Open in new window / Try shogun cloud
--- Log opened Thu Aug 18 00:00:33 2011
-!- serialhex [] has quit [Remote host closed the connection]00:51
-!- f-x [~user@] has quit [Remote host closed the connection]01:08
-!- shogun-buildbot [] has quit [Quit: buildmaster reconfigured: bot disconnecting]03:01
--- Log opened Thu Aug 18 03:49:06 2011
-!- shogun-toolbox [] has joined #shogun03:49
-!- Irssi: #shogun: Total of 2 nicks [0 ops, 0 halfops, 0 voices, 2 normal]03:49
! [freenode-info] channel flooding and no channel staff around to help? Please check with freenode support:
-!- Irssi: Join to #shogun was synced in 6 secs03:49
-!- shogun-buildbot [] has joined #shogun03:49
-!- sonney2k [] has joined #shogun03:49
-!- You're now known as Guest3938003:49
-!- in3xes [~in3xes@] has joined #shogun05:11
-!- in3xes [~in3xes@] has quit [Remote host closed the connection]05:19
-!- mode/#shogun [+o sonney2k] by ChanServ06:22
-!- You're now known as shogun-irclog06:25
-!- blackburn [~blackburn@] has joined #shogun09:23
-!- heiko [~heiko@] has joined #shogun10:04
@sonney2kgood morning heiko10:04
-!- heiko [~heiko@] has quit [Ping timeout: 258 seconds]10:08
blackburnsonney2k: he we've got 364 warnings left10:14
@sonney2kblackburn, and mostly simple ones10:15
@sonney2kit is about time that I enable make check-examples :)10:15
blackburnserialization fixed?10:15
@sonney2kblackburn, no heiko fled when I said good morning :D10:16
blackburnI'm really wondering why there are 5 online here10:16
blackburnless than ever10:16
@sonney2kconsidering that there are 3 bots in here - yes not a lot10:17
@sonney2kI guess I shouldn't write positive emails ;-)10:17
@sonney2kblackburn, once we have make check-examples I will ask everyone again to help fix /port all the examples / bugs10:18
blackburntoday warnings should lower to <20010:19
@sonney2kand once serialization migration works I will enable make tests10:19
blackburnif it is <200 I will be happy hehe10:19
@sonney2kI will be happy when all examples run10:19
blackburnI'll try to finish kpca today10:21
blackburnI hope I understood things right here10:21
blackburna little tricky thing10:22
@sonney2kif we progress as much as we do currently then it is 2-3 weeks until a release...10:22
@sonney2knot too bad I would say10:22
blackburnI should finish all the serious work before 1st of September10:23
blackburnor 5th of September as hard 'deadline'10:23
blackburnbecause it is first day of work :)10:23
blackburnas I will be studying and work 'in parallel'10:24
@sonney2kblackburn, me too actually 31.8 should be our hard deadline10:33
@sonney2kit is do-able10:33
@sonney2kblackburn, I would say most important is to check that we did not break anything that worked before10:34
@sonney2kblackburn, by running make tests when serialization migration works10:38
-!- in3xes [~in3xes@] has joined #shogun10:42
CIA-87shogun: Sergey Lisitsyn master * r560d6cd / (8 files): Various UI warnings fixes -
-!- in3xes_ [~in3xes@] has joined #shogun10:56
-!- in3xes [~in3xes@] has quit [Ping timeout: 240 seconds]10:59
-!- in3xes_ [~in3xes@] has quit [Quit: Leaving]11:01
-!- in3xes [~in3xes@] has joined #shogun11:02
-!- in3xes_ [~in3xes@] has joined #shogun11:07
-!- in3xes [~in3xes@] has quit [Ping timeout: 258 seconds]11:11
-!- heiko [~heiko@] has joined #shogun11:18
heikosonney2k, around?11:26
-!- in3xes_ [~in3xes@] has quit [Ping timeout: 240 seconds]11:29
-!- bettyboo [] has joined #shogun11:29
-!- mode/#shogun [+o bettyboo] by ChanServ11:29
-!- in3xes_ [~in3xes@] has joined #shogun11:29
blackburnsonney2k: 261!11:30
-!- in3xes__ [~in3xes@] has joined #shogun11:36
-!- in3xes_ [~in3xes@] has quit [Ping timeout: 240 seconds]11:39
-!- f-x [~user@] has joined #shogun11:44
-!- in3xes_ [~in3xes@] has joined #shogun12:05
-!- in3xes__ [~in3xes@] has quit [Ping timeout: 240 seconds]12:09
CIA-87shogun: Soeren Sonnenburg master * r9f7740a / (8 files in 4 dirs):12:59
CIA-87shogun: Merge pull request #302 from karlnapf/master12:59
CIA-87shogun: working SGVector serialization (+9 more commits...) -
blackburnsonney2k: what is elastic net?13:01
@sonney2kheiko, hi13:02
@sonney2kagain :)13:02
heikosonney2k, hi13:03
@sonney2kheiko, I am just compiling ...13:03
heikosonney2k, I hope it all works, my tests here went fine - same for the svmlight serial example13:03
@sonney2kheiko, did you manage to fix the destroy() issue?13:04
heikosonney2k, yes I did13:05
@sonney2kI looked at your patch - do you create empty SGVectors - no right?13:05
heikoin new_cont?13:05
heikowhat I basically did is:13:06
heikothere is already code which does the MALLOC/FREE stuff for arrays/matrices13:06
heikobut the pointer was wrong when using SGMatrix13:06
@sonney2kyou mean double* ptrs13:06
heikoit pointed to the method13:06
heikoI now simply changed the pointer the the vector/matrix address13:07
@sonney2kyou mean it pointed to the object right?13:07
@sonney2kheiko, wouldn't it be better to really use the contructor in this case of SGVector etc?13:07
heikothis would be possible13:07
heikobut then I would have to write this for every type13:07
heikothis way, I could simply reuse the old code13:08
heikowhich already works, if the pointer is correct13:08
@sonney2kheiko, so what are the assumption when doing this...13:08
* blackburn hates libshogunui13:08
heikosonney2k, ?13:08
@sonney2k1) the SGVector is correctly initialized before it is being messed with13:09
heikoyes, but NULL also works13:09
@sonney2k2) there is nothing going on after the vector has been set (whatever processing etc)13:09
heikooh you mean, the SGVector has to be there13:10
heikothis is true13:10
heikohow do you mean 2)13:10
@sonney2kblackburn, fix siwg's R wrapper and implement matlab-swig support then it will go13:10
@sonney2kheiko, think of something more complicated than SGCachedVector - where the vector is put somewhere / or processed13:11
blackburnsonney2k: sure13:11
@sonney2kblackburn, it is ok if you finish by the end of today :D13:11
blackburnsonney2k: yeah I plan to finish it in a min13:11
@sonney2kblackburn, as always - you are the man :)13:12
heikosonney2k, currently, the SGVector is treated like a class variable.13:13
heikoonly the content is changed13:13
@sonney2kheiko, yes13:14
@sonney2kI have mixed feelings about this - this will cause trouble at some point13:14
heikoWell, I could change this13:15
heikobut it would not be possible to add class member SGVectors then13:15
heikobecause they are deleted13:15
heikofrom stack memory13:16
heikoCurrently, all SGVector replacements are on stack13:16
-!- in3xes__ [~in3xes@] has joined #shogun13:17
@sonney2kheiko, if you do class->vector = SGVector x(some_memory, len)?13:18
@sonney2kahh not working because not public13:18
@sonney2kok then, so there is no way other than messing around directly in the SGVector13:19
@sonney2kheiko, so lets keep it like this13:20
@sonney2kheiko, I just tried serialization_complex_example.py13:20
@sonney2kit fails in hdf5 and xml13:20
heikosonney2k, yes it complains about missing imprts here13:20
@sonney2kit just works13:20
heikosonney2k, ok great :)13:21
-!- in3xes_ [~in3xes@] has quit [Ping timeout: 240 seconds]13:21
@sonney2kbut I think the vector patch needs to be done for xml and hdf513:21
@sonney2kjson, ascii work13:21
heikoah ok13:21
heikowell, ok then13:21
heikonot too much stuff13:21
@sonney2kheiko, another question - SGNDArray, SGStringList and SGSparseMatrix also need to be treated like this right (but not yet done - or?)13:22
heikoyes, not yet done13:22
heikoshould I also add this?13:22
@sonney2kheiko, the types yes, the support can probably wait - or do you know any class that uses these already?13:23
heikosonney2k, I dont know any13:23
heikosonney2k, adding the types is basically the only step for adding support13:23
heikojust a few switch cases once one knows what to do13:24
@sonney2kheiko, I thought you need to change every serializable file format too?13:24
@sonney2klets prioritize things13:24
heikoyes true, but only minor changes there13:24
@sonney2kI have to enable make check-exampels tonight13:24
@sonney2kfor that I need hdf5/xml support too13:24
@sonney2kthen all examples run through13:25
heikook then13:25
@sonney2kso that first please13:25
heikoI will do the xml/hdf5 stuff for vector and matrix13:25
heikoare there any serialization examples with the other types?13:25
@sonney2kthen you said there are problems with the migration?13:25
heikowell yes, its tricky13:25
@sonney2kheiko, just save a stringfeature object13:25
@sonney2kor sparse feature matrix13:25
heikoehm that means, which types have to work? vector, matrix, sparsematrix?13:26
@sonney2kheiko, but not for make check-examples tonight13:27
@sonney2kfor that we need just vector/matrix13:27
CIA-87shogun: Sergey Lisitsyn master * rbd45c6c / (6 files in 2 dirs): More UI warnings removal -
blackburnbet if it is <20013:27
heikook then13:28
heikoI will finish this now, have lunch and then continue on migration13:28
@sonney2kheiko, ok thank you very much for the great job you are doing!13:29
@sonney2kblackburn, please cheer heiko up too :)13:29
shogun-buildbotbuild #62 of cmdline_static is complete: Failure [failed compile]  Build details are at  blamelist: heiko.strathmann@gmail.com13:29
blackburngenosse heiko you are doing great job for the soviet people!13:29
heikosonney2k, blackburn thanks for the flowers :)13:30
heikoblackburn, lol ;)13:30
blackburnheiko: did I said I want to marry you?13:30
@sonney2kheiko, ^ look at the build failure13:30
@sonney2kblackburn, it is legal even in germany13:30
heikoblackburn, true, so lets go for it :D13:30
@bettybooblackburn, :^)13:30
heikosonney2k, yes the blamelist13:30
blackburnyeah I'll got german citizenship13:30
shogun-buildbotbuild #57 of r_static is complete: Failure [failed compile]  Build details are at  blamelist: heiko.strathmann@gmail.com13:31
@sonney2kblackburn, not so easy you first have to code with heiko 10 years in a row on shogun *while being married*13:31
heikohehe, unpaid :)13:33
heikosonney2k, the build error is strange13:33
@sonney2kheiko, I think some include is missing13:33
heikodo I have to introduce these enums at another point?13:33
heikoi mean if CT_VECTOR is known, how can it be that CT_SGVECTOR is unknown?13:33
@sonney2kheiko, could be that this is from some early git commit that you did - so it is one that doesn't compile that you checked13:35
heikosonney2k, ok13:35
@sonney2kheiko, btw in shogun/lib/DataType.h - in the enum13:35
heikosonney2k, thats where i introduced the new type13:36
@sonney2kyou cannot change the meaning of CT_SGMATRIX13:36
@sonney2kyou have to put these CT_SG* types at the end13:36
heikosonney2k, ?13:36
@sonney2kerr  you cannot change the meaning of CT_MATRIX13:36
heikois the integer saved somewhere?13:36
@sonney2kI mean  CT_MATRIX was 2 before or so13:36
@sonney2kheiko, starting with 1 I suppose13:37
blackburnwhy is it compiles here?13:37
@sonney2kI wouldn't mind if you do CT_MATRIX=2 etc13:37
@sonney2kblackburn, revision f059083741272ad277cdd527935fca4b87c919e0 ?13:37
heikosonney2k, i just moved them to the end13:37
@sonney2kheiko, no I meant you make all these explicit13:38
heikook, will do13:38
@sonney2kheiko, this way numbers are fixed & clear13:38
heikoguys, I will go for lunch now, mensa is closing soon13:38
blackburnehh how to get revision num13:38
@sonney2kjust make sure you don't change the default13:38
@sonney2kheiko, yeah cu13:38
heikosonney2k, will be back in the afternoon for all the stuff we just talked about13:38
@sonney2kblackburn, git log | head -n 113:39
blackburncommit bd45c6ca83b1ab9b8f2f1ed2cef13beaceebb19413:39
@sonney2kheiko, yes thanks13:40
@sonney2kblackburn, so you see - the new builds are already schedules (cf )13:41
blackburnI see13:41
@sonney2kwill build your bd45c6ca83b1ab9b8f2f1ed2cef13beaceebb19413:41
@sonney2kwe really need some more stable buildbot :(13:42
blackburnI wait for <200 warnings here13:42
-!- heiko [~heiko@] has quit [Ping timeout: 258 seconds]14:16
blackburnlibshogun is ahead of 10 builds oh14:20
blackburnbuildbot is slow today14:20
-!- serialhex [] has joined #shogun14:31
-!- in3xes_ [~in3xes@] has joined #shogun14:33
-!- in3xes__ [~in3xes@] has quit [Ping timeout: 240 seconds]14:37
-!- in3xes_ is now known as in3xes14:43
blackburnsonney2k: kernelpca -> dimensionreductionpreprocessor?14:44
shogun-buildbotbuild #63 of cmdline_static is complete: Success [build successful]  Build details are at
-!- in3xes_ [~in3xes@] has joined #shogun14:49
-!- in3xes_ [~in3xes@] has quit [Remote host closed the connection]14:49
shogun-buildbotbuild #58 of r_static is complete: Success [build successful]  Build details are at
-!- in3xes [~in3xes@] has quit [Ping timeout: 276 seconds]14:53
shogun-buildbotbuild #60 of octave_modular is complete: Failure [failed compile]  Build details are at  blamelist: heiko.strathmann@gmail.com15:12
CIA-87shogun: Sergey Lisitsyn master * rbd8f328 / (6 files): Added SGREFing while applying to feature matrices in dimensionality reduction preprocessors, fixed wrong already owned matrices return in MDS and Isomap -
-!- blackburn [~blackburn@] has quit [Quit: Leaving.]15:37
-!- in3xes [~in3xes@] has joined #shogun15:59
shogun-buildbotbuild #61 of octave_modular is complete: Success [build successful]  Build details are at
-!- heiko [~heiko@] has joined #shogun16:19
heikosonney2k, does shogun compile now?16:23
@sonney2kshogun-buildbot, status all16:32
shogun-buildbotcmdline_static: idle, last build 1h26m50s ago: build successful16:32
shogun-buildbotcsharp_modular: idle, last build 17m23s ago: build successful16:32
shogun-buildbotjava_modular: idle, last build 7m54s ago: build successful16:32
shogun-buildbotlibshogun: building(no current step) [ETA 1m57s]16:32
shogun-buildbotlua_modular: idle, last build 12m40s ago: build successful16:32
shogun-buildbotnightly_all: idle, last build 19h02m50s ago: failed compile16:32
shogun-buildbotnightly_none: idle, last build 19h07m41s ago: build successful16:32
shogun-buildbotoctave_modular: idle, last build 22m12s ago: build successful16:32
shogun-buildbotoctave_static: idle, last build 1h36m21s ago: build successful16:32
-!- shogun-buildbot was kicked from #shogun by bettyboo [flood]16:32
@sonney2kheiko ^16:32
-!- bettyboo was kicked from #shogun by sonney2k [bettyboo]16:32
-!- bettyboo [] has joined #shogun16:32
-!- mode/#shogun [+o bettyboo] by ChanServ16:32
heikosonney2k, also just checked it :)16:32
-!- shogun-buildbot [] has joined #shogun16:33
@sonney2kheiko, did you manage to support hdf5 / xml?16:33
-!- bettyboo [] has quit [Remote host closed the connection]16:34
heikocurrently doing ...16:34
-!- in3xes_ [~in3xes@] has joined #shogun16:37
-!- in3xes [~in3xes@] has quit [Ping timeout: 240 seconds]16:40
-!- in3xes__ [~in3xes@] has joined #shogun16:46
-!- in3xes_ [~in3xes@] has quit [Ping timeout: 258 seconds]16:49
-!- in3xes_ [~in3xes@] has joined #shogun17:04
-!- in3xes__ [~in3xes@] has quit [Ping timeout: 258 seconds]17:08
-!- in3xes_ [~in3xes@] has quit [Ping timeout: 240 seconds]17:22
-!- heiko [~heiko@] has quit [Quit: Leaving.]17:36
-!- heiko [~heiko@] has joined #shogun17:36
heikosonney2k, which debian package to install for hdf5 support?17:49
@sonney2kheiko, libhdf5-serial-dev17:56
heikosonney2k, thanks17:57
heikosonney2k, and for xml? my std choices do not work, configure script sais ok but compiler sais cannot find libxml/parser.h18:02
heikoSerializableHdf5 File leaks memory18:21
heikowhich is not tracked by trace-mallocs :(18:22
-!- heiko [~heiko@] has quit [Ping timeout: 258 seconds]19:15
CIA-87shogun: Soeren Sonnenburg master * r12b8f95 / (10 files in 3 dirs):19:55
CIA-87shogun: Merge pull request #303 from karlnapf/master19:55
CIA-87shogun: serialization of SGMatrix/SGVector (+6 more commits...) -
serialhexsonney2k: are you on??21:08
-!- in3xes [~in3xes@] has joined #shogun21:09
@sonney2kserialhex, yes21:09
serialhexok, labels are going to be a vector right?21:10
serialhexi'm like 90% certain but i want to be 100%21:10
serialhexok cool!21:11
serialhexwe're ar v 0.99.9?? :P21:21
-!- blackburn [~blackburn@] has joined #shogun21:31
serialhexhey blackburn, whats up??21:32
blackburnserialhex: hi, i'm fine and you?21:35
serialhexpretty good21:35
@sonney2kblackburn, 177 warnings?!21:35
serialhexworking on these examples21:35
blackburnsonney2k: seems so21:36
blackburnI haven't checked21:36
blackburnyeah as you can see21:36
blackburnhmm lets make it 10021:36
@sonney2kblackburn, heiko committed the fix21:40
blackburnfix about?21:40
@sonney2kso I will try to get some make check-examples into the buildbot21:40
blackburnah, serialization21:40
blackburnsonney2k: what about moving kpca under dimreductionpreprocessor?21:40
blackburnnow it is pretty crazy21:41
blackburnwe set threshold in constructor (sic!) and then use it as int21:41
@sonney2kblackburn, makes sense but PCA then too21:41
blackburnyeah sure21:41
blackburnand I would remove alesis' stuff here21:42
blackburnwith various modes..21:42
blackburnI don't sure if I should21:42
@sonney2kwhy remove?21:42
@sonney2kit makes sense to preprocess and say I want 10 components21:42
@sonney2kor all components with eigenvalues > 0.00121:42
@sonney2k(these are the 2 modes)21:43
blackburnbecause target dimensionality is useless when there are two modes21:43
blackburnnot really predictive acting21:43
blackburnI don't know21:43
@sonney2kblackburn, one has to select one of these modes of course21:44
blackburnwell I don't think preprocessor should think about this thresholding21:44
@sonney2kI don't mind that you make them extra functions21:44
@sonney2ksuch that one has to call them manually like21:45
@sonney2kor so21:46
blackburnyeah ok21:46
blackburnbut this time I have to finish kpca21:46
shogun-buildbotbuild #6 of nightly_all is complete: Success [build successful]  Build details are at
blackburnsonney2k: ?! nighly all compiled!!!21:51
@sonney2kblackburn, no buildbot crashed21:52
@sonney2kI wish I would figure out what the problem is21:52
blackburnserialhex: hey should I write your name in NEWS? or better nickname?21:56
serialhexeither or... you could say: "Justin Patera aka serialhex" or something... or just serialhex... whatever works :)21:57
blackburnserialhex: how can I mention your spiffy trancekoder? :)21:59
serialhexso are we gonna have a newsletter or something blackburn?21:59
blackburnoh I'm talking about this:21:59
serialhexhow?  erm, you mean describe??21:59
blackburnhmm strange I got link to your fork hah22:00
blackburnyeah describe whatever22:00
blackburnnot fair you are not in authors and news22:00
blackburnhmm does "example transition tool" work?22:01
serialhexyeah thats cool22:01
serialhexi'm probably going to kill that file once i'm done with moving the python files to ruby...  though it was a stroke of pseudo-genius getting that done :D22:02
serialhexso have you finished work on the java serialization blackburn??22:04
blackburnseems so22:05
CIA-87shogun: Sergey Lisitsyn master * ra3b6203 / (src/CONTRIBUTIONS src/NEWS): Updated news and contributions -
blackburnserialhex: +1 to fame :D22:11
@sonney2kand flame :D22:12
serialhexsonney2k!!!  i keep getting this error: "/usr/lib/ruby/1.8/i486-linux/ warning: rb_define_const: invalid name `__STDC_FORMAT_MACROS' for constant" when i load and i found the line/place where said error originates: shogun/src/shogun/lib/common.h line # 4322:13
serialhexidk what is to be done about it (if anything) but i'm just letting you know what i've found!22:14
@sonney2kserialhex, looks like I already tried to fix it:22:14
@sonney2k#define __STDC_FORMAT_MACROS 122:14
@sonney2k#include <inttypes.h>22:14
@sonney2k#undef __STDC_FORMAT_MACROS22:14
@sonney2kseems like it didn't help22:15
serialhexyeah, ruby dosnt like constants with names like that... all constants start with a capital letter... hmm...........22:16
blackburnhey guys I've found the strangest function of shogun22:17
blackburnhere it is22:17
blackburnint* cntsrtna(int *sx, int k, int r, int na);22:17
serialhexthough everything else seems to be running peachy-keen!22:17
serialhexdo wahh??22:17
blackburnI'm stucked what to write in its description ahaha22:17
@sonney2kblackburn, guessing from the name it sounds like count sort22:18
@sonney2kna == no idea22:18
blackburnsonney2k: I'll write that it makes CNTSRTNA22:19
@sonney2kblackburn, very useful22:19
@sonney2kserialhex, I don't actually know why we need inttypes.h22:19
serialhexhmm, ok... what's it do?22:19
@sonney2kserialhex, let me try to compile w/o22:20
serialhexoh, and blackburn, when you make yourself famous for all the spiffy docs you can say that i did the russian-english -> english edit :P22:20
serialhexok cool sonney2k, cause it'd take ~ 20 min to compile here :-/22:20
@sonney2kserialhex, now I know what it is for22:22
@sonney2k^ that is a format specifier22:22
@sonney2kfor int3222:22
serialhex...which does?22:22
serialhexahh, i see22:23
@sonney2kfor printf()22:23
serialhexoh, ok22:23
@sonney2kserialhex, still it doesn't make sense that __STDC_FORMAT_MACROS is visible from the swig interfaces22:23
@sonney2klets try a %ignore22:23
CIA-87shogun: Sergey Lisitsyn master * rb0eb823 / (16 files in 7 dirs): Warning removal -
blackburnshould be <100 now22:27
blackburntime to bomb saddam^W^W finish kpca22:29
-!- shogun-buildbot [] has quit [Quit: buildmaster reconfigured: bot disconnecting]22:31
-!- shogun-buildbot [] has joined #shogun22:31
-!- shogun-buildbot [] has quit [Client Quit]22:34
-!- shogun-buildbot [] has joined #shogun22:34
@sonney2kshogun-buildbot, status all22:38
shogun-buildbotcmdline_static: idle, last build 17m02s ago: build successful22:38
shogun-buildbotcsharp_modular: idle, last build 7m08s ago:22:38
shogun-buildbotjava_modular: idle, last build 4m07s ago:22:38
shogun-buildbotlibshogun: idle, last build 7m08s ago:22:38
shogun-buildbotlua_modular: idle, last build 7m08s ago:22:38
shogun-buildbotnightly_all: idle, last build 47m09s ago: build successful22:38
shogun-buildbotnightly_none: idle, last build 47m09s ago: build successful22:38
shogun-buildbotoctave_modular: building(compiling)22:38
shogun-buildbotoctave_static: idle, last build 11m55s ago: build successful22:38
shogun-buildbotpython_modular: idle, last build 7m08s ago:22:38
shogun-buildbotpython_static: idle, last build 12m06s ago: build successful22:38
shogun-buildbotr_modular: idle, last build 7m08s ago:22:38
shogun-buildbotr_static: idle, last build 17m01s ago: build successful22:38
shogun-buildbotruby_modular: idle, last build 7m08s ago:22:38
CIA-87shogun: Soeren Sonnenburg master * r3747c4d / src/interfaces/modular/modshogun_ignores.i : ignore __STDC_FORMAT_MACROS in modular interfaces -
CIA-87shogun: Soeren Sonnenburg master * r1807bc2 / (18 files in 8 dirs): Merge branch 'master' of -
@sonney2kblackburn, do you know if we can ourselves use --rebase when pulling?22:39
@sonney2kI mean not that we screw up history?22:39
blackburnrebase what to what?22:39
@sonney2kblackburn, when you do a git push22:40
@sonney2kand I have some commits already on my local master22:40
@sonney2kand then do git pull github master22:40
blackburnreally? and there wouldn't be merge?22:40
@sonney2kthen would a git pull github master --rebase work?22:40
blackburnahh it was a question22:40
blackburnno, I don't know22:40
@sonney2kblackburn, can you try with some small change?22:41
blackburnI try to git pull things just before commiting22:41
@sonney2k(I just pushed)22:41
@sonney2kso when you change some file and do a --rebase22:41
blackburnso commit then rebase22:41
@sonney2kthen we will see22:41
@sonney2kthen push and pray :)22:41
blackburnI like the concept of push&prahy22:42
* sonney2k praises $DEITY dancing the ~~~lumbojumboshogunolaaaa~~~~22:44
blackburnrefusing to pull with rebase: your working tree is not up-to-date22:45
CIA-87shogun: Sergey Lisitsyn master * rdb25278 / src/shogun/preprocessor/MultidimensionalScaling.cpp : Update MDS thread doc -
CIA-87shogun: Sergey Lisitsyn master * r2ac0b0f / src/interfaces/modular/modshogun_ignores.i : Merge branch 'master' of -
blackburnseems the answer is no22:46
@sonney2kserialhex, what marks a comment in ruby?22:47
blackburnehh how to call static SGMatrix' method?22:48
blackburngot stupidity assault22:49
serialhexsorry about the delay sonney2k, a # marks a comment22:50
serialhexjust like python22:50
@sonney2kserialhex, ok22:51
@sonney2kI am just creating documented examples from the undocumented ones22:51
serialhexok cool22:51
@sonney2k(we have the description in a separate dir and prepend it to each file)22:51
@sonney2kso I need the comment symbol22:51
@sonney2kit is really fun to see22:52
@sonney2kwe have '%' '#' '- ' '//'22:52
@sonney2kbut # is the winner now :)22:52
@sonney2kblackburn, would you mind to move into graphical?22:53
blackburnyeah sure22:53
@sonney2kblackburn, if you want prepend it with dimreduct_22:53
blackburnnot really necessary to mark them anyhow22:53
blackburnjust move it22:53
@sonney2kblackburn, ok I will do22:54
blackburnI thought there will be a millions of examples22:54
blackburnbut I did two22:54
@sonney2kblackburn, but you did one for each in the non-graphical dir22:54
blackburnyeah but useless22:55
@sonney2kwhy useless?22:55
blackburnhow can one use it? :)22:56
@sonney2kblackburn, copy the example in $DIR, load own data - > run22:57
@sonney2keasy :)22:57
blackburnbut no visualization - useless22:57
@sonney2kbut visualization is trivial to add right - everyone can do that22:57
CIA-87shogun: Soeren Sonnenburg master * r33a399d / (5 files in 3 dirs):22:58
CIA-87shogun: create documented examples for all supported languages and move the22:58
CIA-87shogun: dimreduct examples one dir up -
@sonney2kblackburn, the rebase worked here22:58
blackburnsonney2k: what is the best way to implement getters with matrices/vectors?22:58
@sonney2kI just did it22:58
blackburnI did something wrong22:58
@sonney2kI did22:59
@sonney2kgit commit -a22:59
@sonney2kgit pull --rebase github master22:59
blackburnI know22:59
@sonney2kand push22:59
blackburnI forgot to specify what to pull22:59
blackburnjust git pull rebase22:59
@sonney2kblackburn, I don't understand the question - inside SGVector you mean?23:00
blackburnSGMatrix<float64_t> get_transformation_matrix() const23:00
blackburn->->->->->->->->->->->->->->->->->->->->->->->->SGMatrix<float64_t> matrix = SGMatrix<float64_t>(m_transformation_matrix);23:00
blackburn->->->->->->->->->->->->->->->->->->->->->->->->matrix.do_free = false;23:00
blackburn->->->->->->->->->->->->->->->->->->->->->->->->return matrix;23:00
blackburnsonney2k: is it right way to do that?23:01
blackburnor better to make it no-do-free internally?23:02
@sonney2kyeah - not sure if const will work though23:02
@sonney2kblackburn, I would make it no-do-free internally23:02
@sonney2k(I mean you own the object anyways and will do destroy in destructor)23:02
blackburnyeah I luv it much more23:02
blackburnlike heiko23:02
blackburnhey gaytoolbox sounds pretty good23:04
blackburnshogun gaytoolbox23:04
blackburnsonney2k: should kernel be added to m_parameters thing?23:07
blackburnin kpca essentially23:08
@sonney2kblackburn, if the kernel is later needed23:08
blackburnwhen - later?23:08
@sonney2kthink of the preproc is created23:09
@sonney2kand things added23:09
blackburnhmm okay23:09
@sonney2kthen someone wants to save it to disc (can happen at any time)23:09
@sonney2kand then continue23:09
blackburnthis time I will just try to get it work23:09
@sonney2kblackburn, the marriage of heiko and you will end the cold work23:10
blackburnsonney2k: when cleanup is called?23:10
@sonney2kcold war23:10
blackburnyeah berlin wall will fall23:10
@sonney2kthis can only mean hot times and of course that shogun-toolbox is hot :D23:11
@sonney2kblackburn, when the user calls it23:11
blackburnI would say it is perestroyka now23:11
@sonney2kanytime so23:11
blackburnshogun looks like ussr in 89 hahah23:11
@sonney2kit is just the end of the world .... of shogun 0.x23:11
blackburnbtw tomorrow is a 20th anniversary of USSR government take-over23:12
blackburnif I born 6 months later I would born in russia :D23:13
serialhexlol blackburn!23:16
blackburnserialhex: you should hope obama will not do perestroyka :D23:19
serialhexperestroyka?  /searches wikipedia...23:19
blackburnor perestroika23:20
serialhexahhh... i see :P23:20
blackburnserialhex: one of + of perestroika - in sonney2k's city there was no wall since 198923:22
blackburnprobably that was awful thing :)23:22
serialhexcool... i vaguely remember the bringing down of the wall23:23
blackburnI didn't born yet in 1989 hehe23:23
blackburnwhoaa got kpca to compile23:26
blackburnbut still need to get to work haha23:26
blackburnsonney2k: hey but why it doesn't compile if test failed23:28
blackburnI can't count warnings haha23:29
-!- f-x [~user@] has left #shogun ["ERC Version 5.3 (IRC client for Emacs)"]23:31
serialhexhey, i keep getting an error when i run classifier_averaged_perceptron_modular.rb: "[WARN] Averaged Perceptron algorithm did not converge after 1000 iterations."  does this happen normally?23:32
serialhexin the py version?23:32
@sonney2kblackburn, hmmhh how should I fix the issue?23:34
serialhexok, just making sure noting was borked! :D23:34
blackburnsonney2k: which?23:34
@sonney2kblackburn, the linker requires -llapack23:34
@sonney2kfor compiling 1 of the examples23:34
blackburnwhere is it used?23:34
@sonney2kdon't know exactly - I guess dot() for computing kernel23:35
@sonney2kbut that is not the relevant point23:35
CIA-87shogun: Soeren Sonnenburg master * rf5254de / examples/undocumented/libshogun/Makefile : add -llapack in examples -
blackburncan't it be removed?23:38
@sonney2kblackburn, what can be removed?23:40
blackburnlapack usage in examples23:40
@sonney2kblackburn, the problem is the example needs to be linked to lapack23:40
blackburnyeah but can we modify it to not use lapack23:41
@sonney2kblackburn, sure we could but we don't know if atlas is installed (we are not using the configure system for examples...)23:42
-!- in3xes [~in3xes@] has quit [Ping timeout: 240 seconds]23:42
blackburnah okay23:45
-!- in3xes [~in3xes@] has joined #shogun23:47
CIA-87shogun: Soeren Sonnenburg master * r615d2a2 / examples/undocumented/libshogun/Makefile : add flag LIBS_ADD to add libs in addition to LIBS -
@sonney2kblackburn, btw the problem is only due to using an inline function23:48
@sonney2kbecause of that the clbas_ddot is in the header and when linking we need to specify the lib23:49
blackburninline function calls cblas??23:50
blackburnI wouldn't make it inline23:50
shogun-buildbotbuild #72 of libshogun is complete: Success [build successful]  Build details are at
@sonney2khurray :)23:52
@sonney2kblackburn, please create a bug report in trac saying that we need to rework Math.h23:53
@sonney2kand add that we should not have library calls in inline functions23:53
@sonney2k(otherwise we get hard to track linker deps in C++ extensions)23:53
blackburnin one ticket?23:54
@sonney2know with check-examples enabled things become much harder23:54
@sonney2kblackburn, I guess some text would not be bad :)23:55
@sonney2klike why rework and what should be done...23:55
blackburnno idea what to rework23:56
@sonney2kblackburn, split the file up23:56
@sonney2kblackburn, removing inline is not sufficient btw23:56
@sonney2kthese calls have to be in the .cpp file23:56
shogun-buildbotbuild #63 of r_static is complete: Failure [failed test]  Build details are at  blamelist: sonne@debian.org23:57
blackburnhope you satisfied :D23:57
@sonney2kyeah everything fails finally23:58
blackburnsonney2k: perestroika23:58
@sonney2kblackburn, thanks23:59
blackburnthanks to gorbachew :D23:59
--- Log closed Fri Aug 19 00:00:00 2011