Open in new window / Try shogun cloud
--- Log opened Mon Dec 02 00:00:49 2013
shogun-buildbot_build #317 of osx1 - libshogun is complete: Failure [failed compile test shell]  Build details are at  blamelist: Soeren Sonnenburg <>00:35
-!- zxtx [] has quit [Ping timeout: 264 seconds]00:58
-!- Boeke [] has quit [Remote host closed the connection]01:06
-!- zxtx [] has joined #shogun01:10
shogun-buildbot_build #635 of nightly_default is complete: Success [build successful]  Build details are at
-!- Saurabh7 [~Saurabh7@] has quit [Ping timeout: 246 seconds]05:19
-!- sonne|osx [~sonne@] has joined #shogun07:50
sonne|osxwiking: could you tell how csharp modular examples are run via ctest?07:50
sonne|osxwiking: I cannot reproduce any of the crashes when I use the script07:51
@wikingno .sh08:01
sonne|osxwiking: yeah but what is the exact command line ctest is running?08:02
@wikingsee buildbot?08:03
@wikingi mean seriously....08:03
sonne|osxwiking: no08:03
sonne|osxwiking: buildbot just says ctest08:03
@wiking/bin/bash -c 'cd build && ctest --output-on-failure -j2 -R csharp'08:03
sonne|osxnothing like mono <whateveropts> file08:03
sonne|osxwiking: that doesn't help08:03
sonne|osxwiking: I need the command that is run for each test08:04
@wikingfirst line i've copy pasted...08:04
@wikingadd -V08:04
@wikingand u'll see the exact cmd08:04
sonne|osxwiking: when I run that it doesn't find the dll / so08:08
sonne|osx[ERROR] FATAL UNHANDLED EXCEPTION: System.IO.FileNotFoundException: Could not load file or assembly 'modshogun, Version=, Culture=neutral, PublicKeyToken=null' or one of its dependencies.08:08
@wikingenv var...08:08
sonne|osxthere is none on
@wikingyeah sure08:09
@wikingthen there is none08:09
@wikingcome on08:09
@wikingit's really too early for this08:09
@wikingwhy dont u check the example/undocumented/csharp-modular/CMakeLists.txt08:10
@wikingobviously there's no env var:08:10
-!- sonne|osx [~sonne@] has quit [Quit: sonne|osx]08:16
-!- lisitsyn [~lisitsin@] has joined #shogun09:57
-!- lisitsyn [~lisitsin@] has quit [Ping timeout: 252 seconds]10:23
-!- lisitsyn [~lisitsin@] has joined #shogun10:32
-!- besser82 [~besser82@2a02:8108:8840:1800:e8b:fdff:fe16:bb33] has joined #shogun10:33
-!- besser82 [~besser82@2a02:8108:8840:1800:e8b:fdff:fe16:bb33] has quit [Changing host]10:33
-!- besser82 [~besser82@fedora/besser82] has joined #shogun10:33
besser82sonney2k, wiking, lisitsyn, sonne|work: Guten Morgen  :D10:34
lisitsynbesser82: morgen10:35
besser82lisitsyn: Sergej, how is it going?  ;)10:36
lisitsynbesser82: horosho!10:36
lisitsynbesser82: a u tebya?10:37
besser82lisitsyn:  ????  *puzzled*10:38
lisitsynbesser82: hah jk10:38
lisitsynbesser82: sonne|work can understand that though!10:39
besser82lisitsyn: is that some asian lang?  Sounds a bit...10:39
besser82lisitsyn: or is that the comeback for me writing "lizzard-style" sometimes?!?10:40
lisitsynbesser82: asian? russian10:41
besser82lisitsyn: ain't russia in asia, mostly?!?10:41
besser82lisitsyn: the border europe/asia is somewhat near moscow afaik, ain't it?10:43
besser82lisitsyn: but me is fine, too, thanks!10:45
lisitsynbesser82: europe/asia border is ural10:45
lisitsynas considered here10:46
lisitsynbesser82: well I wouldn't like to be anyhow related to asia :D10:47
lisitsynmost of things are european10:47
lisitsynand well some EU countries speak some kind of similar language10:47
besser82lisitsyn: so the border is actually more easter than i thought  ;)10:47
besser82lisitsyn: i know...   Czech, e.g. is somehow related....10:48
besser82lisitsyn: at least all russian-people i know can understand that...10:48
lisitsynbesser82: I think it is easier to understand slovenian actually10:49
lisitsynczech is very influenced by some roman stuff10:49
besser82lisitsyn: I dunno  :D  Don't understand either  :-P10:49
lisitsynand actually our letters were introduced by greeks haha10:49
lisitsynso we have quite similar alphabet10:49
lisitsynbesser82: serbian is even more closer - I can understand some parts of it10:50
besser82lisitsyn: I actually don't understand any cyrilic-based lang, but just some stuff like food...10:51
lisitsynbesser82: I had symmetric stuff when I was in berlin ;)10:52
besser82lisitsyn:  :D    Ausweis, bitte!  XD10:53
lisitsynbesser82: the most 'shameful' when we came to some restaurant10:53
lisitsynactually it was called traube IIRC10:53
lisitsynand there was 'zwei gaenge 15 EUR'10:53
lisitsyndrei gaenge 20 EUR10:53
lisitsynor so10:53
lisitsynI had to ask to explain what is gaenge10:54
besser82lisitsyn:  would be the same for if would come to russia  ;)10:54
lisitsynbesser82: there is a chance people speak english if you come to some big city10:56
lisitsynbut generally yes that's troublesome without understanding here10:56
besser82lisitsyn: I hope so  ;)  But how about places like "Novosibirsk"?!?10:56
lisitsynbesser82: novosibirsk is quite big city and you can find someone able to speak english10:57
besser82lisitsyn: I know, it was just some synonym for some more non-tourist area...10:57
lisitsynbesser82: haha no, why10:58
lisitsynnon-tourist area is say arkhangelsk10:58
lisitsynmay be murmansk10:58
besser82lisitsyn: murmansk...  somewhen heard about that...10:58
lisitsynbesser82: this way you get no service for english speakers + bad weather10:59
besser82lisitsyn: But when I would come to russia, I'd really like to visit Duga-3  ;)10:59
lisitsynwhat's duga-3?10:59
besser82lisitsyn: never heard about that?10:59
lisitsynno :)10:59
lisitsynbut googled10:59
lisitsynbesser82: though it is belarus11:00
lisitsynbesser82: as well as chernobyl which is between ukraine and belarus11:00
besser82lisitsyn: by my knowledge about geography, boarders of russia are actually from my old atlas  ;)11:01
besser82lisitsyn: which shows me some "CSSR" at those locations.  XD11:01
lisitsynbesser82: USSR you mean? :)11:01
lisitsynor CCCP in native11:01
besser82lisitsyn: USSR, rye  ;)11:02
sonne|workhey besser82!11:37
-!- iglesiasg [] has joined #shogun11:39
-!- mode/#shogun [+o iglesiasg] by ChanServ11:39
@iglesiasgwiking, sonne|work, what's the version of Shogun that runs on cloud?12:04
sonne|workiglesiasg: I guess 3.012:05
@iglesiasgsonne|work, so I guess it does not get new stuff as they are implemented in develop12:05
sonne|workiglesiasg: no12:11
sonne|workmanual procedure12:11
besser82sonne|work: Hey Sonne!12:14
@wikingiglesiasg: 3.012:30
@iglesiasgall right, thanks wiking!12:31
@wikingas soon as docker changes the way it does oauth on github we'll be able to do travis like docker image builds12:33
-!- iglesiasg [] has quit [Quit: Leaving]13:18
-!- thoralf [] has joined #shogun15:20
thoralfHey GUIs15:21
besser82thoralf: Hall?chen  ;)15:47
thoralfHey Bj?rn15:48
besser82thoralf: your PR looks good from code-pov  ;)15:48
thoralfHehe, thanks.15:51
thoralfIt's only a part of the solution - if done right, we should make StructuredLabels an interface.15:51
besser82thoralf: sounds good  ;)  but if we want to get that implemented, we should get that done during we rewrite the interfaces for using d-pointers...15:55
thoralfWhat's d-pointers?15:56
besser82thoralf: d-pointers is a technique to wrap all function / object-calls into pointers, so you won't have to deal with external deps anymore...15:57
besser82thoralf: like having an "internal" interface for build-time and a public interface for linking against libshogun15:58
besser82thoralf: qt is one project using this15:58
thoralfbesser82: I must've missed that development.15:58
lisitsynthoralf: in depth description of d-ptrs:15:58
lisitsynclass A { struct AImpl; AImpl* impl; };15:59
thoralfDoesn't look very beautiful.16:00
thoralfIs that idiomatic for c++ programmers?16:00
lisitsynthoralf: yes, this way you hide your implementation/data16:00
lisitsynbecause that impl is incompletely defined16:00
lisitsynyou can use whatever libraries16:00
lisitsynand they are not exposed to headers16:01
lisitsynthoralf: if you use it you get no header dependencies + fast compilation + stable headers/API16:04
besser82thoralf: here is the d-ptr / q-ptr thing explained really verbose:
-!- zxtx [] has quit [Ping timeout: 272 seconds]16:10
thoralfbesser82: Looks like it would break IDEs auto completion.16:22
besser82thoralf: not sure about that, but afaik autocompletion of Eclipse works for Qt.16:23
besser82thoralf: so either there is some addon for Qt in Eclipse or Eclipse can handle it properly...16:23
thoralfMaybe I didn't fully understand it. ;)16:24
@wikingthoralf: it wouldn't... as you would still have the public api16:26
@wikingthoralf: hence if you are just writing an external application it would be just the same16:26
thoralfOkay, so the public API is still exported.16:27
thoralfOnly data is hidden?16:27
@wikingthoralf: no16:27
@wikingthoralf: not only the data16:27
@wikingthoralf: e.g. you would have any private function or member exposed in the public api16:27
@wikingthoralf: there would be nothing like private: void init()...16:27
@wikingas that would be implemented within the private class16:28
thoralfI see.16:28
@wikingof course if you are writing a shogun module16:28
@wikingyou would still work with the shogun source16:28
thoralfAnd no need to make them private in the private class anyway, since its already hidden?16:28
@wikingand there you would have reference on the private class16:28
@wikingthoralf: yes16:28
@wikingthoralf: the only private member of the class would be16:28
thoralfOkay, got it.16:28
@wikinga private class16:28
@wikingsay we have MagicMachine16:29
@wikingit would have like a MagicMachine.h16:29
@wikingthat would have all the public api definitions16:29
@wikingand something like16:29
@wikingprivate: MagiceMachineImpl* dptr;16:29
@wikingand that MagiceMachineImpl.h/cpp would have all the internal stuff of MagicMachine16:29
@wikingwould you would never expose the MagicMachineImpl api16:30
@wikingi.e. it would never be installed to <prefix>/shogun/...16:30
thoralfOkay, now back to the PR: You suggested to apply this d-pointer stuff to the PR classes as well?16:30
@wikingthoralf: well i think the dpointer stuff needs to be discussed16:30
@wikingwe are plannign this for a while16:31
@wikingsince it would help a lot with swig16:31
@wiking(much smaller generated .cxx etc.)16:31
@wikingbut we need to have a discussion about this how exactly we want to do it16:31
@wikingand not just randomly do it16:31
@wikingso i think u just keep it as it is16:32
@wikingbefore we lay down the ground rules of how to do dptrs in shogun16:32
@wikingimho all of us has ideas, but that needs to be fixed16:32
@wikingas you see for example Qt has some magic macros there to do this dptr stuff16:32
@wikingso even if it sounds an easy task16:32
@wikingwe should do some predesign of the whole thing16:34
@wikingso that we do it in a general/robust way...16:34
@wiking</bullshit stuff>16:34
thoralfWait.  I missed the opening tag. ;)16:43
@wikingthoralf: somewhere around 2011 february ;16:48
-!- thoralf [] has quit [Quit: Konversation terminated!]17:03
-!- gideonite [] has left #shogun []17:49
-!- zxtx [] has joined #shogun18:01
-!- sonne|osx [~sonne@] has joined #shogun18:13
sonne|osxbesser82: so how is it going? don't try to dptr stuff before you are done cmaking :)18:17
besser82sonne|osx: doing the final-polish(tm) of first-stage-besser82-cmake-skills(tm)  XD18:17
besser82sonne|osx: will take me 'till ~22:00 CET, after finished i'll push feature/CMakeImproved and paste link to travis-build18:19
sonne|osxbesser82: exciting! I can't wait to try it...18:20
besser82sonne|osx: I can't to hear you say: "bessr82, you f*ckin' skrewed-up everything"  :-P18:21
@wikingbesser82: just push the feature branch into shogun's main repo18:21
@wikingbesser82: it'll autotrigger a travis build18:21
@wikingor same with the PR :)18:21
@wikingwhichever fits u18:21
sonne|osxbesser82: hehe I am through this a couple of times and wiking experienced that too18:22
besser82wiking: I know; but i don't want to rebase around in main-repo, even not in feature-branch  :-P18:22
sonne|osxbesser82: so why should you feel better?18:22
@wikingthat's what feature branch is for...18:22
@wikingso i do really hope that u r going to push into the main repo18:22
besser82wiking: I am..., just be patient a bit ;)18:23
-!- thoralf [] has joined #shogun18:25
sonne|osxthoralf: that was exactly what I meant...18:26
thoralfsonne|osx: The PR?18:26
sonne|osxthoralf: yeah18:26
thoralfsonne|osx: Is this good or bad? ;)18:27
sonne|osxthoralf: depends on whom you ask18:27
-!- zxtx [] has quit [Ping timeout: 272 seconds]18:28
sonne|osxthoralf: but it is really an optimization for a very particular case18:29
@wikingthoralf: kudos for the structdata refactor!18:29
sonne|osxso IMHO this overload is fine18:29
* sonne|osx off train18:29
-!- sonne|osx [~sonne@] has quit [Quit: sonne|osx]18:29
thoralfsonne|osx, wiking: Okay, the Multiclass case is only a proof-of-concept of what could be used for my multilabel stuff.18:31
thoralfwiking: Well, the refactoring is not that good: A multilabel object will have plenty of members: (m_num_classes, m_labels_vector, m_num_labels_set) plus (m_labels) plus (m_current_values)18:33
thoralffrom MulticlassSOLabels, StructuredLabels, CLabels resp.18:33
thoralfSo we're duplicating stuff and having semantically redundant features, but of different types.18:34
thoralfm_labels_vector, m_labels, m_current_values <-- All of them are non-trivial types.18:34
thoralf(Vectors and DynObjArray)18:34
@wikingthoralf: well whatever, but numbers speak for themselves18:39
@wiking136MiB vs 17MiB18:39
thoralfInitially 10GiB vs. 17MiB  ;)18:40
@wikingthat fucking crazy18:42
thoralfAnyway, any suggestions what to change to get this merged?18:43
thoralfTests are already running - the buildbreak is not related to the changes.18:43
-!- thoralf [] has quit [Quit: Konversation terminated!]18:54
-!- sonne|osx [] has joined #shogun19:55
-!- HeikoS [] has joined #shogun20:14
-!- mode/#shogun [+o HeikoS] by ChanServ20:14
besser82HeikoS: Hey, man!  Haven't seen for a long while  :D  How's ya doing?20:35
-!- iglesiasg [] has joined #shogun20:56
-!- mode/#shogun [+o iglesiasg] by ChanServ20:56
@iglesiasggreetings people20:56
-!- sonne|osx [] has quit [Quit: sonne|osx]21:05
besser82iglesiasg: Hey, man!21:07
@iglesiasgbesser82, what did you decide in the end regarding the cmake problem you pointed out yesterday?21:08
besser82iglesiasg: sonney2k decided me to implement d-ptr  :D21:08
-!- HeikoS [] has quit [Quit: Leaving.]21:12
@iglesiasgbesser82, good good21:12
-!- shogun-notifier- [] has joined #shogun22:36
shogun-notifier-shogun: Fernando Iglesias :develop * e8e4244 / src/shogun/labels/LabelsFactory.h:
shogun-notifier-shogun: Fix wrong indentation in LabelsFactory::get_name22:36
shogun-notifier-shogun: Fernando Iglesias :develop * 1090a9f / src/shogun/labels/LabelsFactory.h:
shogun-notifier-shogun: Merge pull request #1775 from iglesias/develop22:36
shogun-notifier-shogun: Fix wrong indentation in LabelsFactory::get_name22:36
shogun-buildbot_build #329 of precise - libshogun is complete: Success [build successful]  Build details are at
shogun-buildbot_build #1433 of rpm1 - libshogun is complete: Failure [failed test]  Build details are at  blamelist: Fernando Iglesias <>23:14
@iglesiasgshogun-buildbot_, force build --branch=develop 'rpm1 - libshogun'23:17
shogun-buildbot_build forced [ETA 16m13s]23:17
shogun-buildbot_I'll give a shout when the build finishes23:17
shogun-buildbot_build #1434 of rpm1 - libshogun is complete: Failure [failed test]  Build details are at
-!- travis-ci [] has joined #shogun23:37
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 []23:37
shogun-buildbot_build #2069 of deb3 - modular_interfaces is complete: Failure [failed test csharp modular]  Build details are at  blamelist: Fernando Iglesias <>23:47
--- Log closed Tue Dec 03 00:00:50 2013