Open in new window / Try shogun cloud
--- Log opened Fri Mar 08 00:00:04 2013
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun00:23
shogun-notifier-shogun: Sergey Lisitsyn :master * 4524426 / src/shogun/lib/tapkee/ (9 files): https://github.com/shogun-toolbox/shogun/commit/45244267d1900180dbd5bad75b78a3d3df6cd8d000:23
shogun-notifier-shogun: Update for tapkee00:23
shogun-notifier-shogun:00:23
shogun-notifier-shogun: - Fixed dense eigenembedding method00:23
shogun-notifier-shogun: - Better parameters macro structure00:23
shogun-notifier-shogun: - t-SNE preparation00:23
shogun-buildbotbuild #871 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/871  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>01:17
-!- Netsplit *.net <-> *.split quits: shogun-buildbot01:30
-!- syst3mw0rm [~quassel@aws.aamirkhan.co.in] has quit [Read error: Connection reset by peer]01:30
-!- aamir [~quassel@aws.aamirkhan.co.in] has joined #shogun01:30
-!- FSCV [~FSCV@173.254.212.46] has quit [Ping timeout: 260 seconds]01:38
-!- Netsplit over, joins: shogun-buildbot01:39
-!- FSCV [~FSCV@204.45.132.11] has joined #shogun01:45
-!- FSCV [~FSCV@204.45.132.11] has quit [Quit: Leaving]02:14
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]03:23
shogun-buildbotbuild #315 of nightly_default is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/31503:37
-!- hoijui [~hoijui@dslb-092-078-043-220.pools.arcor-ip.net] has joined #shogun07:35
-!- in3xes [~in3xes@117.200.6.104] has joined #shogun08:12
-!- in3xes [~in3xes@117.200.6.104] has quit [Remote host closed the connection]08:55
@sonney2kblackburn, seen this: http://shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/871/steps/test%20python_modular/logs/stdio ?09:39
@sonney2kblackburn, some crash in converter_laplacianeigenmaps_modular09:39
@sonney2kwiking, what is CCCSOSVM?09:56
@sonney2kConvexConcaveProcedureStructuredOutputSupportVectorMachine ?09:56
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun10:25
shogun-notifier-shogun: Soeren Sonnenburg :master * 07843b7 / src/shogun/ (9 files): https://github.com/shogun-toolbox/shogun/commit/07843b77b65745eb6045b798e3c3024fa71aae0e10:25
shogun-notifier-shogun: add missing doxygen documentation10:25
shogun-notifier-shogun: Soeren Sonnenburg :master * 9c1840b / tests/unit/Makefile: https://github.com/shogun-toolbox/shogun/commit/9c1840be49d03d5751719dd43f9e630993f4cdd610:27
shogun-notifier-shogun: fix ar warning10:27
shogun-notifier-shogun: Soeren Sonnenburg :master * efed639 / src/shogun/kernel/string/ (2 files): https://github.com/shogun-toolbox/shogun/commit/efed63958cb02e05eff329c4785bd7b08d2c2cd110:30
shogun-notifier-shogun: remove unstable warning about oligo/SLIK kernels10:30
blackburnsonney2k: okay nice I'll take a look10:41
blackburnsonney2k: nice naming, I'd like to put BMRM here to10:42
blackburnBMRMCCCSOSVM10:42
@sonney2kblackburn, I am not 100% sure what CCCSOSVM means / does and wiking didn't document a bit10:48
blackburnha! google now has street view in my town10:48
@sonney2kblackburn, only in mine it will never have studpid germans...10:49
blackburnlet me show you the house I live :)10:49
blackburnsonney2k: https://maps.google.com/?ll=53.546115,49.331797&spn=0.000928,0.0025&t=h&z=19&layer=c&cbll=53.546226,49.331826&panoid=ROzbUHf5vDlDqCSRWQJxdg&cbp=12,104.39,,1,-0.6310:49
blackburnthat one far away just near the trees10:49
@sonney2kblackburn, I've fixed (hopefully) the remaining warnings. now you need to 'only' get tapkee not to crash and we can release10:50
blackburnsonney2k: okay sure I want it not to crash too :)10:50
@sonney2kblackburn, the white one?10:50
@sonney2kstraight or left?10:50
blackburnsonney2k: straight, covered by trees10:50
blackburnsonney2k: below the cloud10:51
blackburnsonney2k: all the city is made of blocks - they didn't enter these blocks though10:52
blackburnsonney2k: we also have factory haha https://maps.google.com/?ll=53.556397,49.253522&spn=0.001683,0.009999&t=h&z=17&layer=c&cbll=53.556397,49.253517&panoid=EBsYw10RYol_l7WV_2C78Q&cbp=11,357.51,,0,-2.810:55
blackburnlook - people are making cars and use them to get to the factory10:55
blackburnwhere they are making more cars10:55
blackburn:D10:55
@sonney2kblackburn, haha10:56
@sonney2kblackburn, this is where I grew up - you have to look up to see the actual building10:56
@sonney2khttps://maps.google.com/maps?q=Fischerinsel,+Berlin,+Germany&hl=en&ll=52.512864,13.40748&spn=0.000301,0.002064&sll=53.556397,49.253517&sspn=0.005213,0.016512&oq=berlin+fischeri&t=h&hnear=Fischerinsel,+10179+Berlin,+Germany&z=20&layer=c&cbll=52.512864,13.40748&panoid=97DicG38TF-kABjp270hDg&cbp=11,222.04,,0,2.1610:56
blackburnsonney2k: looks like a nice place10:57
-!- heiko [~heiko@nat-176-119.internal.eduroam.ucl.ac.uk] has joined #shogun10:57
@sonney2kor look behind you to get an impression how these buildings look like :)10:57
blackburnsonney2k: pretty tall yes10:57
blackburnsonney2k: 22 floors?10:58
blackburnsonney2k: you won't impress me that much as I work in these building: https://maps.google.com/?ll=53.203237,50.155871&spn=0.001697,0.009999&t=h&z=17&layer=c&cbll=53.203508,50.15489&panoid=neRCH6sWeTFSJOLcJBCe2Q&cbp=11,146.75,,0,-34.6710:59
blackburnsonney2k: 27 floors and actually I can see whole city from mine 13th :)11:00
blackburnsonney2k: http://cs7009.vk.me/c407119/v407119002/612d/ia9K30M0pa4.jpg angela is happy with her new phone11:04
blackburnsonney2k: I should tell you a funny story about openmp11:06
blackburnsonney2k: I was using clock() to measure time11:06
blackburnsonney2k: and I tried so hard to make openmp implementation fast11:07
blackburnbut what I was getting is nearly the same time11:07
blackburnand just accidentally I realized what I was measuring11:08
shogun-buildbotbuild #872 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/872  blamelist: Soeren Sonnenburg <sonne@debian.org>11:09
blackburnargh11:18
blackburnI cannot reproduce the error11:18
shogun-buildbotbuild #873 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/873  blamelist: Soeren Sonnenburg <sonne@debian.org>11:22
blackburnheiko: sonney2k: could you please run converter_laplacianeigenmaps_modular.py ?11:23
heikoblackburn: with the latest version or is the one from a few days ago fine?11:30
blackburnheiko: hmm if it'd take time just run the one you have now11:30
blackburnheiko: I just can't reproduce the error11:31
heikoalready updated :) compiling ... will let you know11:31
blackburnthanks11:32
blackburnheiko: btw about openmp story I told to soeren11:32
blackburnheiko: I easily got 2-3x speedup here11:33
heikowhat compared to what?11:33
blackburnheiko: so you may try that too11:33
blackburnheiko: compared to serial11:33
heikoI see11:33
heikommh11:33
blackburnheiko: with one line11:33
blackburnwell two11:33
heikooh11:33
heikothat sounds good11:33
heikohowever, 2-3x doesnt really matter for me11:33
heikoI need 2-3k :)11:33
blackburnheiko: do you have 2-3k cores?11:33
heikoyes11:34
heiko(not on my laptop though)11:34
blackburnheiko: ehm well it should scale11:34
heikothe cores are on different machines11:34
heikolargest machine has 48 or something11:34
blackburnheiko: 2-3x is because I have intel core 2 with hyperthreading11:34
blackburnheiko: then you could get 4811:34
heikono, then I dont move in the cluster queue11:34
heikoit is better to ask for 48 cores which can be located on any machine11:35
blackburnheiko: anyway I can teach you some openmp if you want11:35
heikoyou get this very fast, while asking for one complete machine takes ages11:35
blackburnheiko: the only problem is that Soeren didn't want to have both pthreads and openmp at the same time11:35
heikoblackburn, yes please11:35
heikoI see11:35
heikoblackburn: but openmp should be easier right?11:35
heikopthreads is a pain11:35
blackburnheiko: yes adding one pragma11:36
blackburnheiko: clang doesn't support openmp properly though11:36
heikoblackburn: cool lets talk about that, but lets do it later, I have do do other things now :)11:36
blackburnbut pragmas are effectively ignored11:36
blackburnheiko: if you want an example11:36
heikoblackburn: mmmh would that hurt us? with clang?11:36
blackburnlet me show and then lets get back to that later11:36
blackburnheiko: no, pragmas are ignored if they are not supported11:36
blackburnhttps://github.com/lisitsyn/tapkee/blob/master/tapkee/routines/isomap.hpp11:37
blackburnheiko: all I had to add is line 44  and line 5111:37
blackburnheiko: and it became parallel11:37
heikoblacburn, thats cool!11:37
heikoso that is like parfor in matlab11:37
blackburnno idea about parfor11:38
blackburnparallel for?11:38
heikoyes11:38
heikoof independent things11:38
blackburnheiko: I'd say much more powerful11:38
blackburnheiko: openmp supports not only independent things11:38
blackburnheiko: a variety of locks is supported too11:39
heikoblackburn: cool11:39
blackburnheiko: https://github.com/lisitsyn/tapkee/blob/master/tapkee/routines/locally_linear.hpp#L13811:39
heikoI will get back to you on this tonight11:39
heikoto speedup some MMD things :)11:39
blackburnheiko: that's how you make only one thread doing something11:39
blackburn#pragma omp critical11:40
blackburn{ ... }11:40
heikogotta go now, sorry11:40
blackburnsure11:40
blackburnsonney2k: I am having the same difficulties with io that heiko had11:42
blackburnI see no messages11:43
heikoblackburn11:43
heikopython converter_laplacianeigenmaps_modular.py11:43
heikoLaplacianEigenmaps11:43
heikopython: /usr/include/eigen3/Eigen/src/Core/Block.h:320: Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Block(XprType&, Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Index, Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Index, Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Index, Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Index) [with XprType = const Eig11:43
heikoAborted (core dumped)11:43
-!- n4nd0 [~nando@n145-p185.kthopen.kth.se] has joined #shogun11:44
heikowith latest git11:44
blackburnheiko: hmm okay11:44
blackburnheiko: do you have arpack?11:44
heikoblackburn:  yes11:45
blackburnheiko: okay thanks I'll take a look11:45
shogun-notifier-shogun: Soeren Sonnenburg :master * 688da26 / src/shogun/regression/gp/ (3 files): https://github.com/shogun-toolbox/shogun/commit/688da262f28ef35ef7c6ea7a50893b4d88f6f90111:59
shogun-notifier-shogun: a few more doxygen fixes11:59
blackburnsonney2k: level_above is incorrect12:01
@sonney2kblackburn, -v please?12:02
blackburnsonney2k: I replaced its impl with return true and it worked12:03
blackburnfor more information please add -h flag12:03
@sonney2k./blackburn -h12:03
blackburnsonney2k: I tried to output more information when debugging my stuff and got no output12:03
@sonney2kohh loglevel >= type!12:03
@sonney2knot > type12:03
blackburnhah12:04
blackburnokay12:04
@sonney2kfixing12:04
shogun-notifier-shogun: Soeren Sonnenburg :master * 9cd8b07 / src/shogun/io/SGIO.h: https://github.com/shogun-toolbox/shogun/commit/9cd8b070ae6ca0a45ba5ae881cf841fb23e815bf12:04
shogun-notifier-shogun: fix logging of functions12:04
blackburnsonney2k: may be we should add explicit values to enum12:04
@sonney2kheiko, blackburn, we should be down to 1 warning on python_modular again12:05
@sonney2kthat's as good as it gets12:05
blackburnsonney2k: yes sure, I'll take care of it once I get some tapkee things fixed12:05
@sonney2kblackburn, no we are already I meant12:05
blackburnah12:06
blackburnokay that's even better12:06
blackburn:D12:06
@sonney2kblackburn, for logging enum yes sure just do it12:06
blackburnsonney2k: I will check if it works now12:06
@sonney2kheiko, blackburn - I would like to run generator.py again - all the old tests won't work any longer due to our CLabels -> CBinaryLabels etc changes12:08
@sonney2kso serialization is b0rken anyways12:08
@sonney2kblackburn, for that to work I need you example to not crash though...12:09
blackburnsonney2k: it is *exactly* what I am trying to fix12:10
blackburnsonney2k: could you please run it too?12:11
blackburnconverter_laplacian12:11
@sonney2kblackburn, compiling again...12:12
blackburnsonney2k: it now output *only* SG_INFO12:12
blackburnif SG_INFO is set12:12
@sonney2kMSG_INFO you mean?12:18
blackburnsonney2k: yeah12:18
blackburnsonney2k: still doesn't work12:18
blackburnsonney2k: I set it to MSG_INFO and what I get is *only* info12:19
blackburnI set it to debug and get only debug..12:19
@sonney2kweird12:21
blackburnsonney2k: btw our CTime is incorrect for openmp too12:21
blackburnas clock() will measure everything that was ticked by say 4 CPUs12:22
blackburnmultiplying the time by 412:22
@sonney2kblackburn, no ctime is not incorrect12:25
@sonney2kCTime can measure both wall clock and cpu time12:25
blackburnsonney2k: ah okay there is a method using gettimeofday12:26
shogun-notifier-shogun: Sergey Lisitsyn :master * f9c0960 / src/shogun/lib/tapkee/ (13 files): https://github.com/shogun-toolbox/shogun/commit/f9c0960a2447773e427d210092eafc259b7786b312:31
shogun-notifier-shogun: A few fixes for Tapkee:12:31
shogun-notifier-shogun:12:31
shogun-notifier-shogun: - Correct implementation for NPE, LLTSA and LPP12:31
shogun-notifier-shogun: - Parallel implementation of isomap12:31
shogun-notifier-shogun: - Parallel implementation of lle12:31
@sonney2kblackburn,12:31
@sonney2k$ python converter_laplacianeigenmaps_modular.py12:31
@sonney2kLaplacianEigenmaps12:31
@sonney2kpython: /usr/include/eigen3/Eigen/src/Core/Block.h:320: Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Block(XprType&, Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Index, Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Index, Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Index, Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Index) [with XprType = const Eigen::Matrix<double, -1, 1>;12:31
@sonney2k int BlockRows = -1; int BlockCols = 1; bool InnerPanel = false; Eigen::Block<XprType, BlockRows, BlockCols, InnerPanel, true>::Index = long int]: Assertion `startRow >= 0 && blockRows >= 0 && startRow + blockRows <= xpr.rows() && startCol >= 0 && blockCols >= 0 && startCol + blockCols <= xpr.cols()' failed.12:31
@sonney2kAborted12:31
blackburnsonney2k: is that with debug enabled?12:32
blackburnsonney2k: could you please paste the output of gdb when it gets to the error position12:32
shogun-notifier-shogun: Soeren Sonnenburg :master * ce8fd33 / src/shogun/io/SGIO.h: https://github.com/shogun-toolbox/shogun/commit/ce8fd330e5ce20a6188fdfb61faa84cefffa8e9b12:34
blackburnohhhh sh12:34
shogun-notifier-shogun: fix loglevel_above check12:34
-!- travis-ci [~travis-ci@ec2-23-20-137-122.compute-1.amazonaws.com] has joined #shogun12:35
travis-ci[travis-ci] it's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/533995212:35
-!- travis-ci [~travis-ci@ec2-23-20-137-122.compute-1.amazonaws.com] has left #shogun []12:35
blackburndamn12:35
@sonney2kblackburn, the backtrace is long!12:35
blackburnsonney2k: yeah it should be12:35
blackburnlast commit was stupid12:35
@sonney2khttp://dpaste.com/1016011/12:36
blackburnsonney2k: thanks!12:36
@sonney2kblackburn, ?12:36
blackburnI broke the build argh12:37
blackburnsonney2k: do you mind my adding '-Wno-unknown-pragmas' key/12:39
-!- travis-ci [~travis-ci@ec2-54-242-66-179.compute-1.amazonaws.com] has joined #shogun12:39
travis-ci[travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/534005112:39
-!- travis-ci [~travis-ci@ec2-54-242-66-179.compute-1.amazonaws.com] has left #shogun []12:39
blackburnsonney2k: I reproduced the error!12:40
blackburnwill fix soon then12:41
@sonney2kblackburn, what does -Wno-unknown-pragmas' do?12:42
blackburnsonney2k: ignores unknown pragmas (openmp)12:42
@sonney2kif it still builds with that why not12:42
shogun-notifier-shogun: Sergey Lisitsyn :master * 2b119cc / src/configure: https://github.com/shogun-toolbox/shogun/commit/2b119cc4db723c7d8bad2eaa1062e8832aead13812:43
shogun-notifier-shogun: Added -Wno-unknown-pragmas key to ignore openmp pragmas12:43
blackburnsonney2k: catched the error12:44
@sonney2kblackburn, so waht is it?12:46
blackburnsonney2k: minor thing12:46
blackburnhopefully already fixed, compiling and testing12:46
-!- travis-ci [~travis-ci@ec2-54-242-66-179.compute-1.amazonaws.com] has joined #shogun12:47
travis-ci[travis-ci] it's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/534015412:47
-!- travis-ci [~travis-ci@ec2-54-242-66-179.compute-1.amazonaws.com] has left #shogun []12:47
n4nd0blackburn: the includes in tapkee should be fixed too or not?12:49
blackburnn4nd0: hah yes sure, I fixed that, that was broken commit of mine12:49
shogun-buildbotbuild #874 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/874  blamelist: Soeren Sonnenburg <sonne@debian.org>12:49
n4nd0blackburn: can you do it? otherwise I can, there are also a few other files that need the path changed12:51
blackburnn4nd0: I did that already, will commit in a minute12:51
blackburnjust need to test it12:51
n4nd0cool!12:51
blackburnreally minor thing - wrong segment of eigenvalues12:53
n4nd0blackburn: what?12:54
blackburnn4nd0: laplacian eigenmaps skip one eigenvalue12:55
blackburnhere we go12:56
blackburnshoulda work now12:56
shogun-notifier-shogun: Sergey Lisitsyn :master * 657462c / src/shogun/lib/tapkee/ (7 files): https://github.com/shogun-toolbox/shogun/commit/657462ce433c7edb0a5e1f4f0039774c99e4a4e612:56
shogun-notifier-shogun: Fixed includes in tapkee and fixed wrong segment of eigenvalues in12:56
shogun-notifier-shogun: generalized eigenembedding12:56
shogun-buildbotbuild #875 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/875  blamelist: Soeren Sonnenburg <sonne@debian.org>12:59
-!- travis-ci [~travis-ci@ec2-23-20-137-122.compute-1.amazonaws.com] has joined #shogun13:00
travis-ci[travis-ci] it's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/534037513:00
-!- travis-ci [~travis-ci@ec2-23-20-137-122.compute-1.amazonaws.com] has left #shogun []13:00
shogun-buildbotbuild #876 of deb3 - modular_interfaces is complete: Failure [failed compile libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/876  blamelist: Soeren Sonnenburg <sonne@debian.org>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com>13:05
shogun-buildbotbuild #751 of deb2 - static_interfaces is complete: Failure [failed compile libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/751  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>13:06
shogun-buildbotbuild #752 of deb2 - static_interfaces is complete: Failure [failed compile libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/752  blamelist: Soeren Sonnenburg <sonne@debian.org>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com>13:07
n4nd0blackburn: any problem with travis and Eigen?13:09
n4nd0../shogun/lib/tapkee/tapkee.hpp:55:2: error: 'initParallel' is not a member of 'Eigen'13:10
blackburnahhhh13:10
n4nd0it compiled fine in my box13:10
shogun-buildbotbuild #277 of ubu1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/ubu1%20-%20libshogun/builds/277  blamelist: Soeren Sonnenburg <sonne@debian.org>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com>13:10
blackburnn4nd0: old eigen didn't have it13:10
shogun-notifier-shogun: Sergey Lisitsyn :master * f288268 / src/shogun/lib/tapkee/tapkee.hpp: https://github.com/shogun-toolbox/shogun/commit/f28826896f6de0738ce845b982f9162204fc39e513:10
shogun-notifier-shogun: Commented out initParallel of eigen in tapkee13:10
n4nd0blackburn: hehe it was not used anyway?13:11
blackburnn4nd0: yeah13:11
n4nd0blackburn: yeah==not used?13:12
n4nd0:D13:12
blackburnn4nd0: yes, relevant for standalone but not for shogun13:12
n4nd0ook13:12
blackburnshould compile now13:12
n4nd0blackburn: yeah, nice build now :)13:19
shogun-buildbotbuild #753 of deb2 - static_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/75313:28
wikingheiko: here?13:30
wikingheiko: FYI: http://code.google.com/p/googletest/wiki/AdvancedGuide#Selecting_Tests13:34
wikingheiko: so you can select which unit test u wanna run, e.g. ./shogun-unit-test --gtest_filter=QuadraticTimeMMD.*13:34
wikingin tests/unit13:34
wikingheiko: what we can do is that if you do make unit-tests <extra argument>13:35
wikingthen the <extra argument> is passed along as a parameter for --gtest_filter13:35
shogun-buildbotbuild #278 of ubu1 - libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/ubu1%20-%20libshogun/builds/27813:39
@sonney2kwiking, what is CCCSOSVM ?13:39
@sonney2kcould you please add some doxygen explanation what it is13:40
@sonney2kwhat it can be used for etc13:40
wikingsonney2k: surething13:40
wikingsonney2k: it's just another sosvm using mosek13:40
@sonney2kand the CC what does it stand for?13:41
wikinghad no better name for it13:41
@sonney2kconvex concave?13:41
@sonney2kor what?13:41
wikingyeah it's convexconcave alright ;)13:42
@sonney2kwiking, then please update the dox13:45
@sonney2kwiking, we want to release today13:45
wikingtodaaay?13:45
wikingok13:45
n4nd0sonney2k: there are some warnings in travis that don't apear in the buildbot, should we worry about them?13:47
n4nd0comparison between signed and unsigned integer expressions in tapkee most of them I think13:47
blackburnsonney2k: n4nd0: appears with old eigen3 - I can fix that if you want13:48
blackburncan try to fix that*13:48
n4nd0blackburn: I think it is fine13:49
shogun-notifier-shogun: Viktor Gal :master * 904d75c / src/shogun/structure/CCSOSVM.h: https://github.com/shogun-toolbox/shogun/commit/904d75ccc675ad51e1152f9328f62e378035063313:52
shogun-notifier-shogun: Add some comments for CCSOSVM13:52
shogun-notifier-shogun: Viktor Gal :master * 4a4d217 / src/shogun/structure/CCSOSVM.h: https://github.com/shogun-toolbox/shogun/commit/4a4d217742b88b89ad220eff74ccb51898fe2f8b13:52
shogun-notifier-shogun: Merge pull request #910 from vigsterkr/master13:53
shogun-notifier-shogun:13:53
shogun-notifier-shogun: Add some comments for CCSOSVM13:53
n4nd0wiking: do they explain the convex concave procedure in Cutting-Plane Training of Structural SVMs??13:55
n4nd0mmm I think the method in there is PrimalMosekSOSVM13:56
wikingn4nd0: not exactly13:56
wikingn4nd0: almost13:56
n4nd0wiking: aham! sorry then :)13:56
wikingn4nd0: this one is similar to Dual...13:56
wikingthe one based on BMRM13:57
n4nd0I see13:57
wikingbut this one uses mosek atm13:57
n4nd0ok13:57
wiking!@#13:59
shogun-notifier-shogun: Soeren Sonnenburg :master * 5d4172d / src/interfaces/python_modular/ (2 files): https://github.com/shogun-toolbox/shogun/commit/5d4172dee86ec7fa80640cf43e0b0324e0d2b23713:59
shogun-notifier-shogun: remove NO_IMPORT define and numpy include in sg_print_functions.cpp13:59
shogun-notifier-shogun:13:59
shogun-notifier-shogun: this fixes the long standing warning:13:59
shogun-notifier-shogun: 'int _import_array()' defined but not used13:59
shogun-buildbotbuild #877 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/877  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>13:59
shogun-buildbotbuild #607 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/607  blamelist: Viktor Gal <viktor.gal@maeth.com>14:00
wikingblackburn: sublimetext3 is awesome imo14:01
n4nd0I should give a shot to sublime maybe14:03
wikingn4nd0: the best for me is that it has full vim keymapping suppor ;P14:08
n4nd0wiking: that is very cool indeed14:09
n4nd0the only reason I am resisting to try it out is because I like vim :D14:09
wikingn4nd0: hheh it has a package called vintage14:09
wikingn4nd0: if u enable it u get everything that vim offers14:10
wikingi mean keymapping wise14:10
wikinglike 3dd14:10
wikingand 3dw and stuff like that14:10
wiking(as well as switching between command and insert mode14:11
n4nd0nice14:11
n4nd0I will give it a shot indeed14:11
n4nd0blackburn, sonney2k : is it possible to get access to the project descriptions of last GSoC?14:12
n4nd0nvm14:13
shogun-buildbotbuild #608 of cyg1 - libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/60814:14
wikingrpi is great btw14:18
blackburnwiking: I will give it a shot14:18
blackburnwiking: what about clang_complete?14:19
wikingblackburn: get this: https://github.com/quarnster/SublimeClang14:19
wikingblackburn: install it as a package14:19
blackburnn4nd0: yeah I will check and let you know if I found anything14:19
blackburnwiking: what I am most confused is price :D14:19
wikingblackburn: and u r good to go14:19
wikingblackburn: well u can use it for free14:19
wikingblackburn: after n number of saves it'll tell you to buy it14:20
wikingand u click no thnx14:20
blackburnwiking: heh I see14:20
wikingand for sure you can get a license from somewhere if u r really looking for it14:20
wiking:P14:20
n4nd0blackburn: don't worry about it14:21
n4nd0the project ideas I mean14:22
wikingblackburn: n4nd0 there are the packages i found essential: Git14:22
wikingGitGutter14:22
wikingLaTeXTools314:22
wikingPackageControl14:22
wikingSublimeClang14:22
wikingSublimeLinter14:22
wikingsublime-jsdocs14:22
wikingthere are for ST3 b14:22
wikingbut you have the same ones for ST2 as well14:23
@sonney2kwiking, why get trapped in closed source software14:23
@sonney2kno thanks14:23
wikingsonney2k: mmm afaik it's a matter of opinion... i found this one really useful. there were some thing i really missed from vim and st really fill up that part.14:24
wikings/afaik/imo/g14:24
wiking;P14:25
shogun-notifier-shogun: Soeren Sonnenburg :master * ad1a6e1 / src/shogun/classifier/svm/WDSVMOcas.cpp,src/shogun/structure/CCSOSVM.cpp: https://github.com/shogun-toolbox/shogun/commit/ad1a6e1f128d4c79d0de2912e221a8319effa1f114:27
shogun-notifier-shogun: fix a minor warnings when pthreads are not available14:27
@sonney2kwiking, eclipse is doing that too :P14:28
wikingsonney2k: hahha but that is damnslow :S14:29
blackburnI am actually used to use vim in console (guake) and sublime wont' be nice for that14:33
shogun-buildbotbuild #878 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/878  blamelist: Viktor Gal <viktor.gal@maeth.com>14:38
n4nd0sonney2k, blackburn : it seems that Christoph is interested in the project! :)14:43
@sonney2kn4nd0, what did you propose to him?14:43
n4nd0he is going to let me know within a few days some ideas for a GSoC project14:43
@sonney2kahh so he is thinking about sth14:44
n4nd0yeah14:44
shogun-buildbotbuild #879 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/879  blamelist: Viktor Gal <vigsterkr@gmail.com>14:48
@sonney2kblackburn, looks like we still have an terminate called after throwing an instance of 'tapkee::eigendecomposition_error'14:49
@sonney2kerror14:49
@sonney2kblackburn, this crash / error doesn't happen all the time15:11
@sonney2kis this running parallel / using randomized stuff?15:11
@sonney2kline 1695 in StringFeatures.cpp:     m_parameters->add_vector(&symbol_mask_table, &symbol_mask_table_len, "mask table", "fuck you");15:15
n4nd0sonney2k: wtf?? hahaha15:16
n4nd0sonney2k: how did you find that btw?? grepping fuck in src/shogun :)15:17
@sonney2kn4nd0, attempting to f*x a bug :)15:19
shogun-buildbotbuild #880 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/880  blamelist: Soeren Sonnenburg <sonne@debian.org>15:20
-!- n4nd0 [~nando@n145-p185.kthopen.kth.se] has quit [Quit: leaving]15:38
@sonney2kblackburn, that is in converter_laplacianeigenmaps_modular still15:40
blackburnsonney2k: oh :(15:52
blackburnokay I'll fix that15:52
blackburnsonney2k: "fuck you" is a great message for documentation15:55
shogun-notifier-shogun: Sergey Lisitsyn :master * b560a52 / examples/undocumented/python_modular/converter_laplacianeigenmaps_modular.py: https://github.com/shogun-toolbox/shogun/commit/b560a521af3c165abb4c2fb42b01101cd0f013f715:55
shogun-notifier-shogun: Adjusted width to avoid errors in laplacian eigenmaps15:55
blackburnoooops it is me15:56
blackburn:D15:56
shogun-notifier-shogun: Soeren Sonnenburg :master * 5612279 / examples/undocumented/python_modular/ (20 files): https://github.com/shogun-toolbox/shogun/commit/56122795922c3109fc2fff4c7b5cab748227c0ed15:56
shogun-notifier-shogun: fix various examples15:56
shogun-notifier-shogun:15:56
shogun-notifier-shogun: - to not print anything15:56
shogun-notifier-shogun: - wrong function names (or file name euclidian vs euclidean)15:56
shogun-notifier-shogun: Soeren Sonnenburg :master * a0f490e / src/shogun/multiclass/ShareBoost.cpp: https://github.com/shogun-toolbox/shogun/commit/a0f490e32e3c6b3d645e7a102effe627e98e731915:56
shogun-notifier-shogun: shareboost: don't print progress if not in debug mode15:56
shogun-notifier-shogun: Soeren Sonnenburg :master * d0654a4 / src/shogun/features/StringFeatures.cpp: https://github.com/shogun-toolbox/shogun/commit/d0654a4d5b17cc40b58ac27d338f7f717e4f955415:56
shogun-notifier-shogun: fix serialization of mask table15:56
shogun-notifier-shogun: Soeren Sonnenburg :master * 2f3cbef / src/shogun/classifier/svm/WDSVMOcas.cpp: https://github.com/shogun-toolbox/shogun/commit/2f3cbef4d541c4e4d7c6ee9350cf7514f3a9a05815:56
shogun-notifier-shogun: move more code into #ifdef PTHREAD15:56
blackburnsonney2k: I'm sorry :D15:58
blackburnokay we have no more 'fuck' word15:59
@sonney2kblackburn, so please remove the crasher...15:59
blackburnsonney2k: already15:59
@sonney2kheiko, how do we get evaluation_cross_validation_mkl_weight_storage.py to not output anything15:59
blackburnhttps://github.com/shogun-toolbox/shogun/commit/b560a521af3c165abb4c2fb42b01101cd0f013f715:59
@sonney2kI've removed the print statements but still16:00
heikosonney2k: let me check16:00
@sonney2kblackburn, we still have crashers...16:00
@sonney2kbut in the converter_kernel* thing16:00
@sonney2k*sometimes*16:00
heikosonney2k,     cross_validation.add_cross_validation_output(CrossValidationPrintOutput())16:02
heikothis line causes x-validation to output things16:02
heikothis is the point of the example16:02
heikoremoving it removes outputs16:02
wikingheiko: check logs16:03
@sonney2kheiko, kthx16:03
wikingsonney2k: so switching to clang on bsd bot?16:03
heikosonney2k: say do you know how to import cmake projects in eclipse? I have one where the source.cpp file is at a different folder than the binary and I cannot get hem to work in one single project16:03
shogun-notifier-shogun: Sergey Lisitsyn :master * 7801782 / examples/undocumented/python_modular/converter_laplacianeigenmaps_modular.py: https://github.com/shogun-toolbox/shogun/commit/780178247a0705f89b1ce74e2b6c640217f5344316:03
shogun-notifier-shogun: Correct fix for LA16:03
@sonney2kheiko, no idea16:03
blackburnsonney2k: which one converter_kernel?16:03
heikowiking, nice!16:04
heikothats what I needed16:04
wikingheiko: i could say: rtfm ;)16:04
heikowiking, come on :D16:05
wikingheiko: but i didn't ;)16:05
heikowiking, I tend to be annoyed by those things, I prefer to ask people who know stuff ;)16:05
heikoanyway, thanks!16:05
wikingheiko: lol16:05
wikingyeah i've tested it it works16:05
wikingjust as i said16:05
wikingyou'll have problems16:06
wikingwith the LD_PRELOAD missing16:06
wikingif you call it with ./shogun...16:06
shogun-buildbotbuild #610 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/610  blamelist: Soeren Sonnenburg <sonne@debian.org>16:06
wikingbecause in that case it'll use your shogun lib that is installed on the system not the latest build that you wanna test16:06
wikingso be aware of that16:06
@sonney2kheiko, these are for you:16:07
@sonney2kstatistics_hsic                                              ERROR generating '' using 'statistics_hsic.py'16:07
@sonney2k'module' object has no attribute 'parameter_list'16:07
@sonney2kstatistics_kmm                                               ERROR generating '' using 'statistics_kmm.py'16:07
@sonney2k'module' object has no attribute 'parameter_list'16:07
@sonney2kstatistics_linear_time_mmd                                   ERROR generating '' using 'statistics_linear_time_mmd.py'16:07
@sonney2k'module' object has no attribute 'parameter_list'16:07
@sonney2kstatistics_mmd_kernel_selection                              ERROR generating '' using 'statistics_mmd_kernel_selection.py'16:07
@sonney2k'module' object has no attribute 'parameter_list'16:07
@sonney2kstatistics_quadratic_time_mmd                                ERROR generating '' using 'statistics_quadratic_time_mmd.py'16:07
@sonney2k'module' object has no attribute 'parameter_list'16:07
@sonney2kheiko, please add some reasonable parameter_list16:07
heikosonney, these are the integration tests right?16:07
heikosonney2k: ok will do this evening16:07
@sonney2kyes16:08
shogun-notifier-shogun: Soeren Sonnenburg :master * 655fc3d / examples/undocumented/python_modular/evaluation_cross_validation_mkl_weight_storage.py: https://github.com/shogun-toolbox/shogun/commit/655fc3dc4e0e2bec427d1e481882777bb29f245316:10
shogun-notifier-shogun: make x-val example non-verbose16:10
@sonney2kheiko, I've found another crasher in vw/streaming stuff: http://dpaste.com/1016167/16:12
heikosonney2k: add to the issue, if we have collected enough material, it will be easier to solve16:13
shogun-buildbotbuild #611 of cyg1 - libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/61116:21
-!- hoijui [~hoijui@dslb-092-078-043-220.pools.arcor-ip.net] has quit [Read error: Operation timed out]16:21
shogun-buildbotbuild #881 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/881  blamelist: Soeren Sonnenburg <sonne@debian.org>16:33
blackburnsonney2k: I'm totally back now16:48
shogun-buildbotbuild #612 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/612  blamelist: Soeren Sonnenburg <sonne@debian.org>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com>16:48
blackburnwhat's the status?16:48
-!- FSCV [~FSCV@173.254.212.46] has joined #shogun16:53
shogun-buildbotbuild #882 of deb3 - modular_interfaces is complete: Failure [failed test ruby_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/882  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>16:56
shogun-buildbotbuild #883 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/883  blamelist: Soeren Sonnenburg <sonne@debian.org>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com>17:08
shogun-notifier-shogun: Sergey Lisitsyn :master * 6abf433 / examples/undocumented/python_modular/so_multiclass.py: https://github.com/shogun-toolbox/shogun/commit/6abf4337b902c3226f4639c28046a9d16ce5c21017:15
shogun-notifier-shogun: Fixed SO multiclass example17:15
shogun-notifier-shogun: Sergey Lisitsyn :master * fb68dbc / src/shogun/converter/ (12 files): https://github.com/shogun-toolbox/shogun/commit/fb68dbcbc7514c7a572bfcc2d24184339624a84b17:24
shogun-notifier-shogun: Some fixes for converters doc17:24
shogun-buildbotbuild #613 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/613  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>17:33
shogun-notifier-shogun: Sergey Lisitsyn :master * 116a810 / src/shogun/classifier/svm/WDSVMOcas.cpp: https://github.com/shogun-toolbox/shogun/commit/116a810cd33b91dc72cd0f6748ca792f5b35ee2017:38
shogun-notifier-shogun: An attempt to fix wdsvmocas17:38
shogun-buildbotbuild #614 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/614  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>17:41
shogun-buildbotbuild #884 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/884  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>18:04
shogun-buildbotbuild #885 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/885  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>18:15
shogun-buildbotbuild #762 of deb2 - static_interfaces is complete: Failure [failed test octave_static]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/762  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>18:15
shogun-buildbotbuild #763 of deb2 - static_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/76318:22
shogun-buildbotbuild #615 of cyg1 - libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/61518:31
@sonney2kblackburn, do we need to include libbmrm in the modular interfaces?18:46
blackburnsonney2k: no, I don't think so18:46
@sonney2kfixing...18:47
@sonney2kblackburn, what about libppbm and libp3bm?18:47
blackburnsonney2k: the same.. :)18:47
-!- n4nd0 [~fjig@orange-03.csc.kth.se] has joined #shogun18:48
shogun-notifier-shogun: Soeren Sonnenburg :master * ecc4a61 / src/interfaces/modular/ (3 files): https://github.com/shogun-toolbox/shogun/commit/ecc4a61c095b41ec7293470db5fdb6a6663a034418:50
shogun-notifier-shogun: remove unused (but troublesome) includes of libbmrm, libppbm and libp3bm18:50
@sonney2kblackburn, about this SGVector warning here18:50
@sonney2khttp://shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/886/steps/compile%20ruby_modular/logs/warnings%20%284%2918:50
@sonney2kI am tempted to just remove the static function18:50
@sonney2kwhat do you think?18:50
blackburnsonney2k: what does it do?18:51
@sonney2kblackburn, we have three functions computing product of vector elements18:51
@sonney2khave a look at SGVector.h18:51
blackburnlooking18:51
blackburnsonney2k: what does product() do?18:51
@sonney2kline 43118:52
@sonney2kx_1 * x_2 ... * x_n18:52
blackburnsonney2k: I see no reason to have static func for that18:52
blackburnas one can already call non-static method for that18:52
shogun-notifier-shogun: Soeren Sonnenburg :master * f91efd5 / src/shogun/lib/SGVector.h: https://github.com/shogun-toolbox/shogun/commit/f91efd50d732c78c65ae1f388d1886a5a1934cac18:53
shogun-notifier-shogun: drop static product function from SGVector18:53
@sonney2kblackburn, me neither18:53
shogun-buildbotbuild #886 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/886  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>18:56
@sonney2kblackburn, hmm python ./streaming_vw_createcache_modular.py crashes19:07
@sonney2k==16027== Thread 2:19:08
@sonney2k==16027== Invalid read of size 819:08
@sonney2k==16027==    at 0x86E5683: shogun::CVwParser::read_svmlight_features(shogun::CIOBuffer*, shogun::VwExample*&) (VwParser.cpp:177)19:08
@sonney2k==16027==    by 0x82B713B: shogun::CStreamingVwFile::get_vector_and_label(shogun::VwExample*&, int&, double&) (StreamingVwFile.cpp:63)19:08
@sonney2k==16027==  Address 0x0 is not stack'd, malloc'd or (recently) free'd19:09
blackburnwhoop19:12
@sonney2kheiko, ^ did you see that?19:17
heikosonney2k: looks like the other streaming errors19:17
@sonney2kdo you know when did this last work?19:19
@sonney2kheiko, ohh and any chance you can fix all the statistics_* python modular examples?19:20
heikosonney2k: no idea19:20
heikosonney2k:  yes I can do that, I am currently busy though, when do you want to release? I probably can do this on the weekend19:20
@sonney2kheiko, I wanted to release today19:21
heikoI see19:21
heikook, will go for it now then19:21
@sonney2kblackburn, do you remember who wrote ./classifier_multiclasslogisticregression_modular.py ?19:22
shogun-notifier-shogun: Soeren Sonnenburg :master * 60de6d3 / src/NEWS: https://github.com/shogun-toolbox/shogun/commit/60de6d383f7c633b4a699db4d19ddc762bbb1b1419:24
shogun-notifier-shogun: fix whitespaces in NEWS19:24
@sonney2kblackburn, could you please update NEWS?19:24
@sonney2kblackburn, and prepare an announcement email19:25
blackburnsonney2k: re19:28
blackburnsonney2k: multiclass LR is probably me, I'll take a look19:28
blackburnsonney2k: I'll be back in a half of hour - will update NEWS and write a mail19:29
shogun-buildbotbuild #887 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/887  blamelist: Soeren Sonnenburg <sonne@debian.org>19:33
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 252 seconds]19:36
heikosonney2k: some examples have multiple things in the parameter list19:43
heikohow does this work19:43
heikoe.g.19:43
heikoparameter_list = [[1,7],[2,8]]19:43
heikodef classifier_custom_kernel_modular (C=1,dim=7):19:43
n4nd0hey heiko19:45
heikon4nd0:  hi19:45
n4nd0so the values in the function definition are default values I believe19:45
n4nd0and about the parameter list19:46
heikoI meant the parameter list19:46
n4nd0normally the method is called something like this19:46
n4nd0*parameter_list[0]19:46
n4nd0so I think it is just if you want to try other parameters combinations19:46
@sonney2kyes exactly19:46
n4nd0I am not sure if the tests are running automatically all the elements in parameter_list though19:47
@sonney2kthey do19:47
heikosonney2k: ok cool19:47
@sonney2kit is often fine to just have 1 test19:47
heikothe examples only run the first then I assume?19:47
@sonney2kyes19:47
@sonney2ksometimes you want more, e.g. for an svm two different kernels or so19:48
heikosonney2k: is it sufficient to just init the seed?19:48
heikoif random stuff is going on?19:48
@sonney2kyes19:49
heikoMath.init_random(1) is enought, sonney2k?19:50
@sonney2kblackburn, ./preprocessor_dimensionreductionpreprocessor_modular.py:21: RuntimeWarning: [WARN] The neighborhood graph is not connected.19:51
@sonney2kheiko, yes19:51
@sonney2kthough I would use 4219:51
@sonney2khmmhh serialization to json is b0rken19:51
n4nd0sonney2k: increasing the number of neighbors in that example should make that warning disappear19:53
n4nd0the name should be k19:53
@sonney2kn4nd0, increasing it to >20 helped19:55
n4nd0sonney2k: good :)19:56
blackburnre19:56
shogun-buildbotbuild #888 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/888  blamelist: Soeren Sonnenburg <sonne@debian.org>19:56
n4nd020 is fine here too19:56
@sonney2kso what issues do we have currently: json serialization b0rken, vw & streaming features no longer working,20:00
@sonney2kand these examples don't work as integration tests20:00
@sonney2kmodelselection_grid_search_kernel.py                         ERROR ('module' object has no attribute 'parameter_list')20:00
@sonney2kserialization_string_kernels_modular.py                      ERROR ('module' object has no attribute 'parameter_list')20:00
@sonney2kstatistics_hsic.py                                           ERROR ('module' object has no attribute 'parameter_list')20:00
@sonney2kstatistics_kmm.py                                            ERROR ('module' object has no attribute 'parameter_list')20:00
@sonney2kstatistics_linear_time_mmd.py                                ERROR ('module' object has no attribute 'parameter_list')20:00
@sonney2kstatistics_mmd_kernel_selection.py                           ERROR ('module' object has no attribute 'parameter_list')20:00
@sonney2kstatistics_quadratic_time_mmd.py                             ERROR ('module' object has no attribute 'parameter_list')20:00
heikosonney2k:  can i have multiple function calls in python examples?20:01
heikoif __name__=='__main__':20:01
heikoprint('MMDKernelSelection')20:01
heikokernel_choice_linear_time_mmd_single(*parameter_list[0])20:01
heikokernel_choice_linear_time_mmd_comb()20:01
heiko#show()20:01
heikolike this?20:02
@sonney2kheiko, no20:03
@sonney2kthen make 2 examples out of this20:04
heikosonney2k: thx20:06
heikosonney2k: so if I return an instance of CGaussianKernel20:11
heikois this serialised and the width is asserted to be correct?20:11
@sonney2kheiko, not just the width20:13
heikoand other things20:13
@sonney2keverything that belongs to the gaussian kernel20:13
heikocool20:13
@sonney2ksame features, subsets everything20:13
heikosonney2k: that is really awesome!20:13
blackburnI am surprised heiko is being surprised by shogun :)20:19
heikoblackburn: I tend to not look at things sometimes :)20:20
heikoI knew this existed20:20
heikobut I did not know how it worked and how simple it is20:20
blackburnsonney2k: filling news20:24
heikosonney2k:  is polymorphism correlty handled in python20:24
heikoso if I have classes A->B20:24
heikoand in python, I have an instance of B under the interface of A20:24
heikoand then return this in the example20:24
heikois B serialised or only the A part of it?20:24
shogun-buildbotbuild #889 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/889  blamelist: Soeren Sonnenburg <sonne@debian.org>20:29
@sonney2kheiko, it should work since the serialization underneath is done by the c++ classes20:29
blackburnsonney2k: we should mention perl, right?20:29
@sonney2kblackburn, yes as alpha20:29
shogun-notifier-shogun: Sergey Lisitsyn :master * f7f3431 / src/NEWS: https://github.com/shogun-toolbox/shogun/commit/f7f34318e539b1433ef721636c860a5e365bfdbe20:31
shogun-notifier-shogun: Some updates for NEWS20:31
shogun-notifier-shogun: Heiko Strathmann :master * c6db0b3 / examples/descriptions/modular/ (4 files): https://github.com/shogun-toolbox/shogun/commit/c6db0b3d55ea7db3d6eba37d9c82e393e195c09920:37
shogun-notifier-shogun: example descriptoins for kernel test statistics20:37
shogun-notifier-shogun: Heiko Strathmann :master * 7e2e174 / src/shogun/evaluation/CrossValidation.h: https://github.com/shogun-toolbox/shogun/commit/7e2e1740a60dacba092b0fcb731fe2588f242dde20:37
shogun-notifier-shogun: made CrossValidatioResult serializable20:37
shogun-notifier-shogun: Heiko Strathmann :master * 45e0d17 / examples/undocumented/python_modular/ (9 files): https://github.com/shogun-toolbox/shogun/commit/45e0d171e6f45721e09ff03944bfe5f007262f2320:37
shogun-notifier-shogun: made some python examples work with integration tests20:37
shogun-notifier-shogun: (split the mmd kernel selection example)20:37
shogun-notifier-shogun: Heiko Strathmann :master * 9fe4085 / / (14 files): https://github.com/shogun-toolbox/shogun/commit/9fe4085ef00a8086cebd859c42103f0c1fcfcadb20:37
shogun-notifier-shogun: Merge pull request #911 from karlnapf/master20:37
shogun-notifier-shogun:20:37
shogun-notifier-shogun: added integration tests20:37
heikosonney2k: blackburn ^20:37
blackburnheiko: what should I do? ;)20:38
heikodont know, this should resolve all the integration tests, just wanted to let you know20:38
blackburnheiko: ahh! I see20:38
heikosonney2k: so I haven't run the generator yet20:38
blackburnnice20:38
heikohow do I do it?20:38
heikosonney2k: blackburn, I will go home now and check in online a bit later to check whether things are still needed, I seriously need diner :) see you later!20:42
blackburnheiko: yeah see you20:42
-!- heiko [~heiko@nat-176-119.internal.eduroam.ucl.ac.uk] has left #shogun []20:42
shogun-buildbotbuild #619 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/619  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>20:53
shogun-buildbotbuild #620 of cyg1 - libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/62021:08
shogun-buildbotbuild #890 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/890  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>21:17
shogun-buildbotbuild #891 of deb3 - modular_interfaces is complete: Failure [failed compile r_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/891  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:19
blackburnsonney2k: writing an announcement but not sure about our status21:34
@sonney2kblackburn, still debugging...21:35
@sonney2kblackburn, whats up with ./classifier_multiclasslogisticregression_modular.py:25: RuntimeWarning: [WARN] multiclass/MulticlassLogisticRegression.cpp:MulticlassLogisticRegression: Unstable method!  Please report if it seems to work or not to the Shogun mailing list.  Thanking you in anticipation.21:36
blackburnsonney2k: I didn't have time to debug it probably21:37
blackburnsonney2k: not that I mean it doesn't work21:37
blackburnsonney2k: but I was suspicious iirc, no idea what to do with it21:37
@sonney2kblackburn, does features_dense_protocols_modular.py work for you?21:43
blackburnsonney2k: TypeError: cannot make memory view because object does not have the buffer interface21:44
@sonney2kblackburn, yeah here too :-( Could you ask gsomix what's wrong?21:49
blackburnsonney2k: not crazy close to him as usually but asked via IM :)21:50
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun21:56
blackburnsonney2k: he doesn't know but wants to take a look tomorrow22:03
blackburnsonney2k: I am not sure we should release today22:07
shogun-notifier-shogun: Soeren Sonnenburg :master * 8a5b2b0 / / (10 files): https://github.com/shogun-toolbox/shogun/commit/8a5b2b08dc03e9a498d53c363077d1bb7b4561c522:11
shogun-notifier-shogun: further example fixes22:11
shogun-notifier-shogun:22:11
shogun-notifier-shogun: - completely disable vw examples (broken)22:11
shogun-notifier-shogun: - further muten examples22:11
shogun-notifier-shogun: Soeren Sonnenburg :master * 0e23ae5 / examples/undocumented/python_modular/ (7 files): https://github.com/shogun-toolbox/shogun/commit/0e23ae57de4d7c8e106d932ea0e2b288af91159822:11
shogun-notifier-shogun: muten examples22:11
shogun-notifier-shogun: Soeren Sonnenburg :master * ff46609 / data: https://github.com/shogun-toolbox/shogun/commit/ff466092a9398bc9945f3b199ab88752c1c14ca222:11
shogun-notifier-shogun: require new data version22:11
shogun-notifier-shogun: Soeren Sonnenburg :master * 76d24b2 / src/shogun/evaluation/CrossValidation.h: https://github.com/shogun-toolbox/shogun/commit/76d24b25e83d0cb3bf68cfd2b4b62076d328e50522:11
shogun-notifier-shogun: add doc to obtain_from_generic in crossvalidation22:11
-!- heiko [~heiko@5ac1f59a.bb.sky.com] has joined #shogun22:14
heikosonney2k, blackburn how is it going?22:15
blackburnheiko: bugzz22:15
@sonney2kheiko, I've fixed bugs/warnings and quietened your examples22:17
heikosonney2k: I see, what else is missing?22:18
@sonney2kand ran the generator.py22:18
@sonney2kmost of the stuff seems to work22:18
@sonney2kwhat does not still is the vw stuff22:18
@sonney2k(I disabled that for now)22:18
heikosonney2k: yeah, we kind of waited too long to fix that....22:19
heikosonney2k:  any ideas how to proceed with that?22:19
@sonney2kheiko, argh clash22:19
@sonney2khttp://shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/891/steps/compile%20r_modular/logs/stdio22:19
@sonney2kcould you please rename the variable result to sth else22:20
heikosure22:20
@sonney2k(swig-R uses that already when returning stuff)22:20
@sonney2kheiko, I am tempted to release nevertheless22:20
heikowithout the vw stuff?22:21
@sonney2kbut we have other issues (potentially)22:21
@sonney2kjson (de)serialization22:21
heikoI dont know, maybe we should invest some time into the streaming stuff since it becomes harder and harder to fix22:22
heikobut not sure22:22
@sonney2kthen blackburn 'unstable' marked method22:22
blackburnsonney2k: what?22:22
@sonney2kand some examples fail when running tester.py22:23
@sonney2kand then the buffer protocol issue22:24
@sonney2kno idea if this is important enough for a blocker22:24
heikosonney2k: well are we in a hurry with the release?22:25
blackburnyes22:25
heikobecause?22:25
blackburngsoc22:25
@sonney2kin any case if we don't work on this / fix this stuff in the next few days then  I don't think these would be an excuse to not rlease22:25
@sonney2kheiko, well no feature commits allowed until we release22:25
heikowhat about taking one week to fix things?22:25
heikosonney2k: yes, agreed, only bugfixed22:25
blackburnI don't mind22:25
@sonney2kI don't have time to fix things though22:26
heikosonney2k: blackburn and documentation maybe22:26
@sonney2kif you both can do it?22:26
blackburnI probably can22:26
heikosonney2k: I usually have a few hours per day for shogun, so we could try22:26
heikofixed a couple of non-trivial bugs last week22:26
heikoso what about next sunday then?22:26
heikowe definitely release then and postpone all new features to then22:27
@sonney2kblackburn, when is gsoc application deadline?22:27
@sonney2kblackburn, did you write anything yet?22:27
blackburnsonney2k: 29, no22:27
blackburnyes I know22:27
blackburn:)22:27
@sonney2kI am just saying.... I won't be of much help this year22:28
@sonney2kand it will take much more time than fixing these last bugs22:28
blackburnsonney2k: yes I will start veryson22:29
blackburnsoon22:29
n4nd0blackburn: I can help with gsoc application22:29
blackburnn4nd0: I'll share google doc around later22:29
n4nd0ok22:30
shogun-notifier-shogun: Heiko Strathmann :master * ab46e89 / src/shogun/evaluation/CrossValidation.h: https://github.com/shogun-toolbox/shogun/commit/ab46e89d4f2c55fd32a57ed96a46cbb23680653d22:30
shogun-notifier-shogun: renamed result variable to avoid swig-R crash22:30
shogun-notifier-shogun: Heiko Strathmann :master * f049173 / src/shogun/evaluation/CrossValidation.h: https://github.com/shogun-toolbox/shogun/commit/f049173f7cb84fe016a63fb364206b1fab98c8c922:30
shogun-notifier-shogun: Merge pull request #912 from karlnapf/master22:30
shogun-notifier-shogun:22:30
shogun-notifier-shogun: fix swig-R clash22:30
heikoblackburn: let me know if I should write things apart from my project idea22:31
@sonney2kn4nd0, we also need the cool new website...22:31
blackburnsonney2k: do we have to have all ideas properly described before deadline?22:31
heikowe are getting closer to pull request #1000 :)22:31
@sonney2kblackburn, of course22:32
n4nd0sonney2k: I will have time for that probably during the next weekend22:32
blackburnsonney2k: we should ask possible mentors to describe ideas more then22:32
@sonney2kblackburn, we should have same quality like we had last year22:32
blackburnsonney2k: we will22:32
n4nd0sonney2k: but it is lot of CSS in there.. not sure how long will it take22:33
@sonney2kn4nd0, I spend all my free time on release preparations and won't have any/much time next week22:33
@sonney2kheiko, blackburn - let us please release next thurday22:34
n4nd0sonney2k: IMHO the webpage is fine as it is, the dropdown menu would be nicer yeah22:34
blackburnsonney2k: ok!22:34
blackburnsonney2k: will you be still an admin or backup admin?22:35
heikothursday then, ok22:36
@sonney2kblackburn, if our chances are better if I am still an admin then I am admin otherwise backup22:36
@sonney2kheiko, and no feature commits!22:36
@sonney2kjust bugfixes22:37
@sonney2kheiko, please run22:37
@sonney2kgit pull22:37
heikosonney2k: sure22:37
@sonney2kgit submodule update22:37
blackburnsonney2k: well there are pros and cons - pro is that I am former gsocer 'leaded' the project22:37
@sonney2kand then go to22:37
@sonney2ktests/integration/python_modular22:37
@sonney2kand run22:37
@sonney2kpython tester.py22:37
blackburncons is that I am not known as you22:37
heikosonney2k: I will work on those this week22:38
@sonney2kheiko, it would be a dream if we got all those to working22:39
@sonney2kthen we could integrate that in our tests on the buildbot!22:39
heikosonney2k: that might be a bit of a dream :) but I see what I can do22:39
heikosonney2k: that would be a great idea22:39
@sonney2kheiko, well I wrote that stuff 2 or 3 years back22:39
@sonney2kand wanted them to be all stable :)22:40
heikosonney2k: I am always concerned about making things automatically check themselves22:40
heikoso it is not allowed to change the files I guess :)22:40
@sonney2kexactly22:40
heikoThings might change internally without the result being changed22:40
heikoimagine you are sampling something random (with fixed seed)22:41
heikoand then change the order in the algorithm22:41
heikosame result, fails test22:41
@sonney2kheiko, sure22:41
-!- n4nd0 [~fjig@orange-03.csc.kth.se] has quit [Quit: leaving]22:42
@sonney2kthen just run generator.py with the example as filename22:42
@sonney2kand commit22:42
heikosonney2k:  most tests probably fail due to crashes or other things22:42
@sonney2kit is just important that this is a conscious decision22:42
heikosonney2k: indeed22:42
@sonney2kthe easy bugs fail like this yes22:42
heikois there any way of checking how the file was created? what if the example changed or something?22:43
@sonney2kheiko, well contract is that you update the file when you change the example22:44
@sonney2k(to return e.g. more stuff)22:44
heikoyes22:45
heikosonney2k:  I feel like we should have a "hacking shogun" readme which explains those things22:45
@sonney2klook at this: zero warnings in octave_modular, ruby_modular,python_modular :)22:45
heikoanyway, I know it now22:45
heikosonney2k:  nice :)22:45
@sonney2kheiko, well if it is on the buildslave then one would immediately know22:46
heikosonney2k: yes, thats a good goal, would be very good before gsoc22:46
heikosonney2k btw I always forget the link to the live tutorial pdf22:46
blackburnsonney2k: which email of you I should use to share the doc?22:47
shogun-buildbotbuild #892 of deb3 - modular_interfaces is complete: Failure [failed compile r_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/892  blamelist: Soeren Sonnenburg <sonne@debian.org>22:48
shogun-notifier-shogun: Soeren Sonnenburg :master * dd32469 / src/shogun/evaluation/CrossValidation.h: https://github.com/shogun-toolbox/shogun/commit/dd324691ef7b36451b82357ee8295e5f4079908622:48
shogun-notifier-shogun: fix doxygen doc22:48
@sonney2kheiko, isn't that on the shogun website?22:48
blackburnsonney2k: heiko: wiking: I shared a proposal doc with you - I will put more content here later22:49
blackburnsonney2k: please ask quoc le and olivier chapelle to provide descriptions of projects22:49
heikosonney2k: cannot find22:51
heikoblackburn: thanks!22:51
heikoblackburn: you should consider a doctor with your killing fantasies :)22:53
blackburnheiko: that's ok!22:53
heikoblackburn: sonney2k, could we have access to last's year application?22:53
@sonney2kheiko, no idea then either22:53
blackburnheiko: I'll send it to you22:54
blackburnand n4nd022:54
heikosonney2k, blackburn I have to go now, have a good evening!22:55
blackburnheiko: night :D22:55
blackburnI should have been sleeping already argh22:55
-!- heiko [~heiko@5ac1f59a.bb.sky.com] has left #shogun []22:57
shogun-buildbotbuild #893 of deb3 - modular_interfaces is complete: Failure [failed compile r_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/893  blamelist: Soeren Sonnenburg <sonne@debian.org>23:05
shogun-buildbotbuild #894 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/89423:42
--- Log closed Sat Mar 09 00:00:05 2013