:40
@iglesiasgwiking: re17:42
wikingsomebody should merge the current .travis.yml in develop to master branch17:42
wikingok waith17:42
wikingnot yet17:42
wikingiglesiasg: so should we always try to build master as well?17:42
wikingor only develop?17:43
wikingi would go with master and develop17:43
@iglesiasgwiking: AFAIK travis is going to build only when we push stuff17:43
wikingiglesiasg: yes17:43
@iglesiasgso if we don't have much activity in master for a long while17:43
@iglesiasgas it should be with this git-flow model17:43
@iglesiasgthen it is not going to keep on building the same17:44
wikingnow it builds only master17:44
@iglesiasgthen I think it makes sense to build master too17:44
wikingas we set it up like that17:44
@iglesiasgI issued a pull request so that all branches were built17:44
wikingi think it's not smart enough17:44
@iglesiasgsonne|work wanted to have it like that17:44
lambdayHeikoS: fixed...17:44
wikingto know that which branch is pushed17:44
lambdaysending the PR :)17:44
HeikoSlambday: nice :)17:44
wikingiglesiasg: if travis is being notified that there was a push17:45
wikingthen it'll try to build all the branches that are mentioned17:45
wikingnot only the ones where there's a change17:45
wikingTravis will always look for the .travis.yml file that is contained in the branch specified by the git commit that GitHub has passed to us. This configuration in one branch will not affect the build of another, separate branch. Also, Travis CI will build after any git push to your GitHub project unless you instruct it to skip a build. You can limit this behavior with configuration options.17:46
wikingmaybe it is actually smart enough ;)17:46
wikingor not17:46
wikingi dont really get this explanation17:46
lambdayHeikoS: one more thing, if I add a new file.. all I need to do is run configure again and then make, right?17:47
wikingfrom the last 2 sentence i think it'll build all the branches17:47
wikingregardless of which branch was modified17:47
wikingbut from the first part of this i understood that it'll do the work only on the modified branch17:47
HeikoSlambday: what file and where?17:48
shogun-notifier-shogun: lambday :develop * 7a77a74 / src/shogun/mathematics/Statistics.h:
shogun-notifier-shogun: fixed doxygen error in Statistics.h17:48
shogun-notifier-shogun: Heiko Strathmann :develop * b40b0c2 / src/shogun/mathematics/Statistics.h:
shogun-notifier-shogun: Merge pull request #979 from lambday/develop17:48
shogun-notifier-shogun: fixed doxygen error in Statistics.h17:48
HeikoShaha (Develop)17:48
HeikoSHeiko: Develop17:49
@iglesiasgwiking: it sounds to me like if you push to one branch, then it is going to build for all the branches as specified in the travis.yml of the branch you have pushed to17:49
@iglesiasgdoes that make sense?17:49
wikingso basically you understand17:50
wikingthat it'll build17:50
wikingon the branch where there was modification17:50
wikingand only if it's enabled17:51
@iglesiasgnot really17:51
-!- deerishi [73f88294@gateway/web/freenode/ip.] has quit [Quit: Page closed]17:51
lambdayHeikoS: eigen3.cpp in shogun/mathematics/.. where SGSparseMatrix to SparseMatrix is to be defined..17:51
wikingiglesiasg: then? :)17:51
HeikoSlambday: then yes, just add the file, linking should work automatically17:52
@iglesiasgwiking: let's make some syntax analysis :D17:52
@iglesiasgTravis will always look for the .travis.yml file that is contained in the branch specified by the git commit that GitHub has passed to us17:52
@iglesiasg1) What is the commit that GitHub passes to them?17:52
wikingthat's done by a hook17:52
wikinggithub commit hook17:52
shogun-buildbotbuild #1012 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at
@iglesiasgno idea how the hook works17:53
wikingi think it is executed17:53
wikingas a post commit hook17:53
wikingso when github receives a new commit17:53
wikingit executes the hook of travis17:53
lambdayHeikoS: okay.. :)17:53
wikingand it passes there the branch where the commit happened17:53
wikingso if it receives a new commit to develop17:54
wikingit'll send to travis17:54
wikingthe branch name17:54
wikingand the commit sha117:54
wiking^ branch name = develop17:54
wikingand then travis will check whether that branch has a .travis.yml17:54
@iglesiasgwiking: but that sounds a bit like if it'd do it commit by commit17:55
@iglesiasgand probably it is not17:55
@iglesiasgwiking: what did you mean with post commit hook?17:55
@iglesiasgmm you explained afterwards, nvm17:56
wikingAlso, Travis CI will build after any git push to your GitHub project unless you instruct it to skip a build.17:56
wikingthis should mean for me17:56
wikingi mean this means for me17:56
wikingthat it'll run17:56
wikingafter any git push17:56
@iglesiasgand that makes lot of sense17:56
wikingunless it's specified otherwise17:56
wikingnow the question is17:57
@iglesiasghowever, I get lost in there with this "commit that GitHub passes us"17:57
wikingiglesiasg: github will pass to Travis17:57
wikingin what branch happend the change17:57
wikingand the commit sha-1 string17:57
wikingso that travis can do the building of that given commit17:57
wikingand yes17:58
wikingtravis runs after each commit17:58
@iglesiasgso say you do a push with 20 new commits17:58
@iglesiasgit is going to build 20 times?17:58
wikingor i hope so :)17:58
wikingotherwise i dont understand a fucking thing17:58
@iglesiasgok, there must be a reason to do it like that :)17:58
wikinglet me check something17:59
@iglesiasgwiking: then, I think we don't need to push to master in order to get travis to work in develop17:59
wikingi might be wrong17:59
wikingiglesiasg: there's only one build for that PR18:00
wikingnot 418:00
wikingso maybe it's push per push18:00
wikingok let's try it18:01
@iglesiasgwiking: is travis utterly slow in your browser too?18:02
@iglesiasgI have cliked in pull requests tab one month ago :D18:02
wikingiglesiasg: yes18:03
wikingit's fucking slow18:03
wikingok i'll just do a direct commit now the develop branch18:03
wikingadd develop branch into .travis.yml18:03
wikingeven better18:03
wikingi do your PR :D18:03
@iglesiasgwiking: yeah18:03
@iglesiasgI was going to tell you that :)18:03
wikingand then we will invalidate your PR ;)18:04
-!- deerishi [73f88294@gateway/web/freenode/ip.] has joined #shogun18:04
@iglesiasgbut actually that change should be in all the branches18:04
wikingyeah i know :)18:05
shogun-notifier-shogun: Viktor Gal :develop * adea65e / .travis.yml:
shogun-notifier-shogun: No filtered branches in travis.yml18:08
shogun-notifier-shogun: inspired by Fernando18:08
wikinglets see18:08
wikingtravis should start doing soemthing now18:08
wikingit doesn't do anything yet18:09
@iglesiasgwiking: why didn't you just merge the PR? Just curious because I don't see the difference18:10
wikingiglesiasg: because i've already did some changes earlier18:10
@iglesiasgaaah all right18:10
wikingi've added the env variable18:10
wikingtravis is crazy slow now18:10
@iglesiasgand we want to put a million interfaces :D18:11
-!- hoijui [] has quit [Ping timeout: 245 seconds]18:14
shogun-notifier-shogun: Heiko Strathmann :develop * c1c5078 / data:
shogun-notifier-shogun: updated data with new GPR test example due to implementation change18:19
shogun-notifier-shogun: Heiko Strathmann :develop * 3cfc714 / data:
shogun-notifier-shogun: Merge pull request #980 from karlnapf/develop18:19
shogun-notifier-shogun: updated data with new GPR test example due to implementation change18:19
HeikoSshogun-buildbot: force restart 'deb3 - modular_interfaces' 'please'18:21
shogun-buildbottry 'force build [--branch=BRANCH] [--revision=REVISION] [--props=PROP1=VAL1,PROP2=VAL2...]  <WHICH> <REASON>'18:21
HeikoSshogun-buildbot, force build 'deb3 - modular_interfaces' 'please'18:21
shogun-buildbotbuild forced [ETA 44m36s]18:21
shogun-buildbotI'll give a shout when the build finishes18:21
wikingthe problem is that it's still not building18:22
wikingit only built your PR18:22
wikingand actually based on that PR18:23
wikingit did not build ALL the branches18:23
@iglesiasgwiking: it only builds the branch where it is pushed18:23
@iglesiasgas long as the travis.yml says so18:23
wikingbut then why it did not build my last commit?!18:24
wikingstupid travis18:24
@iglesiasgmatter of time maybe?18:25
-!- zuhao [] has quit [Remote host closed the connection]18:33
-!- deerishi [73f88294@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]18:52
-!- shreyNis [3bb2c21c@gateway/web/freenode/ip.] has joined #shogun18:52
-!- cameron_ [] has joined #shogun18:53
-!- deerishi [73f88294@gateway/web/freenode/ip.] has joined #shogun18:53
@iglesiasghey cameron_18:54
cameron_i am trying to start in shogun18:54
cameron_and are having troubles in compiling the files18:54
cameron_planning to use C++ with g++18:54
cameron_got the basic example working. (the minimal one)18:54
cameron_and i downloaded the package just straight from synaptic package manager18:55
cameron_didn't bother to do make and install18:55
@iglesiasgcameron_: ok18:55
cameron_then was trying to run the gaussian filter example in the tutorial18:55
cameron_and it doesn't work18:55
cameron_a bunch of undefined reference popped up18:55
cameron_i think i just need to link back to the library18:56
@iglesiasgcameron_: mmm it might be that the tutorial is not sync with the version of shogun you are using18:56
HeikoScameron_: yes, -lshogun18:56
deerishiin KernelPCA in python modular, get_transformation_matrix() converts the shogun feature data matrix to the numpy matrix ,rite?(note i am doing get_transformation_matrix() after kernelpca has been done)18:56
HeikoSdeerishi: what exactly do you  mean?18:56
cameron_yes, i already used -lshogun18:57
cameron_still it doesn't work18:57
deerishii want to see the plotted data after kernel pca has been done18:57
deerishii want to confirm if the data has been made linearly separable18:58
HeikoSdeerishi: thats good, what is the question then?18:58
HeikoSshogun matrices are directly translated into numpy18:58
deerishiHeikoS : after doing "preprocessor.apply_to_feature_matrix(features) ",in order to see the results, i need to transform features to a numpy array,by doing get_transformation_matrix().but the thing is the matrix i am getting back is of greater dimension. eg if the data was a array of size 2x302 ,the matrix i am getting back is of size do i plot that to confirm if kernel pca has worked or not?19:02
cameron_any ideas?19:02
-!- shreyNis [3bb2c21c@gateway/web/freenode/ip.] has left #shogun []19:05
shogun-buildbotbuild #1013 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at
@iglesiasgcameron_: can you put your example and the instructions you are using to compile somewhere? e.g. pastebin19:06
-!- deerishi_ [73f88294@gateway/web/freenode/ip.] has joined #shogun19:08
-!- deerishi [73f88294@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]19:09
deerishi_HeikoS : after doing "preprocessor.apply_to_feature_matrix(features) ",in order to see the results, i need to transform features to a numpy array,by doing get_transformation_matrix().but the thing is the matrix i am getting back is of greater dimension. eg if the data was a array of size 2x302 ,the matrix i am getting back is of size do i plot that to confirm if kernel pca has worked or not?19:09
deerishi_HeikoS:any ideas?19:09
cameron_error log :
HeikoSdeerishi_: okay, I get the problem19:10
cameron_command i used: g++ file_name.cpp -o file_name -lshogun19:10
HeikoSdeerishi_: could you paste some code? pastebin19:10
HeikoSwhich illustrates?19:10
HeikoSthen I can verify19:11
@iglesiasgcameron_: what about the source code, can you copy it too please?19:11
HeikoScameron_: maybe your installation did not work properly, you could try compiling shogun from the source19:12
HeikoSiglesiasg: sonney2k: just realising how outdated the developer tutorial is ...19:13
@iglesiasgHeikoS: hehe indeed...19:13
HeikoSdeerishi_: checking ...19:17
HeikoSdeerishi_: sorry will take some time ..will get back to you19:26
deerishi_HeikoS:the data is 2 semicircular arcs of radius 50 and 100 respectively19:27
HeikoSdeerishi_: ok now19:32
HeikoSdeerishi_: so the transformation matrix are the eigenvectors19:35
HeikoSof the kernel matrix19:35
@iglesiasgsee you later people19:35
-!- iglesiasg [] has quit [Quit: leaving]19:35
HeikoSbut you want to plot the projection19:35
HeikoSsay on top of the first two eigenvectors19:35
deerishi_top 2 in the sense, say the first and second row, or according to the eigenvalues19:37
HeikoSno, the transformation matrix are just the eigenvalues19:38
HeikoSif you want to see the projection on the first two you have to multiply the first two19:38
HeikoSbut the KernelPCA class does this for you19:38
HeikoSthis method returns the new feature matrix with the projection19:39
HeikoSfloat64_t* new_feature_matrix = SG_MALLOC(float64_t, m_target_dim*num_vectors);19:39
HeikoSso you have to set the target dimension19:39
HeikoSand then call this method19:39
HeikoSthen your features have another dimension19:39
HeikoSdeerishi_:  gotta go now, hope that helped, if not, ask on the mailing list, I might have a look tonight19:40
-!- Ben___ [8984c79c@gateway/web/freenode/ip.] has joined #shogun19:46
-!- Ben___ [8984c79c@gateway/web/freenode/ip.] has quit [Client Quit]19:47
-!- HeikoS [] has left #shogun []19:56
-!- iglesiasg [] has joined #shogun19:57
-!- mode/#shogun [+o iglesiasg] by ChanServ19:57
-!- lambday [3ba21749@gateway/web/freenode/ip.] has quit []19:58
-!- deerishi_ [73f88294@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]20:00
@sonney2kiglesiasg, no wonder the developer tutorial is outdated :)20:06
@sonney2kthat is the destiny of all documentation20:06
@iglesiasgyeah... the cycle of life20:07
shogun-notifier-shogun: iglesias :develop * d594bb0 / .travis.yml:
shogun-notifier-shogun: No filtered branches in travis.yml.20:10
shogun-notifier-shogun: Soeren Sonnenburg :develop * 7096e33 / :
shogun-notifier-shogun: Merge pull request #978 from iglesias/travis_all_branches20:10
shogun-notifier-shogun: No filtered branches in travis.yml.20:10
@iglesiasgsonney2k: mmm I thought wiking had already applied that change20:11
-!- k_matsuda [] has quit [Remote host closed the connection]20:15
@iglesiasgsonney2k: arrgh you are the fastest in the old western, I had already started to answer Monalisa :P20:17
@iglesiasgsonney2k: the file looks like this btw
@iglesiasgit doesn't look at all like a autogenerated file20:24
@iglesiasgmaybe it is something I downloaded/copied half-asleep and now I don't remember20:24
-!- iglesiasg [] has quit [Quit: leaving]20:25
-!- cameron_ [] has quit [Quit: Leaving]20:37
-!- gsomix [~gsomix@] has joined #shogun20:41
gsomixgood evening20:41
-!- travis-ci [] has joined #shogun20:45
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:
-!- travis-ci [] has left #shogun []20:45
-!- lisitsyn [~blackburn@] has joined #shogun21:23
-!- mode/#shogun [+o lisitsyn] by ChanServ21:23
@sonney2klisitsyn, I am dead. I just answered 13 or so sg emails21:34
@lisitsynsonney2k: about?21:34
@lisitsynsonney2k: redirect to me whether possible21:35
@sonney2khashing/demo etc21:35
@sonney2kI missed foulwall today - did you see him?21:35
@sonney2kand I've seen another cool demo21:35
@lisitsynsonney2k: no21:35
@sonney2kcheck this out ^21:36
@sonney2kgit clone21:36
@sonney2kand then just make21:36
@sonney2k(you need shogun python modular installed)21:36
@sonney2klisitsyn, I am indeed thinking about extending the idea21:36
@sonney2kI mean with real shogun webservice style stuff21:37
@sonney2kand tutorial style intro21:37
@sonney2kand maybe splitting it up21:37
@sonney2kas that is way to big21:37
@lisitsynsonney2k: yes that's a trouble21:37
@sonney2kabout hashing we might also have some nice candidates21:39
@sonney2klisitsyn, do you recall how it was in last years21:39
@lisitsynsonney2k: what exactly?21:39
@sonney2kI mean when we slowly became sure who would be ideal for which slot?21:39
@lisitsynsonney2k: oh last time we had no doubts21:39
@lisitsynsonney2k: but now dates are shifted21:40
@sonney2knot from the very beginning21:40
@lisitsynso I am confused21:40
@lisitsynsonney2k: what I am sure - *after* submissions of proposals21:40
@sonney2kwell it is still 2 weeks to get to know candidates21:42
@sonney2kso maybe it will become more clear21:43
@lisitsynsonney2k: 2?21:43
@sonney2kbtw I don't remember that it was that long last time21:43
@lisitsynsonney2k: everything was much shorter21:43
@lisitsynsonney2k: I guess everyone is confused with that schedule21:43
@sonney2kyes may 3rd21:43
@sonney2kis the deadline21:44
@lisitsynI see21:44
@lisitsynwill try to remember21:44
@sonney2kit is 1 month of getting to know people21:44
@sonney2kvery long21:44
@sonney2kon the other hand they get to know us too :D21:44
@sonney2klisitsyn, did you try the demo?21:45
@lisitsynsonney2k: not yet will try in a minute21:45
@lisitsynI checked the code first21:45
@lisitsynsonney2k: oh that's interesting21:46
@sonney2klisitsyn, what?21:49
@lisitsynsonney2k: that demo21:49
@sonney2knice uh?21:50
@lisitsynyes quite nice21:50
@sonney2kI kind of dreamt of doing this21:50
@lisitsynso you will be in trouble21:50
@sonney2kwe will see21:51
@lisitsynyou are already in trouble with too many candidates :)21:51
@lisitsynsonney2k: worst thing is that they have zero conversion21:51
@lisitsynlike a guy that wants to do hashing can do libqp instead21:51
@lisitsynbut demo people are not that interested in our dirty ml related deals21:52
@sonney2kas I said we could change the topic from ml demos -> machine learning platform21:52
@sonney2kI mean like ML for your grandma21:52
@lisitsynsonney2k: what I am worried is that how does that fit our aim21:55
@lisitsynsonney2k: we have no idea who are our users and what do they want :)21:56
@lisitsynI mean it is nice to have but we have to come up with something that has value21:57
-!- klsv [~klsv@] has joined #shogun21:58
@sonney2klisitsyn, value for us you mean?22:01
@sonney2kor value for users?22:01
@lisitsynsonney2k: in some kind of business sense22:01
@lisitsynso both22:01
@lisitsynlike something that either is good for our users or good for us to reach something more22:01
@sonney2klisitsyn, people certainly need automated tools right?22:01
@sonney2knot everyone is an expert22:01
@sonney2kso users benefit22:02
@sonney2kit is the easy way of using shogun == totally new type of users22:02
@lisitsynyes that's true22:02
@sonney2kit also has lots of challenges so we will learn a lot22:03
@sonney2kand it is what all the big companies are doing too but22:03
@sonney2knot oss of course22:03
@lisitsynsonney2k: actually doing something not OSS in that field looks quite stupid for me now22:04
-!- rsiera [594e79ba@gateway/web/freenode/ip.] has joined #shogun22:05
@sonney2klisitsyn, no why?22:06
@sonney2kif you are yahoo, google, microsoft, ...22:06
@lisitsynsonney2k: being open source offers a lot of advantages22:06
@sonney2ksure but certainly not making money is one of them22:06
@lisitsynsonney2k: that's not true now22:07
@sonney2kwiking, does travis now build python stuff?22:07
@sonney2klisitsyn, how not?22:07
@lisitsynsonney2k: it was true in great days of oracle and etc22:07
@lisitsynwhen companies tried to earn each available cent22:07
@sonney2kwhat changed?22:08
@lisitsynsonney2k: well people are using OSS databases and pay for support22:08
@lisitsynsonney2k: I may be over-positive but model has changed (at least it is moving towards open source with support)22:10
@lisitsynsonney2k: speaking about ML - check what offers any distributed thing like say mahout22:11
@lisitsynand oracle data mining22:11
-!- rsiera [594e79ba@gateway/web/freenode/ip.] has quit [Ping timeout: 245 seconds]22:20
-!- klsv [~klsv@] has quit []22:21
-!- hoijui [] has joined #shogun22:37
gsomixsonney2k, hey, how are you? are there many students this year?23:10
-!- shogun-notifier- [] has quit [Quit: transmission timeout]23:10
-!- dvalcarce [5b757556@gateway/web/freenode/ip.] has joined #shogun23:23
-!- dvalcarce [5b757556@gateway/web/freenode/ip.] has quit [Quit: Page closed]23:34
-!- hoijui [] has quit [Ping timeout: 264 seconds]23:36
-!- dvalcarce [] has joined #shogun23:38
-!- dvalcarce [] has left #shogun []23:45
--- Log closed Wed Apr 17 00:00:28 2013