Open in new window / Try shogun cloud
--- Log opened Mon Nov 11 00:00:19 2013
shogun-buildbot_build #612 of nightly_default is complete: Failure [failed notebooks]  Build details are at
-!- hushell [] has joined #shogun04:40
-!- sonne|osx [~sonne@] has joined #shogun08:19
sonne|osxwiking: could you please update the template?08:19
sonne|osxwiking: it now shows images of the notebooks08:19
sonne|osxwiking: btw we could display abstracts too if we wanted - do we?08:20
-!- sonne|osx [~sonne@] has quit [Quit: sonne|osx]08:39
@wikingsonney2k: will do as soon as i have some time08:47
sonne|workwiking: kthx09:13
-!- iglesiasg [] has joined #shogun10:30
-!- mode/#shogun [+o iglesiasg] by ChanServ10:30
@iglesiasgany idea about some notebooks not being displayed in
@iglesiasgthumbnails are missing too in the demo page10:33
sonne|workiglesiasg: works here10:34
sonne|workiglesiasg: reload?10:34
@iglesiasgsonne|work, already tried mmm10:34
@iglesiasgsonne|work, and about the notebooks?10:34
@iglesiasgI see that at least the LMNN and structured learning ones are missing10:34
@iglesiasgthe dimension reduction one too I think10:35
sonne|workdemos seems to be b0rken10:35
sonne|workiglesiasg: sigh missing " " in the img src10:36
sonne|workare causing this10:36
sonne|workiglesiasg: you mean there are fewer notebooks?10:36
@iglesiasgsonne|work, yeah sure10:36
sonne|workI guess the link on our website is wrong then10:36
sonne|workI mean current should point to 3.010:37
sonne|worksince for some reason FGM's are too slow10:37
@iglesiasgI actually have no idea why fewer notebooks are shown10:39
@iglesiasgbut it should be due to one of the recent changes10:40
@iglesiasgfrom get_notebook_list("*.ipynb") to get_notebook_list(".ipynb")10:41
@iglesiasgsonne|work, is that on purpose?10:41
sonne|workiglesiasg: yes that is correct10:42
@iglesiasgall right10:42
sonne|workiglesiasg: well buildbot fails!10:42
sonne|workand current link seems to be overwritten10:42
sonne|workso fewer notebooks10:42
sonne|workas I said10:42
sonne|workFGM is too slow for soem reason now?!10:42
@iglesiasgit seems so10:43
@iglesiasgnot good10:43
@iglesiasgexecuting locally...10:43
-!- iglesiasg [] has quit [Ping timeout: 244 seconds]11:52
-!- zxtx [~zv@2001:558:6033:150:287e:faea:64e2:d5ec] has quit [Ping timeout: 245 seconds]12:50
-!- iglesiasg [] has joined #shogun13:17
sonne|workiglesiasg: and?14:25
sonne|workhow long did it take?14:26
iglesiasgsonne|work, the time it normally took when it was well built14:26
iglesiasgabout 10 minutes in total14:26
iglesiasgthere is a cell that takes almost 500 seconds14:26
iglesiasgthe rest are ok14:26
iglesiasgbut btw, I got an error in one of the first cells, the one that tries to install networkx14:27
iglesiasgbut that does not seem to be happening in the buildbot so I skipped it14:27
sonne|workyeah no that is not it14:30
sonne|workmaybe lets just try again14:31
sonne|workshogun-buildbot_: force build --branch=develop nightly_default14:31
shogun-buildbot_build forced [ETA 38m09s]14:31
shogun-buildbot_I'll give a shout when the build finishes14:31
iglesiasglet's see14:33
sonne|workiglesiasg: did you check what the status of the k-mean centers is?14:44
iglesiasgsonne|work, uups no14:45
iglesiasgI am doing it right now14:45
iglesiasgit is pretty much ready now14:47
iglesiasgjust a commented SG_UNRED that can be removed14:48
iglesiasgand the fail on the travis build is related to what hushell fixed last wekk I think14:48
-!- benibadman [~benibadma@] has joined #shogun14:48
-!- iglesiasg [] has quit [Quit: Ex-Chat]14:58
shogun-buildbot_build #613 of nightly_default is complete: Failure [failed notebooks]  Build details are at
-!- van51 [] has joined #shogun16:20
van51hello guys16:20
sonne|workhey van51!16:40
van51sonne|work: hello :D16:40
van51sonne|work: I am back, as promised :)16:40
van51I talked to a lecturer here who will be teaching ML next semester and asked him to use shogun in the exercises16:42
van51or show some demos at least16:42
van51and he said he had heard about shogun and is going to participate in the NIPS workshop16:42
van51so, make a good impression!16:42
-!- Saurabh7 [~Saurabh7@] has joined #shogun16:56
sonne|workvan51: HeikoS will :)17:03
sonne|workvan51: ask him what he wants for teaching :)17:03
sonne|workvan51: I think we should put a lot of focus on shogun for teaching for next gsoc17:03
sonne|workas in demos, notebooks17:03
sonne|workbaseline algorithms needed17:03
van51sonne|work: yeah indeed. it would be good to work on those17:07
van51sonne|work: he seemed pretty interested in HMM models17:07
van51sonne|work: when I see him again I will ask him some more17:08
-!- sonne|osx [~sonne@] has joined #shogun17:17
-!- shogun-notifier- [] has joined #shogun17:19
shogun-notifier-shogun-web: Soeren Sonnenburg :master * 8449014 / templates/notebooks.html:
shogun-notifier-shogun-web: add missing '"'17:19
sonne|osxvan51: so when you have time do stuff :)17:22
van51sonne|osx: do you have something in mind now?17:23
sonne|osxvan51: well we need demos / notebooks or you could work on this face gender recognition application17:23
sonne|osxwhat you like more17:23
van51sonne|osx: I like this face/gender recognition application17:24
van51sonne|osx: but I haven't used opencv before17:24
van51sonne|osx: I guess I can learn now17:24
sonne|osxvan51: it is rather easy depending what you want to do17:24
van51sonne|osx: I also have an idea for the kmeans demo but I will need the pending PR merged17:24
sonne|osxvan51: I have some code snippet (which you can find easily on the web too) that detects faces...17:25
van51sonne|osx: will I need to create a dataset for that as well?17:25
sonne|osxvan51: we should maybe aim for sth that does 2 things: a) get the face(s) from an image17:25
sonne|osxb) from a camera17:25
van51sonne|osx: both sound cool17:29
van51sonne|osx: I'll read a bit about it now to get in the spirit17:29
sonne|osxvan51: well for that we just use opencv's detectors17:30
sonne|osxvan51: then the 'real' task starts17:30
sonne|osxto determine the gender of the person17:30
sonne|osxthat is something we might need some manual work for (e.g. by labelling a few hundred faces as male / female)17:31
van51sonne|osx: the detectors will provide us with the faces, right?17:31
van51sonne|osx: no need to work for that17:32
sonne|osx not as good as the one vojtech has but at least open source17:32
van51sonne|osx: ok I got the idea17:33
-!- sonne|osx [~sonne@] has quit [Quit: sonne|osx]17:39
lisitsynwiking: I actually agree this box is a little silly17:41
-!- benibadman [~benibadma@] has quit [Remote host closed the connection]18:02
-!- van51 [] has quit [Quit: Leaving.]18:09
-!- zxtx [~zv@2001:558:6033:150:287e:faea:64e2:d5ec] has joined #shogun18:12
-!- hushell [] has quit [Ping timeout: 245 seconds]18:24
-!- zxtx [~zv@2001:558:6033:150:287e:faea:64e2:d5ec] has quit [Ping timeout: 245 seconds]18:40
-!- hushell [] has joined #shogun18:59
-!- benibadman [] has joined #shogun19:09
-!- benibadman [] has quit [Remote host closed the connection]19:15
-!- benibadman [] has joined #shogun19:16
-!- dsockwell [~twgs@] has quit [Ping timeout: 260 seconds]19:18
-!- benibadm_ [] has joined #shogun19:19
-!- benibadman [] has quit [Ping timeout: 252 seconds]19:21
-!- zxtx [] has joined #shogun19:21
-!- thoralf [] has quit [Quit: Konversation terminated!]19:35
-!- sonne|osx [] has joined #shogun19:54
-!- shogun-buildbot_ [] has quit [Quit: buildmaster reconfigured: bot disconnecting]19:57
-!- shogun-buildbot [] has joined #shogun19:58
@sonney2kshogun-buildbot, force build --branch=develop nightly_default19:58
shogun-buildbotbuild forced [ETA 38m09s]19:58
shogun-buildbotI'll give a shout when the build finishes19:58
-!- shogun-notifier- [] has quit [Quit: transmission timeout]20:19
-!- dsockwell [~twgs@] has joined #shogun20:23
@sonney2kwiking, gives me an internal server error again20:33
-!- hushell [] has quit [Ping timeout: 272 seconds]20:46
-!- hushell [] has joined #shogun20:51
-!- benibadm_ [] has quit [Remote host closed the connection]20:57
-!- benibadman [] has joined #shogun20:58
@sonney2kwiking, not my fault!21:02
-!- benibadman [] has quit [Ping timeout: 272 seconds]21:03
@sonney2klisitsyn, how do you create the markdown->html rendered page?21:03
-!- iglesiasg [] has joined #shogun21:06
-!- mode/#shogun [+o iglesiasg] by ChanServ21:06
@wikingsonney2k: mmm i think it's something to do with yesterday's hickup21:07
shogun-buildbotbuild #614 of nightly_default is complete: Failure [failed notebooks]  Build details are at
@sonney2kiglesiasg, look now it didn't complete after 1hr! 3600s!21:09
@wikingsonney2k: can u try it now21:09
@iglesiasgsonney2k, has someone else tried in his machine?21:09
@sonney2kiglesiasg, don't think so21:11
@sonney2kI guess there is some real issue still21:11
@iglesiasgyes, it must be21:12
-!- shogun-notifier- [] has joined #shogun21:15
shogun-notifier-shogun: parijat :develop * 226cd3d / / (3 files):
shogun-notifier-shogun: initialize cluster centers manually in kmeans21:15
shogun-notifier-shogun: Fernando Iglesias :develop * 6f2b34f / / (3 files):
shogun-notifier-shogun: Merge pull request #1754 from mazumdarparijat/kmeans21:15
shogun-notifier-shogun: initialize cluster centers manually in kmeans21:15
@iglesiasgsonney2k, I am checking the buildbot history21:17
@iglesiasgsonney2k, it must have been any of the changes on the 6th21:17
shogun-buildbotbuild #315 of FCRH - libshogun is complete: Failure [failed configure]  Build details are at  blamelist: parijat <>21:25
shogun-buildbotbuild #316 of FCRH - libshogun is complete: Failure [failed configure]  Build details are at  blamelist: Fernando Iglesias <>21:25
@iglesiasgsonney2k, my guess is that it has to be something coming from this commit
@iglesiasgsonney2k, what do you think?21:31
@sonney2kiglesiasg, I don't see any bug in there21:32
@sonney2kiglesiasg, is the FGM example using sparse features?21:33
@iglesiasgsonney2k, so I have been checking that21:33
@iglesiasgsonney2k, and I don't see it directly in the notebook giving sparse features, although I have not checked it exhaustively21:34
@iglesiasgsonney2k, but I have seen that the hushell's factor graph stuff has sparse stuff in many places21:34
@iglesiasgso, maybe it is used internally?21:34
@sonney2kiglesiasg, but then we should see some valgrind activity21:36
@sonney2kwiking, any chance you can update the template on cloud?21:37
@wikingsonney2k: yeah i just need to sync it first21:37
@wikingas i've got branched since the last push21:37
@sonney2kwiking, well it is just the template ~5 lines of code or so21:38
@sonney2kwiking, we have some issue with nightly_default - we want it to fail when notebooks don't work and not half of the notebooks to be uploaded21:40
@sonney2kor notebook list did shrink due to some failing :/21:40
@sonney2kactually one failing21:40
@wikingsonney2k: easy... set the task a dealbreaker... i.e. buildfailer21:43
@wikingthat way if the notebook generation task has failed then none of the proceeding tasks will be executed21:44
@sonney2kwiking, well how21:45
lisitsynsonney2k: uh some markdown js21:45
lisitsynlet me find21:45
lisitsyn probably this one21:46
@sonney2kwiking, found it haltOnFailure=True21:47
@wikingsonney2k: yep21:47
lisitsynsonney2k: $.get('md/README.markdown', function(data) { var output = markdown.toHTML(data); $("#readme").html(output); });21:48
lisitsynsonney2k: ^ that's how I used it21:48
-!- shogun-buildbot [] has quit [Quit: buildmaster reconfigured: bot disconnecting]22:03
-!- shogun-buildbot [] has joined #shogun22:03
@sonney2kwiking, actually that failure stuff holds for docs too...22:04
-!- hushell [] has quit [Ping timeout: 244 seconds]22:06
-!- shogun-buildbot [] has quit [Client Quit]22:07
-!- shogun-buildbot_ [] has joined #shogun22:07
-!- hushell [] has joined #shogun22:08
@sonney2kiglesiasg, ok I copied the stuff back so we have our 12 notebooks back at least22:15
@sonney2klisitsyn, makes sense not sure yet how to do that with shoguns pages (get won't work the *.md's are not in some static dir22:18
@sonney2k) but I guess I could just put the *.md in a variable and do the same22:18
@iglesiasgsonney2k, good good22:18
@sonney2kiglesiasg, the real fix would be to fix the bug though :/22:19
@iglesiasglet's try to find what is wrong with the FGM notebook in any case22:19
@sonney2ksomeone needs to valgrind22:19
@sonney2khushell, do you have time for that?22:19
@sonney2khushell, the FGM notebook doesn't terminate - even after 1 hour22:19
@iglesiasghushell, oh man you are here! Didn't see you, hey there! :)22:20
shogun-notifier-shogun-web: Soeren Sonnenburg :master * adc094e / shogun/
shogun-notifier-shogun-web: add path to shogun sources`22:21
@sonney2kiglesiasg, if you have time you could do the markdown stuff lisitsyn wrote about - so we can have our in the same way on the homepage22:22
@sonney2kand only maintain *one* readme22:22
@iglesiasgsonney2k, ok, in the home page?22:22
@iglesiasgwe have an extensive readme so I would say not in the home page22:23
@iglesiasgprobably in the about > information page22:24
@sonney2kiglesiasg, well hide it somewhere under some url first and get all the *.md's as pages22:25
@sonney2kthen we have to think about where to put things22:25
@iglesiasgall right22:25
* sonney2k off22:25
shogun-buildbot_build #107 of clang34 - undefined behaviour analysis is complete: Failure [failed test]  Build details are at  blamelist: parijat <>22:27
shogun-buildbot_build #106 of clang34 - thread analysis is complete: Failure [failed test]  Build details are at  blamelist: parijat <>22:42
-!- iglesiasg [] has quit [Quit: Leaving]22:52
shogun-buildbot_build #125 of clang34 - static analysis is complete: Failure [failed analyse]  Build details are at  blamelist: parijat <>23:24
shogun-buildbot_build #108 of clang34 - undefined behaviour analysis is complete: Failure [failed test]  Build details are at  blamelist: Fernando Iglesias <>23:36
shogun-buildbot_build #107 of clang34 - thread analysis is complete: Failure [failed test]  Build details are at  blamelist: Fernando Iglesias <>23:51
--- Log closed Tue Nov 12 00:00:21 2013