IRC logs of #shogun for Tuesday, 2011-07-19

« Monday, 2011-07-18 Index Wednesday, 2011-07-20 »
--- Log opened Tue Jul 19 00:00:39 2011
@sonney2kok now with SGIO00:07
CIA-87shogun: Soeren Sonnenburg master * r6c8f9ca / (203 files in 30 dirs): rename IO class to SGIO to avoid IO (module) vs. IO (class) naming clash - http://bit.ly/njE6Nl00:07
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.]00:18
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun00:19
@sonney2knite blackburn / serialhex00:24
blackburnsee you00:24
* sonney2k is happy with the new code structure00:24
blackburnhhe00:24
@sonney2kblackburn, now letst see if yo manage to get the java examples to compile & run :D00:24
blackburnlet's try00:25
blackburnbut not today ;)00:25
* sonney2k closes one eye00:25
* sonney2k closes one more eye00:25
* sonney2k ...zzzzZZZZZZ00:25
serialhexhmm?? someone asked for me???00:26
blackburnserialhex: I wonder why he did ;)00:28
blackburnhow are you today?00:28
serialhexidk blackburn...00:28
serialhexi'm good00:28
serialhexgot a headache, probably from not enough caffene00:28
serialhexand yourself??00:28
blackburnbe carefull with caffeine ;)00:28
serialhexi love it!00:28
blackburnI'm fine, today met my friend, he is going to usa in next month00:29
blackburnto wisconsin-madison university00:30
blackburnone more russian to your country ;)00:30
serialhexcool :)00:30
bettyboo:*)00:30
blackburnserialhex: tv says your government jokes everyday about these 14billions hehe00:31
serialhexwhat 14 billions?00:32
blackburnof us debt00:32
serialhexoh, those 14 billions... i don't watch much news so i dont *know* but tyhere are soem people taking it sereously, and others probably aren't00:33
serialhexand afaik, it's >> 14billion00:33
blackburnyeap00:33
serialhexthe 14bil is the tip of the iceburg00:33
blackburnI simply can't believe they got in troubles :D00:34
bettyboosmart <:*)00:34
blackburnare they?00:34
serialhexno, they're not smart... and i can believe it... it's fucking stupid!00:35
serialhexanyway, i gotta go for a bit.. i'll catch ya later00:36
blackburnI'm going to bed just like soeren did :)00:36
blackburnsee you tomorrow I guess00:37
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.]01:04
-!- alesis_ [~alesis@188.74.87.206] has quit [Quit: Leaving]05:10
-!- alesis-novik [~alesis@188.74.87.206] has joined #shogun05:10
-!- mlsec [~mlsec@bane.ml.tu-berlin.de] has quit [Read error: Operation timed out]05:32
-!- mlsec [~mlsec@bane.ml.tu-berlin.de] has joined #shogun05:32
-!- mode/#shogun [+o mlsec] by ChanServ05:33
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun07:24
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 276 seconds]07:28
-!- in3xes1 [~in3xes@180.149.49.227] has joined #shogun07:33
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]07:36
-!- gsomix [~gsomix@178.45.80.131] has joined #shogun08:01
CIA-87shogun: Alesis Novik master * rddc2e2a / (2 files): Added examples - http://bit.ly/qLRvrb08:29
CIA-87shogun: Soeren Sonnenburg master * r2a8980b / (2 files):08:29
CIA-87shogun: Merge pull request #209 from alesis/gmm08:29
CIA-87shogun: Added examples for GMM - http://bit.ly/owNCUE08:29
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun08:49
-!- in3xes1 [~in3xes@180.149.49.227] has quit [Ping timeout: 252 seconds]08:52
-!- in3xes_ is now known as in3xes08:54
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun08:59
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun08:59
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 276 seconds]09:03
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds]09:28
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun09:57
CIA-87shogun: Shashwat Lal Das master * r48366d4 / src/shogun/lib/StreamingAsciiFile.cpp : Corrected get_vector and get_vector_and_label to work with new non-null terminated strings from read_line. - http://bit.ly/qnVoMI10:04
CIA-87shogun: Shashwat Lal Das master * r9c42aa2 / (6 files in 2 dirs): Transition from file streams to descriptors for streaming features. - http://bit.ly/oSexYB10:04
CIA-87shogun: Shashwat Lal Das master * r5f2d2b1 / src/shogun/io/StreamingFile.h : Correction in include path in StreamingFile. - http://bit.ly/pdvvfO10:04
CIA-87shogun: Soeren Sonnenburg master * r0a35e71 / (6 files in 2 dirs):10:04
CIA-87shogun: Merge pull request #210 from frx/streaming_110:04
CIA-87shogun: Transition to file descriptors in StreamingFeatures - http://bit.ly/n4kvLy10:04
CIA-87shogun: Soeren Sonnenburg master * r6d22179 / (6 files in 2 dirs): Merge branch 'master' of github.com:shogun-toolbox/shogun - http://bit.ly/ohQUVZ10:05
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun10:06
@sonney2kheiko, in case you are not asleep10:09
@sonney2kI did the renaming10:09
heikosonney2k, hi :)10:09
heikonot asleep10:09
@sonney2ki.e. we have a mathematics folder now10:09
heikohere since one hour :)10:09
heikook cool10:09
heikoI just made the statistics code compiling10:09
@sonney2kso feel free to add some statistics class in there10:09
heikostudents t should work now10:09
@sonney2kand for modular interfaces we now have a Mathematics module too10:10
@sonney2kvery good10:10
heikocool thing :)10:10
heikowill send a pull request later, but I have to test the stuff first10:10
@sonney2kalso we have a new separate io / IO module10:10
@sonney2kand lib now really contains only data structures and other helpers10:10
@sonney2kno longer 80+ misc files10:11
blackburnhi there10:12
bettyboomoin10:12
blackburnsonney2k: after some investigation we've got that C5.0 hard-coded to use files10:13
@sonney2kblackburn, nice - so it is not a long way that you can use shogun's features ;)10:13
blackburnI guess some io of C5.0 should be implemented another way10:14
@sonney2kblackburn, i/o should go away completely...10:15
@sonney2kheiko, great!10:16
blackburnokay gtg now, will buy some flowers for my mom hehe10:22
blackburnshe has a jubilee10:22
@sonney2kblackburn, enjoy!10:23
-!- gsomix [~gsomix@178.45.80.131] has quit [Read error: Connection reset by peer]10:33
-!- gsomix [~gsomix@85.26.232.237] has joined #shogun10:33
-!- gsomix [~gsomix@85.26.232.237] has quit [Ping timeout: 255 seconds]10:56
-!- gsomix [~gsomix@88.200.245.104] has joined #shogun11:02
gsomixforeach in c-code. it's beautiful.11:09
-!- bettyboo [~bettyboo@bane.ml.tu-berlin.de] has quit [Ping timeout: 240 seconds]11:22
-!- mlsec [~mlsec@bane.ml.tu-berlin.de] has quit [Ping timeout: 240 seconds]11:22
-!- mlsec [~mlsec@bane.ml.tu-berlin.de] has joined #shogun11:27
-!- mode/#shogun [+o mlsec] by ChanServ11:27
-!- bettyboo [~bettyboo@bane.ml.tu-berlin.de] has joined #shogun11:37
-!- mode/#shogun [+o bettyboo] by ChanServ11:37
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds]11:50
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.]11:59
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun11:59
-!- sploving1 [~sploving@124.16.139.194] has joined #shogun12:30
sploving1sonney2k, I added two lua examples , lua doc and load.lua. one examples works well. the other one met a problem12:31
sploving1distance_braycurtis_modular.lua, this modular,12:31
@sonney2kjust saw your pull request12:31
sploving1dm_train=distance:get_distance_matrix(). It sadi cannot find get_distance_matrix method12:31
@sonney2ksploving1, can you try a kernel example instead12:32
@sonney2kI might not yet have converted the distance* stuff12:32
sploving1I saw Distance_wrap.cxx file: BaseDistance_get_distance_matrix, but its subcalss get_distance_matrix12:34
sploving1has no12:34
sploving1such method12:34
CIA-87shogun: Baozeng Ding master * r8590caf / examples/undocumented/lua_modular/load.lua : add load.lua - http://bit.ly/r1ECLT12:37
CIA-87shogun: Baozeng Ding master * r1bebd27 / doc/pages/ModularTutorial.mainpage : add lua doc for lua path set - http://bit.ly/nz6W6q12:37
CIA-87shogun: Baozeng Ding master * r190958e / examples/undocumented/lua_modular/classifier_averaged_perceptron_modular.lua : add one classifier example - http://bit.ly/q0KA5Y12:37
CIA-87shogun: Baozeng Ding master * r7118c47 / examples/undocumented/lua_modular/distance_braycurtis_modular.lua : add one distance example - http://bit.ly/neiO6d12:37
CIA-87shogun: Soeren Sonnenburg master * r3c9006f / (4 files in 2 dirs):12:37
CIA-87shogun: Merge pull request #211 from sploving/master12:37
CIA-87shogun: add load.lua, two examples and lua doc - http://bit.ly/pfY4jL12:37
@sonney2ksploving1, could you try a kernel example for now?12:37
sploving1working12:38
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Read error: Operation timed out]12:40
sploving1kernel_gaussian_modular.lua:16: attempt to call method 'get_kernel_matrix' (a nil value)12:41
sploving1sonney2k, the same error12:41
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun12:52
-!- sploving1 [~sploving@124.16.139.194] has left #shogun []12:57
@sonney2kgsomix, foreach?13:10
blackburnsonney2k: yes, there are a lot of ForEach(..) in C5.013:10
gsomixmacro definitions.13:10
@sonney2k*urgs*13:11
@sonney2kseems to be a pure beauty13:11
CIA-87shogun: Baozeng Ding master * r909ad5f / examples/undocumented/lua_modular/kernel_gaussian_modular.lua : add one kernel example - http://bit.ly/naAilX13:11
CIA-87shogun: Soeren Sonnenburg master * rb908d51 / examples/undocumented/lua_modular/kernel_gaussian_modular.lua :13:11
CIA-87shogun: Merge pull request #212 from sploving/master13:11
CIA-87shogun: add a lua kernel example - http://bit.ly/oxFdaB13:11
-!- in3xes1 [~in3xes@180.149.49.227] has joined #shogun13:13
CIA-87shogun: Soeren Sonnenburg master * r2682d26 / src/configure : enable java autodetection - http://bit.ly/o4a76T13:13
-!- in3xes_ [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]13:16
-!- gsomix [~gsomix@88.200.245.104] has quit [Ping timeout: 260 seconds]13:26
-!- gsomix [~gsomix@80.234.26.210] has joined #shogun13:38
CIA-87shogun: Soeren Sonnenburg master * r584c096 / src/.generate_link_dependencies.py : fix build failure w/ {IO,Mathematics,ModelSelection}_doxygen.i - http://bit.ly/rnkltR13:39
-!- blackburn [~blackburn@188.122.238.99] has quit [Ping timeout: 255 seconds]14:00
-!- sploving1 [~sploving@124.16.139.194] has joined #shogun14:01
sploving1sonney2k, what is the matter?14:01
@sonney2ksploving1, thx14:01
@sonney2khow do I run an example14:01
@sonney2k?14:01
sploving1lua *.lua14:01
@sonney2khow do I set the lua path?14:01
sploving1export LUA_PATH=../../../src/interfaces/lua_modular/?.lua\;?.lua14:01
sploving1export LUA_CPATH=../../../src/interfaces/lua_modular/?.so14:01
sploving1I gave it in the doc14:02
sploving1I pushed it in the doc file. maybe you do not see it14:02
@sonney2kyes I missed it14:03
@sonney2kdoes any of the examples work?14:03
sploving1I pushed the kernel example. not work14:04
@sonney2ksploving1, yes14:05
@sonney2kbut is there *any* that works14:05
@sonney2kI couldn't find any so far14:06
sploving1the classifier example I pushed work14:06
CIA-87shogun: Soeren Sonnenburg master * r33f0c26 / examples/undocumented/lua_modular/check.sh : add lua example check script - http://bit.ly/oIsOgh14:06
@sonney2kAveragedPerceptron ?14:07
@sonney2kit fails here too14:07
@sonney2klua: classifier_averaged_perceptron_modular.lua:17: attempt to call global 'AveragedPerceptron' (a nil value)14:07
sploving1sorry. I changed the shogun.lua14:08
sploving1forget to push it14:08
@sonney2kI see  - please push :)14:08
@sonney2ksploving1, so get_kernel_matrix() doesn't work right?14:10
sploving1pushed. yeap14:10
CIA-87shogun: Baozeng Ding master * r360ef02 / src/interfaces/lua_modular/shogun.lua : add AveragedPerceptron for classifier - http://bit.ly/qWJn2x14:10
CIA-87shogun: Soeren Sonnenburg master * r003fbf5 / src/interfaces/lua_modular/shogun.lua :14:10
CIA-87shogun: Merge pull request #213 from sploving/master14:10
CIA-87shogun: add AveragedPerceptron for classifier - http://bit.ly/qtKF9x14:10
@sonney2ksploving1, yes works now14:11
sploving1sonney2k, any idea about the get_kernel_matrix?14:11
@sonney2ksploving1, but why does classifier's apply function work and not get_kernel_matrix?14:11
@sonney2kboth functions are defined in some base class14:12
sploving1I will take a look at the wrapper file14:12
sploving1sonney2k, _wrap_AveragedPerceptron_set_learn_rate14:14
sploving1the wrapper file has such method14:14
sploving1but for the kernel, just Basekernel. you can have a check14:14
-!- blackburn [~blackburn@109.226.102.88] has joined #shogun14:14
sploving1just one _wrap_BaseKernel_get_kernel_matrix method sonney2k,14:16
@sonney2ksploving1, could you please do one quick test?14:17
@sonney2ksploving1, just add to Kernel.h some virtual function    virtual int foo() { return 0; }14:17
@sonney2kand check if it appears in the other kernels - and so visible to lua?14:18
sploving1no typemap change?14:18
@sonney2ksploving1, yes just modify src/shogun/kernel/Kerne.h14:19
@sonney2kKernel.h14:19
@sonney2kI simply want to see if that function appears14:19
@sonney2kif it does you can let foo return SGMatrix<float64_t> or SGVector<float64_t> - just for testing14:20
sploving1my notebook is very slow14:21
@sonney2ksploving1, so we have to wait a little longer than14:22
-!- blackburn1 [~blackburn@109.226.102.88] has joined #shogun14:23
@sonney2kblackburn, I am in your LLE code currently14:24
@sonney2kwhat does this get_distance_matrix / delete distance do?14:24
blackburn1ehh?14:24
blackburn1where?14:25
-!- blackburn [~blackburn@109.226.102.88] has quit [Ping timeout: 260 seconds]14:25
blackburn1ah in apply?14:25
@sonney2kblackburn1, I am trying to 'fix' Distance with the issue sploving reported and so removed one occassion of get_distance_matrix14:26
blackburn1what is the issue?14:26
@sonney2kblackburn1, that you unfortunately use14:26
blackburn1isomap/mds uses it too btw14:26
-!- blackburn1 is now known as blackburn14:27
@sonney2kyeah messing around in there too now14:28
@sonney2kblackburn, you will have to check my code afterwards14:28
blackburnsonney2k: so what is the problem?14:28
@sonney2kstay tuned14:28
blackburnokay14:28
CIA-87shogun: Soeren Sonnenburg master * rdc9ddee / (4 files in 2 dirs):14:31
CIA-87shogun: remove get_distance_matrix(float64_t** dst, int32_t* m, int32_t* n) and14:31
CIA-87shogun: adjust preprocs accordingly - http://bit.ly/pXNN4S14:31
@sonney2kblackburn, please check this patch ^14:32
blackburneverything seems to be ok14:32
blackburnI'll run example in a min14:33
blackburnsonney2k: does it return new matrix now?14:33
blackburnI mean it is ok to delete distance after got it's matrix?14:34
@sonney2kblackburn, I don't know14:34
blackburnchecked14:35
blackburnyes14:35
blackburnCDistance object doesn't store it's matrices14:35
@sonney2kI think you should convert to this free_matrix  / and set  the free flag14:35
blackburnnot really different from the way it is..14:36
sploving1_wrap_AveragedPerceptron_set_learn_rate as set_learn_rate is method of AveragedPerceptron class, we have no _wrap_GaussianKernel_get_kernel_matrix as we have  no such mehtod in the class, just in its faterh class14:39
@sonney2kblackburn, well yes if there is the delete[] m.matrix then it will crash if it is not a new object - that can happen with custom distance matrix14:39
sploving1does this mean current typemy does not support inheriance of C++ method?14:39
@sonney2ksploving1, did you do the test above?14:40
sploving1not yet. just looking at the head file14:40
@sonney2ksploving1, that would tell us if the problem lies in inheritance or typemaps or whatever function name clash14:40
blackburnok it works this time hehe14:42
@sonney2kblackburn, please improve :)14:42
blackburnsonney2k: i would like to remove customdistance there14:43
@sonney2kblackburn, why?14:43
@sonney2kit makes a lot of sense to have it14:43
blackburnin fact I just use it for store matrix14:43
blackburnnot from shogun, from 'dimreductors' :)14:44
-!- serialhex [~quassel@99-101-148-183.lightspeed.wepbfl.sbcglobal.net] has quit [Remote host closed the connection]14:44
blackburnsonney2k: in fact LLE should not use any distances14:47
blackburnbut we don't have any neighbor tree yet14:48
@sonney2kok14:48
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun14:48
blackburnso I have to get matrix, sort it's rows and so on14:48
-!- sploving1 [~sploving@124.16.139.194] has left #shogun []14:49
-!- in3xes1 [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]14:51
blackburnsonney2k: okay distance matrix is always *new* matrix, right?14:54
@sonney2knot always14:54
@sonney2kfor custom matrices it can just return the ptr.14:54
blackburnehrr14:54
blackburnokay for euclidian it is?14:55
@sonney2kcurrently a definitve yes14:58
blackburnokaay14:59
blackburni'm now moving some code to find_null_space14:59
blackburnuseful thing for other LLEs :)14:59
CIA-87shogun: Sergey Lisitsyn master * r9fb1c7c / (2 files): Design improvement for LocallyLinearEmbedding - http://bit.ly/n5Rm4h15:27
blackburnsonney2k: can't push to google cod15:28
blackburne15:28
@sonney2kwhy not?15:29
blackburnerror: GnuTLS recv error (-9): A TLS packet with unexpected length was received. while accessing https://code.google.com/p/shogun-toolbox//info/refs15:29
@sonney2kaha15:29
@sonney2kwhcih git version do you have?15:29
blackburngit version 1.7.0.415:30
@sonney2kok, so git is not the problem15:30
@sonney2kblackburn, I guess you should try to create a git repository at code.google.com to see if ti works at all15:30
blackburnshhhhh15:31
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun15:36
@sonney2khmmh sploving is gone already...15:37
-!- sploving1 [~sploving@124.16.139.194] has joined #shogun15:40
sploving1sonney2k, what is going on?15:40
sploving1I left for some issue15:40
@sonney2ksploving1, I managed to get the example to work but it is weird15:40
@sonney2ksploving1, when I remove the require shogun but add instead15:41
@sonney2krequire("Distance")15:41
@sonney2krequire("Features");15:41
@sonney2kit works15:41
sploving1not change other file, such as typemap or head file?15:41
sploving1sonney2k, I will take a try15:42
@sonney2ksploving https://gist.github.com/109237315:42
@sonney2kthis thing15:42
@sonney2ksploving1, the problem is though that when I change the order of the requires it does not15:42
@sonney2ksploving1, like when I do15:43
@sonney2krequire("Distance")15:43
@sonney2krequire("Features");15:43
@sonney2kit works15:43
@sonney2kbut when I do15:43
@sonney2krequire("Features");15:44
@sonney2krequire("Distance")15:44
@sonney2kit does not15:44
sploving1sonney2k, I will take a try15:44
@sonney2ksploving1, have you ever seen this?15:45
sploving1not ever15:45
@sonney2kit seems like object hierarchy is destroyed when the order is 'wrong'15:46
blackburnsonney2k: what is the version of libgnutls installed on your machine?15:46
@sonney2kii  libgnutls26                          2.8.6-1                                the GNU TLS library - runtime library15:46
blackburnokay I have 2.8.515:46
blackburnwill try to update it somehow15:46
@sonney2kblackburn, debian is just so far ahead of ubuntu ;D15:47
blackburnif it don't help I will have no idea what is wrong15:47
@sonney2kblackburn, it looks like we have another swig problem with lua... the order of loading modules seems to decide whether things work or not15:49
blackburn:D15:49
-!- sploving1 [~sploving@124.16.139.194] has quit [Read error: Connection reset by peer]15:54
-!- sploving1 [~sploving@124.16.139.194] has joined #shogun15:58
sploving1my maches crashed just now15:58
sploving1here:require 'Features' require 'Distance' and require 'Distance' require 'Features'15:59
sploving1works well soney2k in my machine15:59
sploving1sonney2k, the order is the same for me16:00
blackburnsonney2k: got it16:00
sploving1no.16:00
sploving1I forget to cancelt --16:00
@sonney2ksploving1, I just had a look at Features_wrap.cxx and Distance_wrap.cxx16:01
@sonney2kdo you know what these swig_types are?16:01
@sonney2ksploving1, or how I can find out which type an object has?16:01
@sonney2kin lua I mean?16:01
sploving1sonney2k, the order cause the problem too in my machine16:01
sploving1print(type(x))16:02
@sonney2ksploving1, looking at the *_wrap.cxx files there seems to be some type info at the bottom16:02
@sonney2ke.g.16:02
@sonney2k#define SWIGTYPE_p_shogun__CSGObject swig_types[290]16:02
@sonney2k#define SWIGTYPE_p_shogun__CSGObject swig_types[170]16:02
@sonney2kin features / distance respectively16:02
@sonney2knot sure if there are no collisions16:03
@sonney2ksploving1, print(type(x)) just says userdata16:03
@sonney2ksploving1, did you ever hear about that swig_types[] array?16:04
sploving1no. sonney2k16:05
sploving1#define SWIGTYPE_p_EPrimitiveType swig_types[0]16:05
sploving1you can take a look at it in the begin of the wrapper file16:06
sploving1sonney2k, #define SWIGTYPE_p_shogun__CRealDistance swig_types[170]16:06
@sonney2kI guess BrayCurtisDistance can be found in Features_wrap.cxx because we %import Distance from there16:07
sploving1sonney2k, it actlly has . #define SWIGTYPE_p_shogun__CBrayCurtisDistance swig_types[169]16:09
sploving1although I have no idea about swig_type16:09
@sonney2ksploving1, looking at the code it seems to contain the type information16:11
@sonney2kwith name parameter infromation and some callback16:12
sploving1for the kernel_gaussian_modular.lua, require 'Kernel', require 'Features' works, but in another order, it does not work !16:12
@sonney2ksploving1, yeah...16:12
@sonney2ksploving1, I just did some experiment16:12
@sonney2kI removed the %import and %include stuff from Features.i16:13
@sonney2kthen it works16:13
@sonney2kno matter if I first require Features or Distance16:13
@sonney2ksploving1, how can I print a table?16:13
@sonney2kprint(dm_test) just says table: 0x13a717016:14
sploving1for k, v in pairs(z) do print (v) end16:14
sploving1in VectorTest we have such function16:14
sploving1for matrix table, MatrixTest.lua has it16:15
@sonney2kcomplicated - ok16:15
@sonney2kbut works16:15
sploving1sonney2k, when you removed %import or %include, does it affect other modular16:16
sploving1like python, or java16:16
@sonney2kblackburn, who would have thought that all these %include/imports hurt some interface16:16
@sonney2ksploving1, yes these were required for java to work16:16
@sonney2kwe can of course make an exception for lua but it is becomming messy16:17
sploving1why the order affect the result. so strange. sonney2k16:18
sploving1I think I will deep "require" in lua later16:19
@sonney2ksploving1, I suspect that we are doing sth wrong with all these %imports...16:19
sploving1sonney2k & blackburn, can you find a way that compatible to java and lua both16:21
sploving1of course include python and others16:21
blackburnno idea ;(16:22
sploving1I gtg. bye16:23
-!- sploving1 [~sploving@124.16.139.194] has left #shogun []16:24
@sonney2kblackburn, the reason seems to be that type thing16:32
blackburnwhict type thing?16:32
@sonney2konce sth is imported it is registered in the type info16:32
blackburnah'16:32
@sonney2kand it seems taht this is just different16:33
@sonney2kI guess all these problems would go away if we use just a single shogun module16:40
@sonney2kblackburn, ^ /me cries16:40
blackburnI don't know but you can't disagree that in fact the whole shogun is self-dependant16:40
@sonney2kblackburn, can you think of a way to e.g. resolve the CKernel <-> CDistance dependency?16:42
@sonney2kI mean currenty we have a Kernel that can take a distance as input16:42
@sonney2kand vice versa16:42
@sonney2kif we didn't have that - there was no problem16:42
blackburncan't mind any16:42
blackburn 16:44
@sonney2kblackburn, if we use just a single shogun module we will get a C++ wrapper that is 10MB in size16:44
@sonney2kwith a shogun.so of similar size16:44
blackburnit is smaller than we have now, right?16:45
-!- in3xes_ is now known as in3xes16:52
@sonney2kblackburn, sure smaller than the total17:01
blackburnpros:17:01
blackburnsmaller17:01
@sonney2kbut when you change 1 file this whole thing has to be recreated/compiled again17:01
blackburncons:17:01
blackburnwe want to cry17:01
blackburn:D17:01
@sonney2kanyway I don't see any alternative17:01
@sonney2kso I will try to create a shogun monster(tm)17:01
@sonney2kand then we will see if that helped or not17:02
blackburnokaaay17:02
@sonney2kblackburn, recall that is the thing I always tried to avoid and that was causing so much trouble for sploving and us and the crashes in octave_modular17:06
CIA-87shogun: Soeren Sonnenburg master * r6dab54f / (2 files): zero poim before computing it - http://bit.ly/n8stSI17:06
CIA-87shogun: Soeren Sonnenburg master * r8286ce3 / (2 files): Merge branch 'master' of github.com:shogun-toolbox/shogun - http://bit.ly/oMZnH417:06
blackburnit will compile forever..17:07
@sonney2kblackburn, or be faster than ever...17:07
blackburnhmm looks like it have many pros..17:07
@sonney2k+octave will start to work17:08
@sonney2k+lua will17:08
@sonney2k+the include mess will go away17:08
@sonney2k-compile time(?)17:09
blackburn:D no choice17:09
@sonney2kblackburn, we have choice if we manage to split up shogun into disjoint parts (that don't communicate with each other)17:10
blackburncan't see any way17:14
blackburnit is highly related on each other..17:14
@sonney2kme neither17:15
@sonney2kit would work if we had some common interchange format between modules17:15
@sonney2klike numpy matrices :)17:15
@sonney2kanyways I am compiling already17:16
@sonney2kswig is now at 400MB mem requirements17:16
@sonney2kthe c++ file is 13MB17:17
@sonney2kcompiled17:17
@sonney2k18MB17:18
@sonney2knot really slower than multiple files though...17:18
blackburnhm so let's do it17:18
@sonney2kshould we name it Shogun or shogun?17:21
blackburnshogun i guess17:22
@sonney2kanother +17:24
@sonney2k+resolves serialhex's ruby problem17:24
@sonney2klowercase module name :D17:24
@bettyboo;)17:24
blackburnoh haha17:24
@sonney2kthe BaseKernel vs Kernel clash too17:25
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds]17:37
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun17:37
-!- serialhex [~quassel@70-88-36-65-miami-fl.hfc.comcastbusiness.net] has joined #shogun17:40
@sonney2kblackburn, ok it works somehow17:40
blackburnhehe17:40
@sonney2kI have to move the *_includes.i into the respective fiels again though17:41
@sonney2kI guess I will first test if I can modularize it a bit17:41
@sonney2kanyway it is good news for octave an serialhex I guess :D17:41
@sonney2kgtg17:41
@sonney2kl8r17:41
-!- serialhex_ [~quassel@70-88-36-65-miami-fl.hfc.comcastbusiness.net] has joined #shogun17:42
-!- serialhex_ [~quassel@70-88-36-65-miami-fl.hfc.comcastbusiness.net] has quit [Remote host closed the connection]17:43
-!- serialhex [~quassel@70-88-36-65-miami-fl.hfc.comcastbusiness.net] has quit [Ping timeout: 255 seconds]17:45
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds]17:52
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun18:34
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds]18:37
-!- in3xes_ is now known as in3xes18:49
-!- alesis-novik [~alesis@188.74.87.206] has quit [Quit: "I'll be Bach"]19:11
-!- serialhex [~quassel@70-88-36-65-miami-fl.hfc.comcastbusiness.net] has joined #shogun19:23
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun19:29
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Client Quit]19:30
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun19:33
-!- gsomix [~gsomix@80.234.26.210] has quit [Quit: Ухожу я от вас (xchat 2.4.5 или старше)]19:42
-!- f-x [~user@117.192.198.162] has joined #shogun19:45
-!- f-x [~user@117.192.198.162] has quit [Ping timeout: 260 seconds]20:07
-!- blackburn [~blackburn@109.226.102.88] has quit [Ping timeout: 255 seconds]20:46
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds]20:47
-!- blackburn [~blackburn@109.226.102.88] has joined #shogun20:51
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun21:23
-!- serialhex [~quassel@70-88-36-65-miami-fl.hfc.comcastbusiness.net] has quit [Remote host closed the connection]21:24
CIA-87shogun: Sergey Lisitsyn master * r44cf499 / src/shogun/preprocessor/LocallyLinearEmbedding.cpp : Fixed z_matrix size at LocallyLinearEmbedding - http://bit.ly/qqvtYB21:44
CIA-87shogun: Sergey Lisitsyn master * r0454959 / (2 files): Merge branch 'master' of github.com:shogun-toolbox/shogun - http://bit.ly/qZSPhm21:44
-!- gsomix [~gsomix@80.234.26.210] has joined #shogun22:29
alesis-novikIs it ok if the non-graphical examples still generate the data themselves?23:08
@sonney2kalesis-novik, yes if they do it in a non-random way23:09
@sonney2kblackburn, around?23:09
blackburnsonney2k: yeap23:09
@sonney2kblackburn, the shogun.so is bigger than expected23:09
blackburnhow much?23:09
@sonney2k26MB23:10
@sonney2ksource code is 19MB23:10
alesis-noviksonney2k, I guess I should generate data somewhere else then. Now it's generated by sampling, which I guess is random-ish23:10
@sonney2kalesis-novik, just set the seeds23:10
@sonney2kblackburn, but the question now is how do we pretend to have a modular shogun :)23:11
@bettyboorahaha23:11
@sonney2kI don't like doing from shogun import *23:11
blackburnohhh23:11
blackburnit is the thing I had no thoughts about23:11
@sonney2kblackburn, I mean one can do import shogun23:13
@sonney2kand then shogun.GaussianKernel etc will be there23:13
blackburnyes, I understand23:13
alesis-noviksonney2k, cool, I'll look at that and should get the examples done soon. I assume they should look similar to the ones in undocumented/python_modular23:13
@sonney2kbut I would prefer to do from shogun import Kernel or so23:13
blackburnyes, I prefer that too23:13
@sonney2kblackburn, what we had with modules was really the suggested way when wrappers become too big23:14
@sonney2kalso the %include / %import way is suggested on the swig page...23:14
@sonney2kthat doesn't help much now though23:14
blackburnit will break *everything*..23:14
@sonney2kblackburn, no23:14
@sonney2kwe could of course emulate the previous behavior23:15
@sonney2kby simply doing23:15
blackburnhow?23:15
@sonney2kimport shogun as Kernel23:15
@sonney2ketc23:15
blackburnshitty way :D23:15
@sonney2kblackburn, another alternative would be to create a file e.g. Kernel.py with the kernel related objects23:16
@sonney2kthat would then import only those kernel related stuff23:16
@sonney2kbut that has to be done manually...23:16
blackburnbad bad23:17
@sonney2kblackburn, well actually we know - just the classes in Kernel.i that we rename to not have a C prefix :)23:20
blackburnaha23:20
-!- f-x [~user@117.192.198.162] has joined #shogun23:22
-!- gsomix [~gsomix@80.234.26.210] has quit [Ping timeout: 258 seconds]23:28
@sonney2kblackburn, my take on this that we give this a try - in the worst case we have to switch back to what we had. This won't be a user visible change - that is the good thing23:29
blackburnokay23:30
@sonney2kblackburn, how big is your compiled shogun python dir currently?23:31
@sonney2k(all .so files in total)23:31
blackburnblackburn@blackburn-laptop:~/shogun/shogun/src/interfaces$ du python_modular/23:32
blackburn134260python_modular/23:32
blackburnblackburn@blackburn-laptop:~/shogun/shogun/src/interfaces/python_modular$ du -hsc *.so23:33
blackburn...23:33
blackburn44Mtotal23:33
@sonney2kso 26 now is not that bad23:37
@sonney2kconsidering that this is all not yet stripped23:37
@sonney2kstripped 15M btw23:40
blackburn+ size23:54
blackburn;)23:54
blackburn- no modules..23:54
--- Log closed Wed Jul 20 00:00:44 2011
« Monday, 2011-07-18 Index Wednesday, 2011-07-20 »

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