IRC logs of #shogun for Tuesday, 2011-07-05

--- Log opened Tue Jul 05 00:00:46 2011
-!- f-x [~user@117.192.204.74] has quit [Read error: Connection reset by peer]00:05
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.]00:09
CIA-32shogun: Shashwat Lal Das master * r8eec45c / (6 files): Implemented overloading in streaming sparse and string files. Removed the get_vector template declaration in them. - http://bit.ly/l2rRpy08:38
CIA-32shogun: Soeren Sonnenburg master * ra5bffe3 / (18 files in 3 dirs):08:38
CIA-32shogun: Merge pull request #169 from frx/streamingfile08:38
CIA-32shogun: StreamingFile structure changes, updated for String and SparseFeatures - http://bit.ly/m5yodX08:38
CIA-32shogun: Soeren Sonnenburg master * r93ea9dd / src/libshogun/preprocessor/ClassicMDS.cpp : Fix compilation error in classic mds. - http://bit.ly/liWWeI08:44
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun09:49
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]09:51
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]09:56
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun10:01
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun11:09
heikohi there11:19
heikoanybody here?11:19
heikoI got a lot of linker errors11:19
heikoCStreamingFile11:20
@sonney2kyes here11:37
@sonney2kheiko, did you do a git clean -dfx / make distclean?11:39
@sonney2kI compiled for *all* interfaces w/o trouble...11:40
heikocurrently rebuilding :)11:40
bettyboosmart11:40
@sonney2kindeed :)11:42
heikohehe :)11:43
heikosonney2k,11:43
@sonney2kyes?11:43
heikowhich feature type is most important next to simple/string and sparese11:43
heikosparse11:43
@sonney2kthat's all11:43
@sonney2kwe have no more than that11:43
heikohuh but all these that inherit from dotFeatures?11:44
@sonney2kheiko, don't get confused by the streaming/online features  - I don't see a way to do model selection with these11:44
heikoWhat about for example PolyFeatures?11:45
@sonney2kheiko, then I would rather modify dotfeatures directly11:45
heikobut it is abstract11:45
heikoand for example the get_num_vectors method is not implemented in there11:45
heikoI may also just move that method to DotFeatures (if its possible)11:46
@sonney2kheiko, I would say - postpone for now. if things get too complex I would rather modify the subset selection (i.e. one should then manually specify feature objects for training / validation / test)11:48
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun12:09
blackburnsonney2k: thanks for fixing it, strange - was working12:17
@sonney2kblackburn, weird it is a &int vs int - shouldn't work - anyways12:30
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds]12:30
blackburnokay12:30
@sonney2kheiko?12:30
@sonney2kargh12:30
@sonney2kjust wanted to say sth12:30
-!- f-x [~user@117.192.193.217] has joined #shogun13:00
-!- in3xes [~in3xes@180.149.49.227] has quit [Remote host closed the connection]13:20
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun13:22
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun13:41
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun13:52
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 263 seconds]13:55
blackburnsonney2k: heiko13:55
blackburn:)13:55
bettyboo:Q13:56
heikohey13:57
heikowhat?13:57
heikoCould anyone of you help me a little with setting up python_modular for my new classes - i.e. the modelselection folder?13:58
blackburnheiko: soeren wanted to say sth to you13:58
-!- in3xes_ is now known as in3xes13:58
heikook13:58
blackburnheiko: how can I help you?13:59
heikodid not read anything here13:59
heikoblackburn: I want to create some python examples for my modsel stuff13:59
blackburnyes13:59
heikoand I am not sure how this is done13:59
heikoI need a new Modelselection.i file right?13:59
blackburnyeap13:59
blackburnand Modelselection_includes.i14:00
blackburnin modelselection_includes.i just write14:01
blackburn%{14:01
blackburn#include <shogun/modelselection/blablabla.h>14:01
blackburn...14:01
blackburn%}14:01
blackburnin modelselection.i just do as it done in e.g. preprocessor.i14:01
heikook14:01
heikowill try14:01
heikoand then?14:01
heikoeverything ready?14:01
blackburnthen just compile it with python_modular14:01
blackburnyes14:01
heikook14:02
heikocool14:02
heikothanks14:02
blackburnyes pretty cool thing14:02
blackburnjust make a short test14:02
blackburnfrom shogun.Modelselection import [something]14:02
blackburnif it will pass - you could improve it a little :)14:02
heikoyes ok :)14:03
* blackburn should stop using if + future tense14:03
heikoehm :D14:03
heikoyes, but I understood anyways14:03
blackburnit is because you are non-native too :D14:03
blackburnheiko: btw is it the same in german? (I don't know any german grammar except the name of Partizip Zwei)14:08
blackburnI mean is it 'future' in conditionals?14:08
heikoI think not14:09
heikobut I am not the language specialist .)14:09
heikonot sure, but I think its not possible14:10
blackburnI see14:16
blackburnin russian, it is a typical construction: sth like "if I will blablabla"14:17
blackburnso I make mistakes automatically hehe14:17
bettyboo:) blackburn14:17
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 252 seconds]14:35
heikoblackburn, how is the _wrap.cxx file created?14:35
-!- in3xes_ [~dfrx@213.155.190.141] has joined #shogun14:47
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun14:48
-!- in3xes_ [~dfrx@213.155.190.141] has quit [Quit: ZNC - http://znc.sourceforge.net]14:55
-!- sploving1 [~sploving@2001:cc0:2020:2022:5eff:35ff:fe04:f091] has joined #shogun14:56
sploving1hi sonney2k14:56
sploving1coud not apply org.shogun.BaseKernel  save(org.shogun.SWIGTYPE_p_shogun__CFile)  to  (org.shogun.AsciiFile)14:57
-!- f-x- [~somebody@213.155.190.141] has joined #shogun15:06
-!- in3xes [~in3xes@180.149.49.227] has quit [Quit: Ooops..!!]15:10
-!- sploving1 [~sploving@2001:cc0:2020:2022:5eff:35ff:fe04:f091] has left #shogun []15:15
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun15:20
@sonney2kheiko, I was thinking about the DotFeatures stuff15:45
@sonney2kyou actually can call get_num_vectors()15:45
@sonney2kso you could do the subsetting if you want15:46
@sonney2kand then you 'only' need to change the various add_to_dense_vec etc functions15:46
heikoyes15:46
@sonney2kthis would add support for all kinds of features like simple/sparse immediately15:46
@sonney2khowever, simple/sparse have some overloaded get_vector functions IIRC15:47
@sonney2kso these still need to be changed15:47
@sonney2kso much about this15:47
@sonney2kbtw, dynarray is sth I wrote... I think gunnar confuses this with CArray - but could very well be that he did the copy constructor or D.L. did :)15:48
heikoso just implement rhis method in DotFeatures15:48
heikook :)15:48
@sonney2kblackburn, thanks for explaining the swig things to heiko15:48
@sonney2kheiko, but continue the swig thing while you are on it :)15:49
heikosonney2k, did you read my mail about the DynArray<ParameterCombination> from python?15:49
heikook, will do :)15:49
heikoAh sorry, I have to go now, give a math lesson. will be back in an hour15:51
heikosee you15:51
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds]15:56
@sonney2kblackburn, do you have any idea what sploving wanted to say above?16:03
@sonney2kseems he is gone already...16:03
CIA-32shogun: Baozeng Ding master * rfe784a5 / (8 files): add some kernel exmaples for java_modular - http://bit.ly/jYjov516:04
CIA-32shogun: Soeren Sonnenburg master * r41a8c99 / (8 files):16:04
CIA-32shogun: Merge pull request #170 from sploving/master16:04
CIA-32shogun: add some kernel example for java_modular - http://bit.ly/kIADfX16:04
blackburnsonney2k: sorry fell asleep :D16:12
@sonney2kblackburn, office sleep - nice :)16:15
blackburnsonney2k: what about sploving? can't see any msgs from him16:15
@sonney2k<sploving1> hi sonney2k16:15
@sonney2k<sploving1> coud not apply org.shogun.BaseKernel  save(org.shogun.SWIGTYPE_p_shogun__CFile)  to  (org.shogun.AsciiFile)16:15
-!- sploving1 [~sploving@2001:cc0:2020:2022:5eff:35ff:fe04:f091] has joined #shogun16:16
blackburnah16:16
@sonney2ksploving1, ahh16:16
sploving1hi16:16
@sonney2kyou left without explaining16:16
@sonney2k<sploving1> hi sonney2k16:16
@sonney2k<sploving1> coud not apply org.shogun.BaseKernel  save(org.shogun.SWIGTYPE_p_shogun__CFile)  to  (org.shogun.AsciiFile)16:16
@sonney2k?16:16
@sonney2kwhat does this mean?16:16
sploving1a compile error16:16
sploving1the example is :16:16
@sonney2ksploving1, occurring when?16:16
@sonney2k(it compiles here)16:17
@sonney2kshogun compile or java example?16:17
@sonney2kdetails man details please!16:17
sploving1java exmaple16:17
sploving1kernel_io_modular16:18
sploving1System.loadLibrary("Library");16:18
@sonney2kI see16:18
@sonney2kI guess the same issue like always16:18
@sonney2klet me check16:18
sploving1it happens at "kernel.save(f);" OKay16:18
* sonney2k recompiles16:19
-!- alesis-novik [~alesis@188.74.87.206] has joined #shogun16:20
@sonney2ksploving1, fixed16:20
CIA-32shogun: Soeren Sonnenburg master * rb4af03b / src/modular/Kernel.i : add missing %import to fix compilation issue of kernel_io_modular - http://bit.ly/lZ5yBB16:20
sploving1okay.16:20
sploving1thx.16:20
@sonney2ksploving1, it is always the same thing and fix16:20
sploving1now I am working  testcase.16:21
sploving1which file do i need add first?16:21
sploving1test_one.java?16:21
@sonney2ksploving1, did you already do all examples by now?16:21
@sonney2ksploving1, I would write two scripts: tester.java and generator.java - that's all16:21
@sonney2k(no more that is)16:21
@sonney2kf-x, please send a patch to fix the macroish definition16:22
sploving1not yet. they are simple but trival. I would add them not in a day16:22
@sonney2kf-x, and btw thanks for the polished patch16:22
@sonney2ksploving1, how many are left?16:22
sploving1I did not count on them. sparse left and ushort are not supported now16:23
f-xsonney2k: ok.. and btw is the overloading problem fixed? i mean the confusion we had with the compiler using the base/derived class function16:24
@sonney2kf-x, I still don't understand how this can work...16:25
sploving1example are not difficult for me. I would work on  test_suite at the same time16:25
@sonney2ksploving1, ok16:25
@sonney2ksploving1, then just continue16:26
sploving1okay16:26
f-xsonney2k: it could be that i've overlooked some case, but i'm pretty sure it should work.. where's the problem as you see it?16:26
@sonney2kf-x, I mean we use get_vector(bool, int) etc and then we have get_vector(T, int) - I don't think this overloading works.... I am not sure either but it looks weird...16:30
f-xsonney2k: it does since get_vector(T, int) isn't templated... i just imagine each version of the templated class to contain one overloaded implementation of get_vector16:31
f-xwhich overloads only the get_vector version specific to the class type16:31
f-xif get_vector was templated within the class, it wouldn't work probably16:31
-!- blackburn [~blackburn@188.122.238.99] has quit [Read error: Connection reset by peer]16:38
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun16:38
@sonney2kf-x, is this documented somewhere? I mean don't get me wrong - if it is - we have a very elegant solution :)16:53
f-xsonney2k: i read about this kind of thing on the usual forums, i'll give you a link when i find one16:54
@sonney2kf-x, ok - for now I assume it compiles and works with g++ at least :)16:55
f-xsonney2k: compilation works and initial testing too :)16:56
f-xbut gotta go somewhere now16:56
f-xlater16:56
@sonney2kok cu16:56
-!- f-x [~user@117.192.193.217] has quit [Read error: Connection reset by peer]16:56
blackburnsonney2k: very lame question: why we use float64_t, blabla?? :D16:58
sploving1hello sonney2k,  a java simple question: fnam.find(supported) (python). does java string has similar function? I did not see it. http://download.oracle.com/javase/1,5.0/docs/api/java/lang/String.html17:05
bettyboohug17:06
blackburnfnam.find?17:07
sploving1yeap. test_one.py17:08
blackburnwhat should this function does?17:08
blackburnsonney2k: bam!17:09
blackburnndefined reference to `shogun::CStreamingFile::get_longreal_sparse_vector_and_label(shogun::SGSparseVectorEntry<long double>*&, int&, double&)'17:09
blackburncollect2: ld returned 1 exit status17:09
sploving1it findes if a substring exits17:10
blackburnindexOf17:11
blackburnah17:11
blackburnsubstring17:11
blackburnhm17:11
blackburnyes, it works with substrings too17:11
blackburnit returns -1 if there are no such substr17:12
sploving1indexof17:13
sploving1great17:13
sploving1blackburn, does java have strip method in string class?17:26
sploving1sonney2k, how to write this in java? name=line.split('=')[1].strip().split("'")[1]17:29
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun17:31
blackburnsploving1: split17:31
blackburnreturns String[]17:31
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Read error: Connection reset by peer]17:31
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun17:33
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Read error: Connection reset by peer]17:35
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun17:36
sploving1blackburn, I mean strip not splite17:37
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Read error: Connection reset by peer]17:37
blackburnwhat is the diff?17:37
sploving1strip could delete the whitespace or other character in the end of string17:38
sploving1"x   ".strip return "x"17:38
blackburnsplit will return the same if you will set regexp right17:40
sploving1blackburn, yeap. I will have a try17:43
@sonney2kblackburn, git clean -dfx17:43
@sonney2kas usual17:43
-!- sploving1 [~sploving@2001:cc0:2020:2022:5eff:35ff:fe04:f091] has left #shogun []17:49
blackburnsonney2k: it not helped18:33
blackburnClassifier_wrap.cxx: In function ‘void* _p_shogun__CArray3T_unsigned_long_long_tTo_p_shogun__CSGObject(void*, int*)’:18:34
blackburnClassifier_wrap.cxx:83572: error: ‘CArray3’ is not a member of ‘shogun’18:34
blackburnsonney2k: is it compiles on your machine?18:34
-!- in3xes [~in3xes@180.149.49.227] has quit [Quit: Leaving]18:39
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun18:40
-!- f-x- [~somebody@213.155.190.141] has quit [Quit: ZNC - http://znc.sourceforge.net]20:00
@sonney2kblackburn, hmmhh it did before let me check20:04
* sonney2k does git clean -dfx20:04
blackburnI have no idea ffdddddddddddddtrtryrrertj20:07
blackburnwas cleaning my keyboard :D20:07
@sonney2kindeed - so it is the %import game again20:08
* sonney2k sighs20:08
blackburnbitchy thing20:09
@sonney2kblackburn, do you recall how we fixed it last time20:09
@sonney2kI added %import "Library.i" to Kernel.i20:09
@sonney2kand I guess that this is now the culprit20:09
blackburnthere was a preprocessor thing20:09
blackburnand you added preprocessor_includes almost everywhere20:09
@sonney2kblackburn, I thought I removed this later - no?20:10
@sonney2klemme check20:10
@sonney2kahh I see20:10
@sonney2kscrew it  - I have no idea what is going on20:13
@sonney2klet me remove this %import and all the preproc imports20:13
blackburnhehe20:21
blackburnit is really strange20:21
blackburnwe add it, remove it ehhr20:21
@sonney2kblackburn, if you can make sense of this problem let me know20:22
blackburnsonney2k: of course20:33
blackburnsonney2k: I don't like this practice of including *_includes almost everywhere20:46
blackburncan we avoid this completely?20:46
@sonney2kblackburn, if I knew how I would do it20:46
blackburnhmm20:47
blackburnI will make a tri20:47
blackburntry20:47
@sonney2kblackburn, don't try more than a week :)20:51
blackburnseems adding Library_includes help, right?20:51
@sonney2kI dont' know20:52
blackburnI'll let you know in a minute20:52
blackburnwe should get some more developers!20:52
blackburnokay our bugfixes for most of errors is to add some include :D20:57
blackburnsegfault? add include!20:57
@sonney2kblackburn, or have everything in one file20:58
CIA-32shogun: Sergey Lisitsyn master * rfbbd60b / (src/modular/Classifier.i src/modular/Regression.i): Get this shit to compile - http://bit.ly/j1S1Rl20:58
blackburnthe whole shogun in one file could be funny20:58
@sonney2kblackburn, pretty polite commit msg20:59
blackburnsonney2k: but informative :D20:59
@sonney2know there for eternity :D20:59
@sonney2kyeah - it basically says: if you do swig wrapping do one big blob21:00
@sonney2kand include everything in there21:00
blackburnthe most strangest thing - it emerges in some completely accidental moments21:01
@sonney2kblackburn, no21:04
@sonney2kblackburn, I needed to %import Libary in Kernel.i because Kernels need Files21:04
blackburnah21:04
@sonney2kand voila classifiers need kernels21:04
blackburnhmm21:04
@sonney2kand regression needs kernels21:04
@sonney2kso these includes were missing21:05
blackburnI see21:05
@sonney2kthe problem is that swig does not do recursive inclusion without the -includeall switch21:05
@sonney2kbut when I enabled that switch it tried to wrap also glibc and other things21:05
blackburnfrom linux.Kernel import Drivers :D21:06
blackburnsonney2k: I think it is ok - we don't really add those 'modules' every day21:08
@sonney2kblackburn, yes only as long as sploving is working on the java examples21:08
blackburnsonney2k: what is his mid-term task&21:09
@sonney2kjava_modular21:11
blackburnand next?21:11
blackburnruby?21:11
@sonney2kmissing examples and 2 paragraphs documenting how to use it21:11
@sonney2kahh and tester.java / generator.java21:12
blackburnthe whole project is about java??21:12
@sonney2kso that the scripts can be tested21:12
@sonney2kafter mid-term lua and ruby21:12
blackburn2 paragraphs documenting? ;D21:12
blackburnoh sh~21:12
@sonney2kblackburn, yeah it is just java for  this http://www.shogun-toolbox.org/doc/modular_tutorial.html and http://www.shogun-toolbox.org/doc/installation.html21:13
blackburnsonney2k: do you have some snippet related to some dna possible to port as graphical example in clojure?21:14
blackburnI'm pretty interested in clojure and I will make some pretty example with it some day21:15
@sonney2kgraphical example of DNA?21:15
blackburnyes why not?21:15
blackburnthe first idea came in my mind: you did some splice site detection, right?21:16
blackburncorrect me if I go wrong, it is some chunk of dna and we can show/zoom it somehow21:17
blackburnno?21:17
blackburnI believe that pretty graphical examples is a key to popularity :D21:19
-!- f-x [~somebody@213.155.190.141] has joined #shogun21:25
-!- f-x is now known as Guest3207921:25
in3xesGuest32079, got yourself a shellmix account? Can I use that znc service too?21:40
in3xeslooks like a free service21:42
-!- blackburn [~blackburn@188.122.238.99] has quit [Remote host closed the connection]22:06
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun22:06
-!- f-x` [~user@117.192.193.217] has joined #shogun22:07
-!- Guest32079 [~somebody@213.155.190.141] has quit [Quit: ZNC - http://znc.sourceforge.net]22:13
-!- f-x` is now known as f-x22:17
CIA-32shogun: Sergey Lisitsyn master * r0440c90 / (src/libshogun/lib/arpack.cpp src/libshogun/lib/arpack.h): Arpack wrapper improvements - http://bit.ly/nLmmJC22:30
CIA-32shogun: Sergey Lisitsyn master * rf97613f / src/libshogun/lib/arpack.cpp : Removed junk at arpack wrapper - http://bit.ly/pU5BT822:32
blackburni got tired with these arpack affairs..22:35
-!- blackburn1 [~blackburn@188.122.238.99] has joined #shogun22:39
-!- blackburn [~blackburn@188.122.238.99] has quit [Ping timeout: 255 seconds]22:42
-!- blackburn1 [~blackburn@188.122.238.99] has quit [Read error: Connection reset by peer]22:48
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun22:49
-!- f-x__ [~f-x@213.155.190.131] has joined #shogun22:50
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.]23:04
--- Log closed Wed Jul 06 00:00:49 2011

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!