Open in new window / Try shogun cloud
--- Log opened Tue Aug 27 00:00:32 2013
-!- votjakovr [] has left #shogun ["WeeChat 0.4.0"]00:02
shogun-buildbotbuild #1900 of deb1 - libshogun is complete: Success [build successful]  Build details are at
shogun-buildbotbuild #1535 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Thoralf Klein <>00:30
shogun-buildbotbuild #1323 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>00:35
-!- pickle27 [] has left #shogun []00:50
shogun-buildbotbuild #1617 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at
-!- travis-ci [] has joined #shogun01:00
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 []01:00
-!- thoralf [] has quit [Ping timeout: 256 seconds]01:02
shogun-buildbotbuild #1324 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Thoralf Klein <>01:18
shogun-notifier-shogun: Thoralf Klein :develop * d343f68 / tests/unit/lib/
shogun-notifier-shogun: Removed "auto" type declaration from unit test.01:19
shogun-notifier-shogun: Thoralf Klein :develop * ef2a225 / tests/unit/lib/
shogun-notifier-shogun: Removed all the beautyful c++09/c++11 achievements from test.01:19
shogun-notifier-shogun: Fernando Iglesias :develop * a488c17 / tests/unit/lib/
shogun-notifier-shogun: Merge pull request #1469 from tklein23/build_hotfix01:19
shogun-notifier-shogun: Fixed sg_sparse_vector unit tests to be c++98 compliant01:19
shogun-buildbotbuild #1536 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Fernando Iglesias <>, Thoralf Klein <>01:37
-!- hushell [] has quit [Ping timeout: 264 seconds]01:44
-!- iglesiasg [] has quit [Quit: Leaving]01:58
shogun-buildbotbuild #1537 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Thoralf Klein <>02:13
shogun-buildbotbuild #1325 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Fernando Iglesias <>, Thoralf Klein <>02:16
-!- travis-ci [] has joined #shogun02:18
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:
-!- travis-ci [] has left #shogun []02:18
-!- hushell [] has joined #shogun02:25
-!- hushell [] has quit [Ping timeout: 256 seconds]02:33
shogun-buildbotbuild #1326 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Thoralf Klein <>03:00
-!- shogun-notifier- [] has quit [Quit: transmission timeout]04:19
-!- hushell [] has joined #shogun04:25
-!- pickle27 [] has joined #shogun04:47
-!- pickle27 [] has left #shogun []04:56
hushellHeikoS: Thanks for fixing the bugs for me! So the initialization lists are risky?05:28
hushellHeikoS: Sometimes I forgot hanging out on IRC, just send me an email I'll show up and debug :)05:30
-!- sonne|osx [~sonne@] has joined #shogun08:17
-!- shogun-notifier- [] has joined #shogun08:22
shogun-notifier-shogun: Soeren Sonnenburg :develop * d1be92c / examples/undocumented/libshogun/parameter_iterate_sgobject.cpp:
shogun-notifier-shogun: move n into local context to fix warning08:22
shogun-notifier-shogun: Soeren Sonnenburg :develop * 72bdc18 / src/shogun/machine/gp/EPInferenceMethod.cpp:
shogun-notifier-shogun: fix unsigned/signed warning08:24
shogun-buildbotbuild #1538 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>08:33
shogun-notifier-shogun: Soeren Sonnenburg :develop * 58498fa / src/interfaces/ (2 files):
shogun-notifier-shogun: use dim_vector(1,n) instead of deprecated (n)08:36
-!- sonne|osx [~sonne@] has quit [Quit: sonne|osx]08:36
-!- travis-ci [] has joined #shogun09:01
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 []09:01
shogun-buildbotbuild #1539 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>09:13
shogun-buildbotbuild #1327 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>09:15
shogun-buildbotbuild #1540 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>09:19
-!- travis-ci [] has joined #shogun09:28
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:
-!- travis-ci [] has left #shogun []09:28
@lisitsynwiking: yoyo09:49
-!- iglesiasg [~iglesias@2001:6b0:1:1da0:1917:e0bc:9ae3:411c] has joined #shogun09:52
-!- mode/#shogun [+o iglesiasg] by ChanServ09:52
@iglesiasggood morning!09:52
@wikingi've just created this agenda09:53
@wikinganything else we'd need to discuss:
-!- travis-ci [] has joined #shogun09:53
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 []09:53
@wikingah yeah i know09:53
@HeikoSwiking, I'll be there in 30 ....09:57
shogun-buildbotbuild #1328 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>09:58
@wikingeh what?09:59
@wikingaccording to doodle the meeting should be 10 UTC :)09:59
@wikingor you used your timezone? :)))09:59
@iglesiasg10 UTC?09:59
@wikingiglesiasg: yes (12 CEST10:00
@iglesiasgI thought it was going to start now :)10:00
@iglesiasgaaaah ok10:00
@iglesiasgmy bad10:00
@wikingiglesiasg: yeam i mean i had the same impression10:00
@HeikoSwiking, no but the doodle doesn't do 1030 I sent you all these messages yesterday to tell you10:00
@wikingbut then again doodle says 12-14 would fit everybody10:00
@wikingHeikoS: ok10:00
@wikingHeikoS: i'll schedule it from 10:3010:00
@wikingor GMT?10:01
@wikingi mean 10:30 GMT or CEST HeikoS10:01
@wikingok scheduling it for 10:30 :)10:02
@wikingheh ok it's 10:30 then10:03
@wikingsent a reminder10:03
@HeikoSI just a found a computer, can also do now if everyone is there10:03
sonne|workErrm wiking it said 10 CEST10:06
@wikingsonne|work: what did? :)10:06
@wikingdeveloper developers developers!!!10:06
sonne|workwiking: your doodle10:06
@wikingsonne|work: no it said 12-1410:06
sonne|workthat too yes10:06
@wikingcheck it out... now it only says 1210:07
-!- Heiko [5accf27c@gateway/web/freenode/ip.] has joined #shogun10:07
sonne|workTime zone: (GMT+02:00) Berlin10:07
Heikowiking: ok, here I am on an ancient machine :)10:07
sonne|workand then 10am-12pm10:07
sonne|workso we do it in 2 hours now?10:07
@wikingit's just a block of time10:08
@wikingand we do it10:08
@wikingin 20 mins10:08
Heikowiking: can also do now as said, but in 20 is also fine10:08
@wikingbut then again if everybody is here10:08
@iglesiasgI can start now too10:08
@wikingwe can just start right away10:08
@wikinglisitsyn: ?10:08
@iglesiasglisitsyn, hello?10:08
@iglesiasgwiking, the fast guy :P10:08
@wikingok he is 'brb' i guess ;)10:09
sonne|workok then in 20 mins10:09
Heikook then ill get breakfast and coffee, see you in 20 :)10:09
@wikingHeiko: :)10:10
Heikowiking: could you also add obtain_from_generic to the list?10:13
@iglesiasgand the NIPS workshop?10:13
Heikowiking: and I also have a couple of other things: modlselection and sergeys auto getter/setter10:13
Heikoand NIPS10:13
Heikoiglesiasg: yeah :)10:13
@wikingHeiko: oh yeeah nips10:13
@wikingHeiko: okok doing it10:13
@lisitsynI am here10:14
Heikoand maybe even discuss how we can get rid of old non-working code10:14
Heikowiking: see all those general things to discuss issues10:15
Heikobut maybe not all, its too much :D10:15
@wikingok i've updated10:17
@wikingbut then again10:17
@wikingwe'll start discussing10:17
@wikingand i'll do the notes10:17
@wikingand will update it after the meeting10:17
@lisitsynwho's on duty next time?10:17
@wikinglisitsyn: we'll discuss this as well during the meeting10:18
@wikingi'm just getting a coffee10:18
@wiking5 mins and i'm back10:18
@iglesiasgal right10:19
@wikingok back10:24
@wikingsonne|work iglesiasg lisitsyn Heiko shall we start?10:24
@lisitsynI am ok to start10:24
@wikinganyhow i wanted there's a 0. point where i'd be writing the most and then you can catch up10:24
@wikingso i just wanted fast review of what has happened with the last meetings points10:25
@wikingso doc-sprint application again went to south :( but again thanks a lot for Heiko to putting the applications together!10:25
@wikingcmake (we'll get back to this anyways)10:26
@wikingand unfortunately as until now i have no updates on the shogun fundation, but i guess sonne|work and Heiko would know about that10:26
@wikingand as a remaining point10:26
@wikingwho wants to organise the next meeting?10:26
@lisitsynI can10:26
@wikingok it's yours then10:26
Heikook to start10:27
sonne|workwiking: any news on the video upload?10:27
@wikingsonne|work: yeah that's still 0.10:27
@wikingit's my bad we have these delays but i promise that the first ones are going up today10:27
Heikoshogun foundation is in (slow) progress10:28
Heikosonne|work: see my email to you the other day10:28
@wikinghad to put together a python script because my internet here at my place now is really shit10:28
Heikowiking: good idea to review the old meeting points10:28
@wikingok then10:28
@wikingeverybody is here so let's kick it off10:28
Heikowiking: about the appllocation to docsprint, we will meet the people at gsoc summit and ask them what we need to do to get in :)10:28
@wikingehehh cool!10:29
@lisitsynyeah directly ask carol hah10:29
sonne|workHeiko I mean all we need is the updated letter right? Signatures is nothing you have to take care of10:29
Heikosonne|work: I need the names10:29
sonne|worklisitsyn: I did last year - not so easy...10:30
Heikoyou are right, not the signatures, so if you could send me the list with the names (for voting, satzungsupdate etc)10:30
@lisitsynsonne|work: I see10:30
sonne|workHeiko ahh ok. Let me add it to my todo10:30
@wikingah and one more thing about 0.: we are getting a new buildbot server for shogun10:30
Heikocool, what exactly?10:30
Heikojust heard its 35 euros a month10:30
@wikingsonne|work: ordered a 36euros/month server10:31
@iglesiasgyep, which one is at the end?10:31
@wikingthe cheaper option10:31
@iglesiasgaah the one of the page in German IIRC10:31
sonne|work24G core i710:31
@wikingok so first point in the agenda10:31
@wikingnext release10:31
sonne|workI guess we will have it next week10:31
@wikingwhen do we want to have it10:32
@wikingas just a vague deadline10:32
HeikoI vote for 1-2 weeks after gsoc10:32
sonne|workbut it is not 'just' a buildbot but a machine we all will have logins to10:32
sonne|workHeiko when is end of gsoc?10:32
sonne|work1 month still?10:32
@wikingHeiko: so which week?: )10:32
@iglesiasgyes, I think the end of September is the right reference date10:32
@wikingokey so let's put it around 30th of sept ;)10:32
Heikoeval deadline10:33
@iglesiasglet me check a second10:33
@wikingso basically there's a lot of things to do before we can actually release10:33
HeikoShould we maybe start having lists of things that need to be done before release?10:33
@iglesiasgso September 23rd is the last day10:33
@wikingi've tried to put together in the agenda10:33
Heikowiking: cool, where to put it? wiki?10:33
@iglesiasgmaybe we should push it a little bit10:33
@wikingso that's a reference point10:33
@wikingbut let's discuss now what is needed in more detail10:34
Heikowhat about having a seperate wiki page where we collect things needed for next release10:34
@iglesiasga vague deadline about say, no sooner than 30th and no later than 15th Oct10:34
@wikingHeiko: i'm ok with that10:34
@lisitsynI think we should just use milestone10:34
@iglesiasgthat sounds good10:34
@wikinglisitsyn: yes10:34
Heikoyeah milestone is good then we also have connected issues10:34
sonne|worklisitsyn: agreed10:34
sonne|workmilestone & issues10:34
@wikingok so instead of wiki just add issues and add it to the milestone10:34
Heikoso all the gsoc projects need to be working10:34
@wikingi'll do this10:34
sonne|workwiking: I guess it is clear that we need the buildbots to finally work10:35
@wikingi mean adding the milestone + issues10:35
@wikingsonne|work: yeah10:35
@wikingso who wants to join the mess of buildbot fixing with me?10:35
sonne|workwe still have failing tests / issues on cyg/bsd/static interfaces not tested10:35
HeikoWe can push the mentors and students on the GSoC meeting fridays that things are really ready by then, including all the little bits and pieces10:35
sonne|workwiking: well I am doing all my spare time10:35
Heikowiking: whats wrong with the buildbots currently?10:35
@wikingHeiko: we need to fix this10:35
sonne|workHeiko look at
@wikingstatic interfaces, cygwin10:35
@wikingstatic interfaces are not yet tested10:36
@wikingcmake scripts are needed for that10:36
HeikoI see10:36
@wikingbut it's more like a copy paste job10:36
@wikingwith minimal understanding10:36
sonne|workwiking: yeah that is for you10:36
Heikothat sounds like you are the best man for that :D10:36
@wikingfuck you all10:36
sonne|workwiking: we love you too10:36
@wikingok i've put myself there10:36
HeikoI can help though, would like the understand that stuff better anyway10:36
@wikingand if anybody has the need to help just join me10:36
sonne|workwiking: I tried to fix the cyg one but when running the examples locally on the cyg bot they just worked!10:36
Heikoyou have to show me how to do things probably thouggh10:36
@wikingsonne|work: fuck :(10:37
@wikingsonne|work: yeah i've read that the other day10:37
@wikingthat's totally weird10:37
@wikingbut i'll try to set an env variable or something10:37
@wikingmaybe that'll help10:37
sonne|workthe bsd1 fails one test for no reason10:37
@wikingsonne|work: i know why10:37
sonne|workthe osx1 - I am on it is easy10:37
@wikingsonne|work: some weird fbsd+gcc+gmock combination bug10:37
@wikingsonne|work: yeah how about that with gsomix?10:37
@wikingafaik it's his unit tests that are failing10:37
@wikingwho's gonna take care of that10:38
sonne|workwiking: he is busy with his non-gsoc live/job :/10:38
sonne|workwiking: fix is easy - just the usual c locale issue10:38
@wikingsonne|work: ok you take it10:38
sonne|workand we have a macro for that since we needed that for other ascii files10:38
@wikingok i've put you up for that10:38
sonne|workthen we have tons of warnings like
sonne|workwe need to get warnings down to 0 for the release again10:39
HeikoI hate these warning runs10:39
Heikowhy should prevent collecting them in the first place10:39
@wikingHeiko: we'll push the students10:39
Heikowiking: you know how we usually "fix" them?10:39
Heiko" / ** TODO */10:40
@wikingwell yeah10:40
@wikingok next bigger step10:40
sonne|workHeiko I don't know about you but I fix them properly when I do that10:40
@wikingwho's doing the merge10:40
@wikingof the git flow branches10:40
HeikoI am still for making errors out of them, then people directly document and do not push half documented code ;)10:40
@wikingfor the release10:40
sonne|workHeiko not possible - some cannot be fixed10:40
@wikingsonne|work: will u do the release this time again?10:41
sonne|workalright what else do we need for the release10:41
sonne|workwiking: you should do it10:41
@wikingi.e. merging the branch10:41
@wikingah ok10:41
@wikingi'll do the next release10:41
Heikosonne|work: the problem is that people (including) me dont even realise these warnings are there, maybe just a message or so would do it10:41
sonne|workwell look at the buildbot10:41
@wikingokey so next step before release10:41
@wikingcreate a page on our site where we will upload the nightly built packages10:41
sonne|workI do this daily and try to fix the issues...10:42
@wikingwho's taking this ?10:42
HeikoI have another suggestion for the release10:42
sonne|workwiking: copying is not the isse but the page is10:42
@wikingsonne|work: indeed10:42
Heikolast time the NEWS file was a bit chaotic, I would like to write a proper version of it, cleaned up and structured10:42
sonne|workwe would need sth similar to what we already have10:42
@wikingso who wants to do some webpage work?10:42
sonne|workok so Heiko writes NEWS10:43
@wikingdown with it10:43
@iglesiasgwiking, I am not sure what it really consists of but I guess I can do it after GSoC10:43
sonne|workthen we also need the python notebooks10:43
sonne|workI mean a page for them10:43
@wikingiglesiasg: just create some page somewhere on the website where we can put the nightly packages10:43
sonne|workwe have them autogenerated already10:43
@iglesiasgwiking, we are talking about some code so the nightly build is packaged and uploaded to website?10:43
@wikingiglesiasg: sonne|work will do that10:43
sonne|workiglesiasg: no not the upload but the list the stuff on the website10:44
@iglesiasgaaaah ok, that sounds easy10:44
Heikowe will just have 5 notebooks after gsoc10:44
@wikingsonne|work: i've put you for script writing10:44
@iglesiasgHeiko, just 5?10:44
@wikingok so ipython notebooks10:44
sonne|workiglesiasg: well it needs to be some dynamic script going through some dir listing releases etc10:44
@wikingHeiko: doesn't matter how much10:44
sonne|workwiking: yeah one liner10:44
Heikolater we can aim for something like
@wikingwe need to discuss where are the ipython notebooks are going to reside10:45
Heikoiglesiasg: ah sorry 810:45
sonne|workwell I would do it the same way we do docs now10:45
sonne|workwe have one doc per stable release10:45
@wikingsonne|work: you have the ipythons autogened nightly?10:45
sonne|workso 2.1.0  etc10:45
sonne|workand two symlinks10:45
sonne|workone pointing to the current release10:45
-!- Heiko [5accf27c@gateway/web/freenode/ip.] has quit [Quit: Page closed]10:46
sonne|workand one to the latest (nightly)10:46
@wikingah about the ipython notebooks one more thing10:46
@wikingnightlies should be diffed10:46
-!- Heiko_reload [5accf27c@gateway/web/freenode/ip.] has joined #shogun10:46
@wikingi've seen a script like this10:46
sonne|workwiking: why diff?10:46
@iglesiasgsonne|work, that makes sense, doing it as the docs10:46
@wikingwhere the notebook is diffed to a 'stable' version10:46
sonne|workwiking: I mean if we change a script it is on purpose no?10:46
@wikingyes but in case of nightlies10:47
@wikingit can been that something got broken10:47
@wikingin libshogun10:47
@iglesiasgwiking, anyway, write me to do that and we will work out the details in another moment. I am in for it *after* GSoC, so in a month10:47
sonne|workI mean we don't want tests for notebooks like we have for examples or do we?10:47
@wikingiglesiasg: ok! :)10:47
sonne|workwiking: well unit & integration tests should detect that10:47
@wikingsonne|work: but what if ;)10:47
Heiko_reloadsonne|work: its possible but not really needed10:47
@wikingthen it's off the list10:47
@wikingeasy peasy10:47
@wikingso who's doing the webpage part of this?10:47
Heiko_reloadjust test them for API consistency10:48
@wikingiglesiasg: you again? :)10:48
Heiko_reloadand people can put asserts into them10:48
@iglesiasgwiking, I got lost what is the second thing :D10:48
@wikingas we need to put it somewhere on our webpage...i.e. integrate it nicely etc10:48
sonne|workno asserts in notebooks please10:48
@iglesiasgaaah notebook10:48
@wikingiglesiasg: well notebooks are generated :)10:48
sonne|workiglesiasg: webpage is missing10:48
@wikingiglesiasg: they need to be integrated somehow into our webpage10:48
@wikingiglesiasg: i can help out10:49
@iglesiasgwiking, all right, sure10:49
@wikingiglesiasg: so i'll put both of us10:49
Heiko_reloadiglesiasg: like this, and one notebook per entry :)10:49
sonne|workfor the release I would also like to have our live demos integrated in the webpage10:49
@wikingokok we are over this10:49
@iglesiasgHeiko_reload, yeah that was my idea :) check how they were doing it10:49
sonne|workwiking: live demos for the release10:49
@wikingsonne|work: ok who wants to do that? :)10:49
Heiko_reloadiglesiasg: man if we had this, all features of shogun with a notebook, how cool would that be ? :)10:49
Heiko_reloadsonne|work: what is the state of the live demos?10:50
@iglesiasgtoo cool for school lol10:50
@wikingsonne|work: can we use the gsoc student? :D10:50
Heiko_reloadbtw do we plan that every student does a live demo?10:50
@wikingHeiko_reload: did u have it scheduled?10:50
@wikingi mean in the initial gsoc plan of each student?10:50
Heiko_reloadwiking: not yet, but we can do this10:50
sonne|workwiking: no this is sth we have to do10:50
@wikingsonne|work: okey10:50
Heiko_reloadno the notebook came in afterwards10:50
sonne|workwiking: he did the demos we have to figure out how we want things10:50
@wikingso live demos integration into the webpage10:51
Heiko_reloadhow much work is it to create a demo for the project?10:51
sonne|workHeiko_reload: which project?10:51
@wikingso who wants to join me for that?10:51
Heiko_reloadgsoc project10:51
sonne|workHeiko_reload: which gsoc project?10:51
Heiko_reloadhow mnuch work to create a new demo?10:51
@wikingHeiko_reload: as much as a notebook imo10:51
@wikingyou need some nice data etc10:51
Heiko_reloadlike say we ask gsoc students to create a demo for their project , how much work would that be?10:52
sonne|workwell only certain things are supported10:52
sonne|worklike loading data from some file / display a curve / heatmap10:52
Heiko_reloaddo you guys think we should do a demo for every student?10:52
sonne|workbut if you follow this then it is like notebook so rather easy10:52
Heiko_reloadmandatory or not?10:52
@iglesiasgHeiko_reload,  I think it should be doable10:52
@wikingif it's doable10:53
sonne|workHeiko_reload: well we need more - we want it for all algorithms in shogun10:53
@wikingi mean we cannot ask something that wasn't in the plan really :P10:53
sonne|workHeiko_reload: but yes for gsoc presentation it would be very nice so yes10:53
Heiko_reloaddoes it even make sense for all projects?10:53
@wikingHeiko_reload: well yes it does :)10:53
@wikingit makes sense for anything10:53
@wikingit's rather just a showoff10:53
@lisitsynnot for all I guess10:53
@wikinghow you can do shitz :)10:53
@lisitsynwhat would you show in IO (gsomix) project :)10:54
sonne|workshow off is important10:54
@wikinglisitsyn: loading data from web? :)10:54
Heiko_reloadlamday is also hard to show in a demo10:54
Heiko_reloadsince its just an estimator :)10:54
@wikinglisitsyn: load a feature matrix via json :P10:54
sonne|worklisitsyn: not a live demo sure10:54
Heiko_reloadmaybe plot convergence :D10:54
Heiko_reloadhow many demos do we have currently?10:54
@wiking404 :)10:55
@iglesiasg404 here too hehe10:55
sonne|workwell you have to choose one of the urls10:55
@iglesiasgbut the list is there10:55
@wikingyeah i've got it :)10:55
@wikingnoworkie for me10:56
sonne|workohh that is missing10:56
sonne|workmulticlass is left over stuff10:56
@wikingbut yeah the page looks awesome ;)10:56
Heiko_reloadgp also doesnt work10:56
sonne|workit did last time I checked10:56
sonne|workworks here...10:57
Heiko_reloadah entrance10:57
@wikingdo we hve some kind of an automated test to check10:57
Heiko_reloadwe have to have them on the website for release, just too nice10:57
@wikingwhether it's broken or not?10:57
sonne|workwiking: no10:58
sonne|workno idea how to do that10:58
@wikingcan we have one?: )10:58
@wikingi'll look into it10:58
@iglesiasgthe demos are pretty awesome10:58
@iglesiasgthese guys visible somewhere in the webpage would be a killer feature IMHO10:58
sonne|workMy dream would be that we had these demos when you click on the images at the bottom10:58
@wikingso i've put down some stuff fo rthe demos10:58
sonne|workso on our current page10:58
@iglesiasgI like very much that idea10:59
sonne|workthe screenshots at the bottom10:59
@iglesiasgyes yes10:59
sonne|workwhen you click on this some overlay opens10:59
sonne|workand you see the demo10:59
Heiko_reloadsonne|work: how much work is it to tune the demos, for example for GP, some parameters are missing etc, is it easy to change that?10:59
sonne|workmaybe with a little code snippet showing the real code or the link to a python notebook or so for further information10:59
@iglesiasgthe overlay could be like lisitsyn did in tapkee10:59
sonne|workHeiko_reload: yes parameters are a dictionary10:59
@iglesiasgoepn any of the graphical examples of usage examples11:00
@lisitsynwhat overlay?11:00
Heiko_reloadsonne|work: cool!11:00
sonne|workyeah like this11:00
@lisitsynthat's easy with bootstrap11:00
Heiko_reloadsonne|work: could you tell foulwall to do a heatmap for GP regression? like the one in the notebook I did for the workshop?11:01
@wikingso we need a complete redesign of the webpage i see that :D11:01
Heiko_reloadhaha :D11:01
@wikingbut yeah anyhow11:01
Heiko_reloadpoor iglesiasg11:01
@wikingwe'll need to do an integration of demos11:01
@wikingalong with nightly packages and ipythons11:02
sonne|workno it is not a complete redesign11:02
@wikingso it's quite some work on the webpage part11:02
sonne|workshouldn't be too hard11:02
@wikingyeah but working time11:02
@wikingi've put down iglesiasg and myself for that11:02
@iglesiasgHeiko_reload, well I have here with me until November the guy who did the webpage. He should be able to back me up if I ask him :)11:02
@wikingbut i'll do issues for all these11:02
@wikingand then anybody can join as well11:02
sonne|workwiking: ok so I guess the release plan is clear?11:03
@wikingnot yet11:03
@wikingfix CPACK scripts for creating tar.gz and .deb, .rpm and .mpkg packages11:03
@wikingi know i know11:03
@wikingi have to do it11:03
sonne|workwiking: yes11:03
sonne|workso clear :P11:03
@wikingbut i need access to redhat/fedora machine11:03
@wikingi dont have that11:03
@wikingmaybe we the new machine?11:03
sonne|workwiking: well install one on the new machine in a chroot11:03
@wikingsonne|work: ok then i'll w8 for that one11:03
@wikingi guess we have plently of stuff to do till the next release :)11:04
@wikingso afaik we can get to the point in the agenda11:04
sonne|workyeah I guess it is 1-2months work...11:04
@wikingi would go directly to11:04
@wikingApplication to NIPS workshop11:04
@wikingas it has a priority11:04
sonne|workWho wants to go to NIPS?11:05
@wikingas most of the things we discussed for the release11:05
@wikinglike demo + ipython notebooks11:05
@wikingwould be required for the NIPS workshop presentation11:05
@iglesiasgbefore going we should prepare something and get it accepted, right?11:05
@wiking2 things11:05
@wiking1) when is the deadline in the application11:05
@wiking2) when is nips :)11:05
@iglesiasglike a paper I guess11:05
@lisitsynwiking: 9 oct iirc11:05
sonne|workwell if   no one wants to go we don't have to right?11:05
Heiko_reloadso application deadline is early oct11:05
@lisitsynwiking: nips is early december11:06
@wikingHeiko_reload: good11:06
@lisitsyn10-12 or so11:06
Heiko_reloadcall for bla is not out yet11:06
@wikingthen we should have already the next release11:06
Heiko_reloadI will be there very likely anyway11:06
@wikingor at least  the demos fixed up wiht the notebooks11:06
Heiko_reloadcheng said we should submit something11:06
Heiko_reloadso since the workshop focus is on integrated software11:06
@wikingthen we can use all that material11:06
@wikingin the application11:06
Heiko_reloadwe can focus on ipython, demos, file formats11:06
Heiko_reloadand then connect them with backends: libsvm for example11:07
@wikingi would go with shogun itself11:07
Heiko_reloadyes true11:07
@wikingipython+demos are just a great way11:07
sonne|workHeiko_reload: so you will go?11:07
@wikingto show off11:07
sonne|workand give the talk?11:07
Heiko_reloadBut focus on shogun connects many existing open-source software tools11:07
@wikingHeiko_reload: will you take care of the application. organizing etc?11:07
Heiko_reloadsure I can do that :)11:07
Heiko_reloadwiking: I will write a draft11:07
@wikingHeiko_reload: great11:08
Heiko_reloadwe have to have the release before that so that I can rely on things :)11:08
@wikingok so who wants to go11:08
Heiko_reloadas said, I will be around anyway11:08
@lisitsynI am off11:08
@wikingi dont have any more funds11:08
@wikingHeiko_reload: good11:08
sonne|workok so heiko goes does the work and gives the talk11:08
Heiko_reloadsonne|work: are you going to nips?11:08
sonne|workHeiko_reload: your paper got accepted?11:08
sonne|workHeiko_reload: no11:08
Heiko_reloadIll do a draft, you gotta help me with that!11:08
@iglesiasgI don't think I can get funding as a master's student to get there :S11:08
@iglesiasgbut I would love going11:08
Heiko_reloadnever presented software to conferences11:08
Heiko_reloadiglesiasg: you can apply11:09
Heiko_reloadiglesiasg: they do give funds to students11:09
@wikinglisitsyn: dontgo :(11:09
@lisitsynHeiko_reload: I will run a daemon in my mind to think what should we present11:09
sonne|workHeiko_reload: well it is just show off time11:09
@lisitsynwiking: what?11:09
Heiko_reloadlisitsyn: haha :)11:09
Heiko_reloadsonne|work: ok11:09
@wiking11:08 <@lisitsyn> I am off11:09
sonne|workwe should next year have someone go to scipy or so11:09
@lisitsynwiking: ahh no I mean11:09
Heiko_reloadsonne|work: btw the ipython guy will be there, fernando perez or so11:09
@lisitsynwiking: I won't be able to go11:09
sonne|workor europython11:09
-!- lambday [67157e4c@gateway/web/freenode/ip.] has joined #shogun11:09
@wikinglisitsyn: ah ok11:09
-!- naywhayare [] has quit [Read error: Operation timed out]11:09
Heiko_reloadhe will like how we (ab)use his notebooks :)11:09
@lisitsynwiking: otherwise get fired  :D11:09
@wikingHeiko_reload: nips gives out funding for travel?11:09
Heiko_reloadwiking: yes11:10
@wikingHeiko_reload: ok i can join then if i get one :)11:10
Heiko_reloadsonne|work: they do11:10
@wikinganyhow we'll see11:10
sonne|workHeiko_reload: well not for a workshop contribution11:10
@wikingHeiko_reload: you just lead this one11:10
sonne|workbut yes I've received NIPs travel funding a couple of times11:10
@wikingsonne|work: not me :)11:10
@lisitsynit is a good chance to have a 'new' shogun paper11:10
@lisitsynwith all the people joined recently11:10
Heiko_reloadlike to meet the guys anyway11:10
sonne|workHeiko_reload: if you have a nips paper and a workshop contribution you would get it11:10
Heiko_reloadlisitsyn: no there are no proceedings11:10
@wikinglisitsyn: good idea ! :)11:11
@lisitsynHeiko_reload: ohh bad11:11
@lisitsynI though there are11:11
@wikinglisitsyn: webpage proceedings ;)11:11
Heiko_reloadnot for this one, it sais on the webpage11:11
@lisitsyntoo bad11:11
Heiko_reloadanyway, we will put the video up11:11
sonne|worklisitsyn, Heiko_reload we should rather write a new JMLR paper about shogun then11:11
Heiko_reloadoh I have more thing for the release11:11
Heiko_reloadthe README11:11
Heiko_reloadis ooooold11:11
@wikingHeiko_reload: who's doing that? :)11:11
@lisitsynsonne|work: reviewers wouldn't like it11:11
sonne|workand the INSTALL files11:11
sonne|worklisitsyn: they would11:11
Heiko_reloadbut README is first impression11:12
@lisitsynsonne|work: why? we haven't changed much11:12
Heiko_reloadtalks about svms mainly11:12
Heiko_reloadbut we can do so much more by now11:12
sonne|worklisitsyn: just everything11:12
Heiko_reloadalso on the webpage there is an old text11:12
Heiko_reloadshould be updated before release11:12
@iglesiasgHeiko_reload, which one? URL?11:12
@lisitsynsonne|work: no we are still interfaces + SVM + one-core computing11:12
@lisitsynnothing great11:12
@wikinglisitsyn: GP?11:12
sonne|workit is the usual lisitsyn whine11:13
-!- naywhayare [] has joined #shogun11:13
@wikingi mean there's really some more shitz in shogun now then just simple svms11:13
@lisitsynwiking: doesn't sound like something great ;)11:13
@lisitsynmore shit is not the other shit11:13
@lisitsynif we went distributed or GPU or whatever11:13
@wikinglisitsyn: as well as fancy structural learning etc11:13
Heiko_reloadiglesiasg: this text is not really a good overview anymore11:13
@lisitsynit would be a change11:13
Heiko_reloadwould prefer some less technical overview and then a detailed features page with references etc11:13
@wikingi would suggest that i put up issues for that11:14
@lisitsynI have some idea for workshop Heiko_reload11:14
@wikingand put it into the milestone11:14
@wikingand whoever grabs it11:14
@wikingdoes it11:14
Heiko_reloadlisitsyn: lets talk later11:14
Heiko_reloadlets press on with the meeting, I gotta go soon11:14
@wikinglisitsyn: i would start a wiki or issue for that11:14
Heiko_reloadlisitsyn: Ill be back this afternoon11:14
@wikingand let's discuss it there or?11:14
@wikingnext poing11:14
@lisitsynplease add issue11:14
@wikingSerialisation framework11:14
@wikingso we have this nice new feature branch11:15
@lisitsynoh really11:15
Heiko_reloadsonne|work: why did you close the issue on json?11:15
sonne|workwe are talking shogun 0.10.0 vs shogun 3.0 you know that right?11:15
Heiko_reloadjson breaks when there is inf11:15
Heiko_reloadwhich is a bug11:15
sonne|workHeiko_reload: there is no bug11:15
sonne|workHeiko_reload: in json you cannot store inf / nan11:15
Heiko_reloadyes there is, shogun crashes on save and then load11:15
Heiko_reloadwell its written to the file by shogun11:15
Heiko_reloadwe cannot write it and then say on loading "its not supported"11:16
sonne|worktrue then we should just fail writing11:16
@wikingHeiko_reload: so i would just change the save part11:16
Heiko_reloadyeah thats fine, thats why I created the issue11:16
@wikingreopen it!11:16
Heiko_reloadits the last error before the automated serialization tests work11:16
@wikingand do the fixing there :)))11:16
Heiko_reloadall other things work11:16
@wikingHeiko_reload: what was the name of the issue?11:16
@wikingi mean number11:16
Heiko_reloadwhat about this idea of dropping json?11:17
Heiko_reloadforgot, ill check11:17
@wikingso yeah once we've got the travis going 100% of the SerialUTests branch we'll merge it11:17
@wikingand according to Heiko_reload it's really only that one issue with json11:17
Heiko_reloadwiking: ah and we have to update 10 integration tests or so11:17
@wikingno i think it's not this one11:17
Heiko_reloadsince I had to remove spaces from parameter names11:18
Heiko_reloadwe now allow "_", "." and alnum11:18
@wikingHeiko_reload: ah no worries11:18
@wikingHeiko_reload: that'll be done by sonne|work  ;)11:18
Heiko_reloadoh and the only reliable serialization is hdf511:18
@wikingHeiko_reload: as soon as we get travis running we are good11:18
@wikingto merge...11:18
-!- van51 [] has joined #shogun11:19
Heiko_reloadall others only work up to 1e-6 or 1e-811:19
@wikingyeah that was expected11:19
@wikingok so we've talked with sonne|work11:19
@wikingand we'll keep json11:19
@wikingjust add new one11:19
Heiko_reloadok then11:19
@wikinglike msgpack11:19
Heiko_reloadwhat do we do about inf11:19
Heiko_reloadsince some classes contain it11:19
@wikingthat's a more precise JSON like11:19
@wikingok it was this issue
@wikingreopened it11:20
@wikingbasically either we fail at writing11:20
sonne|workHeiko_reload: well fail writing11:20
Heiko_reloadand then?11:20
@wikingor just put there MAX_DOUBLE11:20
@wikingor something11:20
Heiko_reloadno serialization?11:20
Heiko_reloadno max double11:20
sonne|workyes fail11:20
@wikingi mean it's pretty stupid to have a serialization framework11:20
@wikingthat only works partly11:20
@wikingi mean not framework11:21
@wikingbut output11:21
@wikingin this case just lets drop json11:21
Heiko_reloadfine for me11:21
@wikingand use something that supports all the things we need11:21
Heiko_reloadleaks also btw11:21
@wikinglike inf and NaN11:21
Heiko_reloadas some of the other writers do11:21
@wikingbecause to have now a half working serialisation writer11:21
@wikingit's just no use11:21
@wikingwe open here pandora's box11:21
sonne|workwell people use json for serialization11:22
Heiko_reloadI am fine with dropping if it cannot support inf11:22
sonne|workI am against dropping11:22
@wikingok let's do a vote11:22
@wikingiglesiasg: lisitsyn11:22
Heiko_reloadI dont really care to be honest11:22
sonne|workit is an easy way to transfer from js11:22
sonne|workerr to js and de-serialize11:22
Heiko_reloadbut what about the inf then?11:22
@lisitsynHeiko_reload: +11:22
sonne|workNO INFs11:23
@wikingok so we have nocares11:23
sonne|workI don't have SVMs with INFs11:23
@wikingbut that doesn't mean that other parts of shogun doesn't have that either11:23
Heiko_reloadso a users starts using json for his stuff11:23
@wikingi mean we just cannot go11:23
Heiko_reloadputs in some work11:23
@wikingworksforme wit this11:23
Heiko_reloadand then discovers that he cannot use it since there are infs11:23
Heiko_reloadthat is annoying11:24
@wikingHeiko_reload: +11:24
sonne|workthen don't use json11:24
sonne|workit is that simple11:24
sonne|workif you want a non-lossy format use that one11:24
sonne|workand use that by default11:24
@wikingok so let's say we fix json serialization11:24
@wikingby not supporting NAN and INF11:24
@wikingand just die with it11:24
Heiko_reloadwell ok, i still dont like that though11:24
@wikingand add a more mature serialization backend11:24
@wikinglike msgpack11:24
Heiko_reloadwiking: can you blacklist the inf classes?11:24
@iglesiasgwe should just fix json standard, easy :D11:24
Heiko_reloadfine with me11:24
@wikingiglesiasg: but we cannot as json doesn't support NaN and INF11:25
sonne|workI mean it is like saying python should not support json because it cannot encode inf11:25
@wikingso there's really no way around this11:25
@iglesiasgwiking, that's why, fix it so it supports it :D jk11:25
@wikingiglesiasg: aaah json itself11:25
@wikingok i get it11:25
@wikingyeah that's a good task for another month :P11:25
Heiko_reloadjust throw an error11:25
Heiko_reloadon writing inf11:25
@wikingsonne|work: ok let's just fix then by SG_ERROR if inf/nan comes11:25
@wikingand that's it11:25
Heiko_reloadsince currently it DOES write inf11:26
Heiko_reloadto the file11:26
@wikingok cool11:26
@wikingwe fix it by exit(1) :)11:26
Heiko_reloadand then blacklist the classes11:26
Heiko_reloadplease do a proper error message :)11:26
@wikingHeiko_reload: blacklist? why?11:26
@iglesiasgerror message yeah11:26
@iglesiasgexit(1) lol11:26
@wikingi mean it should be just SG_ERROR and that's it11:26
Heiko_reloadwiking: since some classes contain inf by default11:26
@wikingHeiko_reload: that's ok11:26
Heiko_reloadthen we get an error in automated tests11:26
@wikingHeiko_reload: if JSON sees INF11:26
@wikingah fuck11:26
@wikingi dont waaaaaant11:27
@wikingdrop this fucking json11:27
Heiko_reloadthats my point11:27
sonne|workdrop the automated tests11:27
@wikingsonne|work: really? :D11:27
Heiko_reloadhaving serialization that only support half of the classes is weird11:27
Heiko_reloadno, these tests are very helpful11:27
Heiko_reloadI found 100 bugs through them11:27
sonne|workHeiko_reload: live is weird11:27
Heiko_reloadI am for blacklisting11:27
@wikingi mean yeah11:27
sonne|worksomeone introduced floats11:27
@wikingHeiko_reload: you write it11:27
sonne|worksome people use ascii to store floats11:27
Heiko_reloador maybe dont allow inf in default initialization?11:28
@wikingi dont care11:28
sonne|workand we still use floats11:28
Heiko_reloadalso not good11:28
@wikingjust fix 146411:28
Heiko_reloadwhat about11:28
Heiko_reloadignoreing the parameter :D11:28
@wikinghahah no fucking way11:28
@wikingthat's not serialization11:28
Heiko_reloadno idea11:28
@iglesiasgit would be nice if we can have a way that a problem exits with an error and it can be controlled from a test11:28
Heiko_reloadblacklisting seems the only way11:28
@iglesiasgso the test doesn't break11:28
@wikingwhat is the point of JSON serialization11:28
@wikingthere's no way you can deserialize a class11:28
@iglesiasgproblem -> program11:28
@wikingw/o shogun11:28
@wikingi mean lets say i deserialize an SVM11:29
sonne|worknext topic please11:29
@wikingwhat could i do with that in .js ?11:29
@wikingsonne|work: seriously11:29
@wikingwhat you do with a JSONed SVM11:29
@wikingw/o shougn11:29
sonne|workwiking: run it in .js11:29
@wikingsonne|work: it will auto deserialize it?11:29
@wikinginto what?11:29
sonne|workwith code I write?11:29
@wikingsonne|work: for .js ?11:29
sonne|workjs does this for me11:30
@wikingi mean i dont see how u use a serialized shogun class in .sj11:30
sonne|workI have the whole object already11:30
@wikingso you just do deserialize(JSONedSVM) in .js11:30
Heiko_reloadwiking: blacklisting inf classes11:30
@wikingand you have it there11:30
* iglesiasg brb11:30
@wikingand you can run that svm ?11:30
Heiko_reloadits the best choice in this situation11:30
@wikingif that's so it's cool11:31
Heiko_reload1.) throw error upon writing inf11:31
Heiko_reload2.) blacklist json serialization classes11:31
Heiko_reload3.) be happy again11:31
@wikingi mean i dont see how it's possible to run a serialized SVM w/o libshogun11:31
Heiko_reload4.) continue with meeting since I have to go in a few mins :)11:31
@wikingbut it's maybe just my lack of knowledge...11:31
sonne|workyeah continue11:31
@wikingnono pelase11:32
@wikingi want to know this11:32
Heiko_reloador even finish here?11:32
Heiko_reloadalready enough, we can have anothoer meeting in a few weeks11:32
@wikingauto generate obtain_from_generic function for all classes11:32
Heiko_reloadah yes11:32
@wikingi'll do this11:32
@wikingso the conclusion on the end11:32
Heiko_reload+1 for the global functions11:32
Heiko_reloadand optional with ifdef11:32
Heiko_reloadso that we dont need jinja11:32
@wikingdo we have a separate .h+.cpp for this right?11:32
@wikingand there just do the ifdef11:33
Heiko_reloadwiking: yes I think thats best11:33
@wikingok and it'll go into shogun/base11:33
@wikingok done11:33
Heiko_reloadthe things are for modular interfaces anyway11:33
* iglesiasg back11:33
@wikingokok i'll do that one11:33
@wikingok next one11:33
@wikingSergey's idea about auto setter/getter11:33
@wikingso lisitsyn go ahead11:33
Heiko_reloadlisitsyn: told me he would write a feature branch with that :)11:33
@lisitsynyeah I'd rather try11:34
@lisitsynand check what happens11:34
@lisitsynhave all of you seen the issue?11:34
@wikinglisitsyn: vaguely11:34
@wikingi have no fucking clue11:34
@wikingwhat it's all about11:34
@lisitsynwiking: okay so idea11:34
@lisitsynis to stop writing get/set methods11:34
@wikinglisitsyn: + for that one (see c#) :D11:34
@lisitsynI came with some solution11:34
@lisitsyninstead of adding getter11:35
@lisitsynwe add some say11:35
Heiko_reloadwiking: first, it would be nice to have getter/setter for all parameters since the code is cleaner11:35
@lisitsynKeyword<float> my_parameter;11:35
@lisitsynin sgobject we would have11:35
Heiko_reloadsecond, it would be nice if one could rely on those, for example in model-selection frameowkr11:35
@lisitsynfloat get(Keyword<float> kw);11:35
@lisitsynand the usage is11:35
Heiko_reloadlisitsyn: important is that users devs should be able to overload setters to add checks11:35
@wikingbtw: Modern C++ guidelines hold that your class shouldn't have getters and setters. In fact, setters are almost completely useless, and getters have only very limited usefulness.11:36
@wikingIf you want to change the state of a class instance, there are two ways to do this:11:36
@lisitsynyes I keep that in mind11:36
@wikingYou perform an action on it that mutates its state. You create a new instance.11:36
sonne|worklisitsyn: better send a link to the issue where you explained it11:36
@lisitsynsonne|work: haha yes11:36
-!- shogun-notifier- [] has quit [Quit: transmission timeout]11:36
@lisitsynthat's the issue11:36
@lisitsynwiking: yes but we should have them11:36
sonne|workwiking: well you know ML methods have parameters one needs to specify11:36
@wikingyeah of course11:37
sonne|workapart from that yes getters/setters should be avoided to the extend possible11:37
@wikingi was just copy-paste :)11:37
@wikingi would say lets go with an issue + feature branch11:37
@wikingand let's discuss experiment there11:37
@lisitsynokay I have to leave in a few minutes11:37
@lisitsyndo we have anything else?11:37
@wikinglast big one11:37
@wikingIntegration tests11:38
@wikingthis is an ongoing one11:38
@lisitsynyeah I wouldn't mind if we dropped them11:38
Heiko_reloadsorry I have to go now11:38
@wikinghehehe ok11:38
Heiko_reloadwe have to discuss that later11:38
@wikingthen let's say we have all discussed11:38
Heiko_reloadand until then, sonne|work has to fix broken tests :)11:38
@wikingintegration + migration11:38
@wikingbut yeah11:38
Heiko_reloadwiking: next time11:39
sonne|workHeiko_reload: well you fix the broken tests11:39
@wikingsonne|work: will do the broken tests11:39
sonne|workI didn't break them11:39
@wikingsonne|work: but u can fix them :)11:39
Heiko_reloadi didnt break them11:39
Heiko_reloadi just changed parameter names11:39
@wikingok so i'll write the transcript of this meeting today11:39
Heiko_reloadthat shouldnt break anything11:39
sonne|workwhich breaks tests11:39
Heiko_reloadanyway I gotta run now11:39
sonne|workof course11:39
@wikingand start putting up issues + milestone for the next release11:39
sonne|worksince serialization will no longer work when you do this11:39
@wikingthere's like fucking lot of work here to do11:39
@wikingfor the next 1 month11:40
@wikingbut it's going to be a good release11:40
@wikingif we make all of this11:40
@wikingthanks all to be here on such short notice11:40
@wikingthe next meeting is going to be organized by lisitsyn11:40
@lisitsynI'll set up a doodle11:40
@lisitsynI guess middle of sept?11:41
@wikingi think we should schedule it somewhere mid sept?11:41
@wikingsame idea i head11:41
@wikingok i'm off to get my coffee11:41
@wikingand then i'll start writing issues :))011:41
@wikingone more thing11:42
@wikingwhen are we bumping new major version11:42
@wikinglike when we have d-pointers? :)11:42
sonne|workwiking: for the next release already11:42
sonne|worktoo many changes11:42
sonne|workbuild system completely new11:42
@wikingsonne|work: so we go 3.0  now?11:43
@wikingah so NEWS is at the moment misleading11:43
@wikingso in september we'll have shogun 3.011:43
@wikingnice one11:43
sonne|workwell I didn't expect it to be that much11:43
@wikingfor linux it took like fucking 15 years to do this11:43
-!- Heiko_reload [5accf27c@gateway/web/freenode/ip.] has quit [Ping timeout: 250 seconds]11:44
sonne|workshogun is 14 years old...11:45
-!- lambday [67157e4c@gateway/web/freenode/ip.] has quit []11:59
shogun-buildbotbuild #1623 of deb3 - modular_interfaces is complete: Failure [failed compile csharp_modular]  Build details are at  blamelist: Soeren Sonnenburg <>12:03
-!- van51 [] has quit [Quit: Leaving.]12:30
-!- thoralf [] has joined #shogun12:41
thoralfwiking: Around?13:07
thoralfwiking: I compiled on a new system and cmake did not detect that gtest is not available.13:07
thoralfTook quite a long time to compile all the stuff just to see it fail with "fatal error: gtest/gtest.h: No such file or directory"13:10
@wikingthoralf: and?13:21
@wikingeither give me a full output or debug it13:21
thoralfwiking: What is the expected behaviour?13:22
@wikingthoralf: -DENABLE_TESTING=ON was given to the cmake?13:23
thoralfNo gtest/gmock in the cmake output.13:24
@wikingwell cmake should not put anything about gtest/gmock in it's output13:24
@wikingonce you start make13:24
@wikingthen it should start doing shit with gmock/gtest13:24
thoralfsimply typed "make shogun-unit-test" in the cmake directory.  I was only surprised that it failed that late... after 10 minutes compiling other stuff.13:39
thoralfwiking: Which files are expected in third_party so that cmake detects google mock automatically?13:39
thoralfwiking: RTFM me if you like, but I didn't find it in the documentation. ;)13:40
@wikingthoralf: third_party/GoogleMock should have the svn checked out14:13
@wikingby cmake14:13
thoralfwiking: Only after running make GoogleMock manually, but indeed, this worked fine.14:13
@wikingmmm interesting14:14
@wikingi'll try it locally14:14
-!- HeikoS1 [] has joined #shogun14:19
HeikoS1wiking: pls update the wiki with the meeting summary :)14:24
HeikoS1irc logs would be good, you can generate a link to the time when the meeting started then its easy to re-cap14:25
thoralfwiking: Small PR for you, no need to wait for travis:
thoralfwiking: Hoping that people will start improving it. :)14:26
thoralfiglesiasg: Thanks. ;)14:32
@iglesiasgthoralf, ;) thanks to you14:33
-!- iglesiasg_ [] has joined #shogun15:05
-!- iglesiasg [~iglesias@2001:6b0:1:1da0:1917:e0bc:9ae3:411c] has quit [Ping timeout: 240 seconds]15:05
-!- sonne|osx [~sonne@] has joined #shogun15:24
-!- shogun-notifier- [] has joined #shogun15:25
shogun-notifier-shogun: Thoralf Klein :develop * 23e0864 / README.cmake:
shogun-notifier-shogun: Started a small readme how to use cmake build system.15:25
shogun-notifier-shogun: Thoralf Klein :develop * 1c16681 / README.cmake:
shogun-notifier-shogun: Added more information to README.cmake - how to get a list of possible interfaces.15:25
shogun-notifier-shogun: Soeren Sonnenburg :develop * 758b542 / README.cmake:
shogun-notifier-shogun: Merge pull request #1472 from tklein23/cmake_readme15:25
shogun-notifier-shogun: Started a small readme how to use cmake build system.15:25
HeikoS1wiking: around?15:26
sonne|osxthoralf: btw do you want to take part / join the next developer meeting?15:33
sonne|osxthoralf: you are currently rather active so you would be warmly welcomed :)15:33
* sonne|osx off15:34
-!- sonne|osx [~sonne@] has quit [Quit: sonne|osx]15:34
-!- votjakovr [] has joined #shogun15:36
shogun-buildbotbuild #1541 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Thoralf Klein <>15:38
thoralfsonney2k:  Thanks for inviting.  Yeah, I'd like to... depends on place/time.15:38
HeikoS1thoralf: we try to do them every few weeks15:40
HeikoS1thoralf: see our wiki page for summaries on the first two ones15:40
thoralfHeikoS1: Can you tell me the link?15:41
HeikoS1pages->shogun dev meetings15:41
HeikoS1thoralf: you will see that we just started15:43
thoralfHeikoS1: Yes.15:43
thoralfHeikoS1: Maybe add a link on the wiki landing page?15:43
thoralfHeikoS1: I thought the wiki was empty. ;)15:43
shogun-buildbotbuild #1542 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>15:43
HeikoS1its only important for devs and they know :)15:43
thoralfHeikoS1: The next meeting is or was today?  When?15:44
HeikoS1this morning15:45
HeikoS1next one will be in a few weeks15:45
HeikoS1there is also the gsoc meeting on friday where everyone might be around15:45
HeikoS1thoralf: we will let you know before the next one15:47
HeikoS1but careful, there always lots of things to do :D15:47
thoralfHeikoS1: Thanks, please do.15:48
HeikoS1thoralf: suggestions are also very welcome15:48
HeikoS1like your hacking Tuesday :)15:48
thoralfHeikoS1: Don't mind - working people don't scare me ;)15:48
votjakovrHeikoS1: hi! Please, have a look at inheritance CGradientCriterion <- CEvaluation. If i understand right CGradientCriterion is used just for tell us about direction. So why is it inherited from CEvaluation?15:53
HeikoS1votjakovr: that is because the model selection framework is based on that interface15:53
HeikoS1votjakovr: have a look into gradient model selection15:56
HeikoS1in there, parameters are applied to the machine15:56
HeikoS1and then CMachineEvaluation interface is used to evaluate these parameters15:57
HeikoS1this "evaluation" happens to be the gradient in this case15:57
HeikoS1its a bit abstract but a nice representation in fact15:57
HeikoS1votjakovr: evaluating a machine can return some arbritary things, accuracy, ROC, or even the gradient15:58
HeikoS1everything that fits into CEvaluation15:58
HeikoS1ehm CEvaluationResult15:58
-!- thoralf [] has quit [Quit: Konversation terminated!]15:58
-!- thoralf [] has joined #shogun15:58
votjakovrHeikoS1: yeah we have GradientEvaluation16:01
votjakovrHeikoS1: but i thought, that GradientCriterion is just helper class to know the direction16:01
HeikoS1votjakovr: but it is16:02
HeikoS1votjakovr: i dont get the question then16:02
votjakovrHeikoS1: ok, thanks, i understand now :)16:06
HeikoS1votjakovr: how do you like this framework?16:06
HeikoS1votjakovr: it fits well into existing parts16:07
HeikoS1votjakovr: my only convern is that jacob abused the parameter combinations a bit16:07
shogun-buildbotbuild #1624 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at
-!- iglesiasg_ [] has quit [Ping timeout: 240 seconds]16:15
shogun-buildbotbuild #1329 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Thoralf Klein <>16:17
shogun-buildbotbuild #1543 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Thoralf Klein <>16:24
-!- votjakovr [] has quit [Quit: WeeChat 0.4.0]16:26
-!- iglesiasg_ [] has joined #shogun16:28
thoralfwiking: I was tired of the empty cxx11_* temp directories below inside the build directory:
shogun-buildbotbuild #1330 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>17:00
HeikoS1wiking: around?17:14
-!- thoralf [] has quit [Quit: Konversation terminated!]17:23
shogun-buildbotbuild #1331 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Thoralf Klein <>17:42
-!- iglesiasg_ [] has quit [Quit: Ex-Chat]17:56
-!- shogun-notifier- [] has quit [Quit: transmission timeout]18:25
-!- lisitsyn [] has quit [Quit: Leaving.]18:28
-!- lisitsyn [] has joined #shogun18:29
-!- pickle27 [] has joined #shogun18:41
@sonney2kwiking, the server is there18:42
-!- lisitsyn [] has quit [Quit: Leaving.]18:46
@sonney2kHeikoS, please send me your ssh public key19:13
-!- Broschki [d99e8f4b@gateway/web/freenode/ip.] has joined #shogun19:17
HeikoS1ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDgoFlUBxPYyMuVc0ZqDA0Gt5GVT9vPfHyapZvZApxmn4aoNGfNusgJWcPbJjwcl5r15qLYfE7ZpQUC8xFrafpFK8h+1u+kg1aSlj/Qe9VCtamFaR/I+VI9XvpdgXfkahbFy/dk7a3YAwoGO4bCs4Z2KpQSamSM2k61CkLzT5iFEr+yJ5OHYansr65vjD0t0kmcenyPeCK+NoxUpFO1JfBHEXvBxhVDzZ9kZdCoXJ34z1RHfwpybKyDSKPOxzogjVVdSZhdHjMVoyu8t4EYKnyzCmX4qNT7budx0itn4RUz92ZgJp2KE1HOB3HI3edEquymarxjIm9XWhC9diaWxUFx heiko@heiko-ThinkPad-T42019:17
@sonney2kHeikoS, per mail please19:17
@sonney2kBroschki, ?19:17
HeikoS1Broschki: ?psel!19:19
BroschkiHeikoS1: Shogun ist eine tolle Sache!19:19
* sonney2k wonders what is going on19:19
-!- Broschki [d99e8f4b@gateway/web/freenode/ip.] has quit [Client Quit]19:19
HeikoS1sonney2k: that is my girl stalking me :)19:20
@sonney2kHeikoS, please send it to me19:20
@sonney2knot the mailinglist19:20
@sonney2knot IRC19:20
HeikoS1well you have it now dont you?19:20
@sonney2kI don't trust it19:20
@sonney2kso private mail to me again please19:21
@sonney2kand if someone wants to trust your key it is up to him/her19:21
HeikoS1got it?19:25
@sonney2kHeikoS, try ssh heiko@bug.7nn.de19:26
HeikoS1sonney2k: nope19:27
@sonney2kHeikoS, nope as in?19:27
HeikoS1Permission denied (publickey).19:27
@sonney2kHeikoS, sure you did that from the machine w/ that ssh key?19:28
HeikoS1ill double check19:28
HeikoS1so its the file in .ssh/ids_rsa.pub19:30
HeikoS1and I am using that for github too19:30
@sonney2kohh maybe bug did not yet propagate19:30
@sonney2kin DNS19:30
HeikoS1so wait?19:30
@sonney2kHeikoS, so try ssh heiko@
@sonney2kHeikoS, should be quite fast19:32
HeikoS1what system is that?19:33
@sonney2kHeikoS, we can run the demos on it19:33
@sonney2kbut I am still installing shogun deps19:33
HeikoS1ok ill try to compile it ...19:33
@sonney2kno compiler no git yet :D19:33
HeikoS1ah :)19:33
@sonney2kbut yes try it and complain19:33
HeikoS1-bash: git: command not found19:33
HeikoS1grrrrrrrrr!!!!! :D19:33
@sonney2kI just got it a an hour ago so don't complain19:34
@sonney2kyou all have logins already19:34
@sonney2kI will also put the buildbot on this19:34
HeikoS1cool, thanks for that!19:34
@sonney2kand remove it from the poor webserver19:34
@sonney2kHeikoS, git and cmake are installed now19:35
@sonney2kwhat else?19:35
HeikoS18 cores19:35
HeikoS1lets see19:35
@sonney2kg++/gcc are there already19:36
HeikoS1CMake Error: Could not find cmake module file:/home/heiko/shogun/build/CMakeFiles/CMakeCXXCompiler.cmake19:36
HeikoS1CMake Error: CMAKE_C_COMPILER not set, after EnableLanguage19:36
HeikoS1CMake Error: CMAKE_CXX_COMPILER not set, after EnableLanguage19:36
HeikoS1-- Configuring incomplete, errors occurred!19:36
@sonney2kdpkg -l | grep g++19:36
@sonney2kii  g++                             4:4.7.2-1                     amd64        GNU C++ compiler19:36
@sonney2kii  g++-4.7                         4.7.2-5                       amd64        GNU C++ compiler19:36
HeikoS1never saw this error before19:37
HeikoS1I just re-logged so its not the bash session19:37
HeikoS1ah there we go19:37
HeikoS1python libs?19:37
HeikoS1ccmake would be good19:37
HeikoS1just did just did a make -j 8 :)19:38
@sonney2kfeels good doesn't it :D19:38
HeikoS1yes :)19:39
@sonney2kok swig is on19:39
@sonney2kR develop19:39
@sonney2kwhat is ccmake?19:39
HeikoS1a "graphical" tool for configureing cmake options19:40
HeikoS1compiling went through19:40
HeikoS1ccache somehow is not installed by cmake19:41
HeikoS1recompiling re-does everything19:41
@sonney2kok installed19:41
@sonney2klua too19:41
@sonney2kis on19:42
HeikoS1 CMake Error at19:45
HeikoS1 /usr/share/cmake-2.8/Modules/FindPackageHandleStandardArgs.cmake:9719:45
HeikoS1 (MESSAGE):19:45
HeikoS1   Could NOT find Jinja2 (missing: JINJA2_IMPORT_SUCCESS)19:45
HeikoS1 Call Stack (most recent call first):19:45
HeikoS1   /usr/share/cmake-2.8/Modules/FindPackageHandleStandardArgs.cmake:28819:45
HeikoS1   cmake/FindJinja2.cmake:17 (find_package_handle_standard_args)19:45
HeikoS1   CMakeLists.txt:723 (FIND_PACKAGE)19:45
HeikoS1this just started popping up19:45
@sonney2kHeikoS, installed19:47
@sonney2kbtw how do you like the new cmake summary?19:47
@sonney2kany ideas for improvement?19:47
HeikoS1sonney2k: I like that a lot19:48
HeikoS1so-- Could NOT find Subversion (missing:  Subversion_SVN_EXECUTABLE)19:48
HeikoS1CMake Error at /usr/share/cmake-2.8/Modules/ExternalProject.cmake:1071 (message):19:48
HeikoS1  error: could not find svn for checkout of GoogleMock19:48
HeikoS1Call Stack (most recent call first):19:48
HeikoS1  /usr/share/cmake-2.8/Modules/ExternalProject.cmake:1542 (_ep_add_download_command)19:48
HeikoS1  cmake/external/GoogleTestNMock.cmake:6 (ExternalProject_Add)19:48
HeikoS1  tests/unit/CMakeLists.txt:7 (include)19:48
@sonney2kheh it really has 10MB/s downloading deb packages19:50
HeikoS1sonney2k: ^19:52
HeikoS1wiking: around?19:53
@sonney2kHeikoS, long installed19:54
HeikoS1wow now ccache kicks in19:55
HeikoS1think I will buy a new notebook with 8 cores soon thats really convienient19:55
@sonney2kget 32 cores ;)19:56
HeikoS1in a notebook? :)19:56
@sonney2kmy phone has 8 cores :P19:56
HeikoS1I remember something back from a computer science lecture that around 32 to 64 cores it start saturating too much due to shared memory19:57
@sonney2kHeikoS, tales of ...19:57
HeikoS1well, 8 is far from that19:57
HeikoS1I in fact have 219:57
@sonney2kno seriously it depends on the application19:57
HeikoS1with this hyperthreading so linux sees 419:57
HeikoS1yeah of course19:57
@sonney2kI observed this around 8-16 cores when I did the COFFIN experiments19:58
HeikoS1ah interesting19:58
@sonney2kbut it depends really19:58
HeikoS1I think so19:58
@sonney2kif you can squeeze everything in cache19:58
HeikoS1I mean its also really nice for just working19:58
@sonney2kno shared memory issues at all19:58
@sonney2kfor compiling yes19:58
HeikoS1since currently when I compile shogun, my computer getts so slow19:58
@sonney2kbut otherwise hmmh19:59
HeikoS1also I tend to have 20 program open19:59
HeikoS1and you know I like eclipse :D19:59
HeikoS1+20 open tabs in browser19:59
HeikoS1and things start to get a bit slower19:59
HeikoS1just passed the unit tests19:59
HeikoS1wow they were lightning fast19:59
HeikoS1oh, but I compiled without debug19:59
@sonney2kdebug is always on IIRC20:00
HeikoS1default is release20:01
HeikoS1chcek ccmake20:01
@sonney2kbut debug symbols20:02
@sonney2k-g is20:02
-!- gsomix [~gsomix@] has joined #shogun20:02
@sonney2kgsomix, !20:02
gsomixsonney2k, hey!20:03
@sonney2kgsomix, I hope you are OK?20:03
@sonney2kyou seemed quite busy / distracted...20:03
gsomixsonney2k, yep. working, cleaning code, preparing PR. so, I finished a lot of my not-shogun works.20:06
gsomixnow I have a lot of `free' time.20:06
@sonney2kgsomix, OK sounds good!20:06
@sonney2kgsomix, so lets fix the bugs and then get at least protobuf stuff to work (or matlab if you prefer that)20:07
@sonney2kgsomix, btw it is .mat files reading and I guess the easiest is steal code from octave and python scipy for that20:07
@sonney2kgsomix, but first things first20:07
@sonney2kgsomix, did you fix the locale issue already?20:08
@sonney2kgsomix, you can just use the SG_SET_LOCALE_C20:08
@sonney2kand SG_RESET_LOCALE macros20:08
@sonney2kwhich you should use before reading/writing ascii20:09
gsomixsonney2k, cool! I wanted to use setlocale.20:10
@sonney2kgsomix, we had that issue already w/ asciifile20:11
@sonney2kthat is why I know it20:11
gsomixsonney2k, OK, everything fine then20:11
@sonney2kHeikoS, what is colpack?20:12
gsomixsonney2k, btw, maybe you knows. can I use Eigen3 in commercial software?20:14
@sonney2kgsomix, sure20:14
HeikoS1sonney2k: a library for computing graph colourings20:14
@sonney2kHeikoS, yeah but it is not in debian?20:14
@sonney2kwhere did you get it from?20:14
HeikoS1sonney2k: I never installed it20:15
HeikoS1sonney2k: and it broke travis, wiking did some cmake magic to download it20:15
HeikoS1so its currently an open problem20:15
HeikoS1but for lambday it worked20:15
-!- thoralf [] has joined #shogun20:15
@sonney2kthoralf, hey20:15
thoralfHey S?ren20:16
@sonney2kHeikoS, I think I have all deps installed now20:17
@sonney2kexcept colpack and cplex *har* *har*20:17
@sonney2kwell ok mosek also not but I think wiking didn't take care of that20:17
thoralfCan someone please  restart the builds for  I think the breaks are not related to my changes:
thoralfsonney2k: Do you need trial licenses for cplex?20:18
@sonney2kwe should just get rid of the mosek/cplex parts20:18
@sonney2kit is no use20:18
thoralfOkay.  Just in case: ilog is very generous with (short-term) trial licenses.20:19
@sonney2kthoralf, ok20:20
@sonney2kHeikoS, I will explode swap a bit20:20
HeikoS1sonney2k: man, the ipython guys are now thinking of adding d3js to the notebooks20:20
@sonney2k512MB is not enough20:20
HeikoS1sonney2k: what?20:20
@sonney2kok so we have ~2TB on this machine20:21
@sonney2kHeikoS, what do they add d3 support for? I mean they can hardly replace matplotlib or will matplotlib get a d3 backend?20:25
HeikoS1no idea, just talking about it20:26
HeikoS1I think its in addition20:26
HeikoS1being able to create these things in notebooks20:26
@sonney2kwiking, you have!20:30
@wikingwhere? :)20:30
@sonney2kman machine fast20:30
@wikingwhat's the ip :)20:30
@sonney2k163MB/s for creating swap20:30
@sonney2kssh wiking@bug.7nn.de20:31
@wikinghehe in20:31
HeikoS1wiking: haha :)20:31
@sonney2kwiking, I wish gunnar did transfer the domain name20:31
@sonney2kwiking, I just installed the system20:32
@wikingme too20:32
@wikingsonney2k: cool20:32
HeikoS1wiking: could you update the wiki on the meeting? would be good to keep good track of everything, with links to irc logs etc etc20:32
@wikingHeikoS1: yeah i will20:32
@sonney2kwiking, all shogun deps should be installed now20:32
@sonney2kwiking, I also put buildslave etc on20:32
@wikingHeikoS1: i just still have to do some work20:32
@wikingsonney2k: \o/20:32
@wikingHeikoS1: non shogun related20:32
@wikingHeikoS1: but i have to eat something you know; )20:32
HeikoS1wiking: what? you are actually doing something else? :D20:32
@wikingyeah unfortunately20:33
@wikingon windows man20:33
@wikingi'll fucking kill MSVC20:33
@wikingthat's like the biggest crap compiler ever i've seen20:33
@wikingok bbl20:33
@wikingi gotta finish this one20:33
@sonney2kwiking, check /home/wiking/p20:34
@sonney2kthen remove the file20:34
@sonney2kand yes you can just do20:34
@sonney2ksudo su20:34
@wikingsonney2k: done20:34
@wikinghahah hat paswd20:34
@sonney2kwiking, if you have time you can setup the buildslave20:34
@wikingsonney2k: mmm i doubt man20:34
@wikingi really have some shit here to do20:34
@sonney2kyeah I guessed so20:34
@sonney2kotherwise I will do it but not today20:35
@wikingsonney2k: lol20:35
@wikingcannot su20:35
@wikingwith that passwd20:35
-!- iglesiasg [] has joined #shogun20:35
-!- mode/#shogun [+o iglesiasg] by ChanServ20:35
@wikingsonney2k: just change my passwd to something easy peasy and i'll passwd then20:36
@sonney2kwiking, try again20:36
@sonney2kwiking, I had a typo in your name20:37
@sonney2kiglesiasg, send me your ssh pub key20:37
@sonney2kwiking, so you could create chroots for fedora /  ubuntu etc under /home/fedora etc20:38
@wikingiglesiasg: u sure about that? :)20:38
@sonney2kiglesiasg, email please20:38
@wikingsonney2k: will do20:38
@wikingiglesiasg: i mean it's really not 100% that nobody can do prime factorization in O(n^p) ;)20:38
@iglesiasgwiking, mm I thought there is no problem sending the public key, that's why it is called public hehe20:38
@sonney2kwiking, at least he didn't send it to the mailinglist and IRC like Heiko did20:38
@wikingsonney2k: hahaha yeah that's another security guy ;)20:39
@wikingway to go HeikoS1 ;)20:39
@iglesiasgwiking, let's hope none can then20:39
@wikingiglesiasg: afaik the ppl owning that first quantum D-128 machine20:39
@wikingthey can alreeady factor it quite nicely20:39
@iglesiasgnot good20:39
@wikingsee shor algo20:39
HeikoS1guys i dont care about security, I got other things to do here :D20:39
@wikingor how it is called20:39
HeikoS1also I can post my public key where I want20:40
HeikoS1nothing unsecure about that20:40
HeikoS1wiking: I just got the independent computation engine running on a PBS in python20:41
HeikoS1and that rocks20:41
HeikoS1going home, bye20:42
-!- HeikoS1 [] has quit [Quit: Leaving.]20:43
@sonney2kiglesiasg, try ssh iglesias@bug.7nn.de20:43
@wikingok back to work20:43
@iglesiasgsonney2k, permission denied20:43
@sonney2kiglesiasg, host bug.7nn.de20:44
@iglesiasgand another line20:44
@sonney2kssh iglesIas20:45
@sonney2kiglesiasg, you mistyped20:45
@sonney2knot iglesas20:45
@sonney2kbut sias20:45
@iglesiasgsonney2k, true20:45
@sonney2kiglesiasg, ok now?20:45
@iglesiasgsonney2k, yes20:45
@iglesiasgsonney2k, sorry about that20:46
@iglesiasgsonney2k, is it the new machine?20:46
@sonney2know only sergey is missing20:46
@sonney2kiglesiasg, yes - try to compile shogun on it20:46
@sonney2kit is really an enjoyable experience20:46
@iglesiasgsonney2k, are we each keeping our own copy in our home?20:46
@sonney2kiglesiasg, yes sure20:47
@sonney2kiglesiasg, machine has 2 TB so stop caring20:47
@sonney2kahh but no backups!20:47
@iglesiasghehe sure20:47
-!- lambday [67157f4c@gateway/web/freenode/ip.] has joined #shogun20:54
lambdayHeikoS: just checked your comments20:54
lambdayHeikoS: I will test with the ozone matrix soon20:54
@sonney2klambday, if you send me your pubkey we have a rather fast machine now20:55
lambdaysonney2k: wow!! :D20:55
lambdaysonney2k: its the same, without the space in the end :(20:55
lambdaysonney2k: you're great!20:55
-!- arka [73bb3974@gateway/web/freenode/ip.] has joined #shogun20:55
lambdaysonney2k: I am mailing you again20:56
-!- gsomix [~gsomix@] has quit [Ping timeout: 260 seconds]20:56
@sonney2klambday, try ssh lambday@bug.7nn.de20:59
@iglesiasgsonney2k, compiling..20:59
arkahi..m interested in this project for the upcoming 14 gsoc...can u deliver me ur mail id so that i can contct u soon..20:59
@sonney2kiglesiasg, me too20:59
@sonney2kwith 8 cores20:59
@iglesiasgarka, you can find our mails in the web site20:59
@sonney2kI guess we can kill any machine :D20:59
@iglesiasgsonney2k, I am also doing make -j8 hehe20:59
lambdaysonney2k: yayy I am there21:00
@sonney2klambday, alright then enjoy21:00
@sonney2klambday, but note - no backups!21:00
lambdaysonney2k: awesome!21:00
lambdaysonney2k: alright21:00
@iglesiasgarka, but please consider using the mailing list, and more importantly reading the mailing list logs21:00
@sonney2kiglesiasg, I think we have slowed down the machine quite a bit21:01
@iglesiasgsonney2k, I hope so, in the machine I normally work on these days the build would have finished by now :)21:02
@iglesiasgbut it also has a powerful processor -- i7 2.8GHz21:02
@sonney2kiglesiasg, well I see ~16 jobs running cc1plus and ccache etc21:03
@sonney2kload 14...21:03
@sonney2kit is time that we get 64 core machines ;)21:04
@iglesiasgsonney2k, in any case it is awesome having access to a machine like this, so thank you sooo much for taking care of it!21:04
@iglesiasgit is like Christmas again :)21:04
@iglesiasglast week wiking made Christmas21:04
@sonney2kwe need the buildbot running on this and our demo server21:05
@sonney2kso I guess the machine will always have some load21:05
@iglesiasgsonney2k, and what will happen with the machine currently running the buildbot?21:05
@iglesiasgis it time to become pensioner?21:06
thoralfHow many cores/threads/RAM?21:06
@sonney2kiglesiasg, well it can be our webserver21:06
@sonney2kiglesiasg, but *only* webserver :)21:06
@sonney2kI think this separation makes a lot of sense21:06
@sonney2kthoralf, 24G core i721:07
@sonney2kso 4 cores + 4 HT21:07
@sonney2kthoralf, I know you have much better stuff at ZIB so I am not even asking21:07
thoralfNot bad.21:07
@sonney2kthoralf, 36 EUR/month21:07
@sonney2kthanks to google's gsoc money we can afford this :D21:08
thoralfsonney2k: No, we usually don't work on the supercomputer. ;)21:08
thoralfSince it's to much work preparing experiments for that.21:08
-!- gsomix [~gsomix@] has joined #shogun21:08
thoralfPowerful commodity hardware rocks.21:09
@iglesiasgomg valgrind is not installed!!21:12
@sonney2kiglesiasg, liar ;)21:14
@iglesiasgsonney2k, thanks! :)21:16
lambdaywiking: :(21:16
@sonney2klambday, ?21:17
lambdaywiking: I have made a few more classes only visible if eigen3 is there...21:17
lambdaysonney2k: without colpack, I can't get it compiled21:17
thoralfAs everyone got a chrismas present now, may I ask for mine?  Could someone restart the builds for PR ;)21:17
@iglesiasgwow valgrind didn't like EPInferenceMethod tests21:17
@sonney2kwow ccache does it in 9 secs if cache is filled21:17
thoralfsonney2k: Which target?21:18
lambdaysonney2k: class_list.cpp seems to be trying to instantiate a class which should not be there in absence of an external library21:18
@sonney2kthoralf, can't you do it yourself?21:18
lambdayexternal library=colpack (for graph coloring)21:18
@sonney2kthoralf, anyway restart clicked21:18
thoralfsonney2k: No, I can't.  I refused travis to access my profile/email-adresses/followers on github.21:18
@sonney2klambday, how did you get colpack installed then?21:19
thoralfsonney2k: Thanks.21:19
@iglesiasghushell, hey!21:19
@sonney2kthoralf, why that?21:19
@iglesiasghushell, did you find out if there was something wrong with the unit tests? I just run valgrind on them and they were all good mmm21:19
lambdaysonney2k: I installed it beforehand... in my system it works fine since it finds colpack and sets HAVE_COLPACK=121:19
thoralfsonney2k: Why should travis know my private mail addresses registered in github?21:19
thoralfsonney2k: There's no need for that.21:20
@sonney2kthoralf, well.. I get your point21:20
lambdaywithout this, shogun should compile just fine, because the class I have used this library for, I've kept this inside a #if.....#endif block21:20
thoralfsonney2k: I'm curious: Which target finished in 9 seconds?21:21
lambdaysonney2k: its a minor thing anyway... but should work fine21:21
lambdaysonney2k: I am afraid if things work fine in absence of eigen3, because there are a few classes I only made visible in presence of eigen321:21
lambdaysimilar situation21:21
* sonney2k just ordered a tomtom multi-sport gps watch
@sonney2kthoralf, all interfaces21:22
@sonney2klambday, but eigen3 is installed21:22
thoralfsonney2k: Wow.21:22
lambdaysonney2k: yes that's why it works fine... if eigen3 is NOT installed, I think I'd get similar type of errors for few more classes as well21:22
@sonney2kthoralf, I would wish I had that back in the summer21:22
@sonney2kthoralf, in winter it is rather boring - since I use this for outdoor swimming21:23
lambdayreally excited about the 3.0 release21:25
lambdaysonney2k: would it include the additions that we made in this gsoc?21:25
@sonney2klambday, of course21:25
lambdaysonney2k: yayy :D21:25
@sonney2kthat is why you all have to finish you projects :D21:25
lambdaysonney2k: hehehe :D21:26
lambdaysonney2k: well, regarding mine, all modules have been implemented.. :-/ I gotta add more and more tests...21:26
lambdayand think of optimitions21:26
@sonney2kno worries21:26
-!- arka [73bb3974@gateway/web/freenode/ip.] has quit [Quit: Page closed]21:26
@sonney2kwe still ahve a month at least21:26
lambdaythis week after this PR gets merged, I'll work on practical scenarios (the ozone matrix) where the project is actually going to be used21:28
lambdaygetting really excited, and a bit scared as well..21:29
thoralfsonney2k: I didn't add my name to none of the changed files' preamble - is there a policy for that?21:29
thoralfsonney2k: Didn't take any credits so far. ;)21:29
lambdayI was reading a HKL paper and read your reference :D21:31
lambdayI wanna understand this MKL thing...21:32
@sonney2kthoralf, take credit dammit21:36
-!- sonne|osx [] has joined #shogun21:38
shogun-buildbotbuild #53 of osx1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Soeren Sonnenburg <>, Thoralf Klein <>, Kevin <>, Heiko Strathmann <>, Fernando Iglesias21:50
shogun-buildbot<>, Roman Votyakov <>21:50
hushelliglesiasg: hi, valgrind is fine, I think the bugs are initialization21:54
hushelliglesiasg: I found HeikoS corrects my mistakes
hushellHeikoS: Thanks! :)21:55
hushelliglesiasg: Shall I send a PR right now? If not urgent I'll include it with my PR a bit later today21:56
@iglesiasghushell, it is better if you send a pull request just with that21:56
@iglesiasghushell, it is a good practice to keep pull requests and commits in general with small functionality as purpose I think21:57
@iglesiasgeasier to revise, also easier to debug in case it turns out that there was a bug in any of them21:58
hushelliglesiasg: okay, I'll do it right away21:58
-!- lambday [67157f4c@gateway/web/freenode/ip.] has quit [Ping timeout: 250 seconds]22:04
shogun-buildbotbuild #1627 of deb3 - modular_interfaces is complete: Failure [failed install test python_modular]  Build details are at
-!- pickle27 [] has left #shogun []22:30
-!- gsomix [~gsomix@] has quit [Ping timeout: 245 seconds]22:33
@iglesiasgall right people, good night!22:34
-!- iglesiasg [] has quit [Quit: Ex-Chat]22:34
-!- gsomix [~gsomix@] has joined #shogun22:50
thoralfsonney2k: Apropos big machines.  This is what we want:
-!- hushell [] has quit [Ping timeout: 264 seconds]23:48
shogun-buildbotbuild #1626 of deb3 - modular_interfaces is complete: Failure [failed compile csharp_modular]  Build details are at
-!- thoralf [] has quit [Quit: Konversation terminated!]23:58
--- Log closed Wed Aug 28 00:00:34 2013