IRC logs of #shogun for Sunday, 2013-06-09

« Saturday, 2013-06-08 Index Monday, 2013-06-10 »
--- Log opened Sun Jun 09 00:00:37 2013
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]01:01
-!- naywhayare [~ryan@spoon.lugatgt.org] has quit [Ping timeout: 248 seconds]01:51
-!- naywhayare [~ryan@spoon.lugatgt.org] has joined #shogun03:30
-!- nube [~rho@49.244.74.114] has quit [Quit: Leaving.]04:00
-!- naywhayare [~ryan@spoon.lugatgt.org] has quit [Ping timeout: 276 seconds]05:50
-!- naywhayare [~ryan@spoon.lugatgt.org] has joined #shogun05:57
-!- gsomix_ [~gsomix@95.67.165.97] has joined #shogun06:55
-!- gsomix [~gsomix@109.169.139.233] has quit [Ping timeout: 248 seconds]06:58
-!- naywhaya1e [~ryan@spoon.lugatgt.org] has joined #shogun09:35
-!- naywhayare [~ryan@spoon.lugatgt.org] has quit [Ping timeout: 276 seconds]09:38
-!- vgorbati [~vgorbati@212.2.159.34] has joined #shogun09:53
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun10:34
van51hey guys10:34
@lisitsynhey van5111:59
@lisitsynwhy the hell you are all with "_" gsomix_ sonney2k_ wiking_12:07
-!- lisitsyn is now known as lisitsyn_12:07
@lisitsyn_:D12:07
-!- vgorbati [~vgorbati@212.2.159.34] has quit [Quit: vgorbati]12:10
-!- gsomix_ [~gsomix@95.67.165.97] has quit [Ping timeout: 240 seconds]12:10
-!- vgorbati [~vgorbati@212.2.159.34] has joined #shogun12:37
-!- lisitsyn_ [~blackburn@109-226-114-235.clients.tlt.100megabit.ru] has quit [Quit: Leaving.]13:11
-!- nube [~rho@49.244.41.114] has joined #shogun15:14
-!- vgorbati [~vgorbati@212.2.159.34] has quit [Quit: vgorbati]15:37
-!- iglesiasg [~iglesiasg@213.143.50.92] has joined #shogun15:51
-!- mode/#shogun [+o iglesiasg] by ChanServ15:51
-!- vgorbati [~vgorbati@212.2.159.34] has joined #shogun16:31
-!- HeikoS1 [~heiko@nat-160-110.internal.eduroam.ucl.ac.uk] has joined #shogun17:25
-!- gsomix [~gsomix@185.6.239.170] has joined #shogun17:39
gsomixhey, hey, hey17:39
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has joined #shogun18:33
-!- mode/#shogun [+o lisitsyn] by ChanServ18:33
@lisitsynnow debian powered :D18:34
gsomixlisitsyn, huh.18:46
gsomixlisitsyn, how does it feel?18:47
@lisitsyngsomix: everything is ok but wifi was an issue18:47
* gsomix doesn't want prepare for exam. :(18:48
@lisitsynkaboom instant crash of gnome shell18:49
@lisitsyn:D18:49
@lisitsynrock stable fucker18:49
gsomixlol, it's gnome, baby18:49
@lisitsynahh my eyes18:52
gsomixlisitsyn, m? Шг.18:52
@lisitsyngsomix: totally18:55
@lisitsynno that's unacceptable19:00
@lisitsyntwo times per last 10 minutes19:00
gsomixlisitsyn, http://www.youtube.com/watch?v=MIZbGSXeWWE19:00
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has quit [Quit: Leaving.]19:08
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has joined #shogun19:20
-!- mode/#shogun [+o lisitsyn] by ChanServ19:20
@lisitsynokay repaired fonts19:20
@lisitsyntotally unusable someone has to do something with it19:20
-!- n4nd0 [~iglesiasg@213.143.50.92] has joined #shogun19:25
-!- iglesiasg [~iglesiasg@213.143.50.92] has quit [Ping timeout: 255 seconds]19:27
HeikoS1lisitsyn: see email19:34
@lisitsynHeikoS1: workshop?19:35
HeikoS1lisitsyn:  no clone method19:35
HeikoS1just wrote19:35
@lisitsynahh I see19:36
HeikoS1I think most of them are yours19:36
HeikoS1so the questoin is whether you want to have deep copies or not19:37
HeikoS1lisitsyn: apart from that19:38
HeikoS1clone is ready19:38
HeikoS1and equals, which is nice19:38
@lisitsynHeikoS1: yeah that's good19:39
@lisitsynHeikoS1: I'd have to check where it is used19:39
HeikoS1just search for "clone()" in all src files19:39
@lisitsynHeikoS1: thanks :D19:40
HeikoS1lisitsyn:  do you know of a way to create unit tests for *all* non abstract classes?19:43
HeikoS1with same content?19:44
HeikoS1I will just modify class_list.py I guess?19:44
@lisitsynHeikoS1: do you want to test all classes' clone?19:44
HeikoS1yep and then equals19:44
HeikoS1thats a useful thing to do I guess19:45
HeikoS1that at least all std things should work19:45
@lisitsynHeikoS1: yes and no :D19:45
@lisitsynbut rather yes19:45
@lisitsynHeikoS1: yeah I think it would be good to put it to class_list.py19:45
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has quit [Quit: Leaving.]19:48
wiking_fuckthis shiatz19:49
-!- wiking_ is now known as wiking19:49
-!- wiking [~wiking@info2k1.hu] has quit [Changing host]19:49
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun19:49
-!- mode/#shogun [+o wiking] by ChanServ19:49
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has joined #shogun19:49
-!- mode/#shogun [+o lisitsyn] by ChanServ19:49
@wikinglisitsyn: yo19:49
@lisitsynwiking: hey19:49
@wikingwanna cmake? )19:49
@wiking:)19:49
@lisitsynwiking:  depends what do you mean :D19:50
@wikingmmm fucking hell19:50
@lisitsynwiking: I am afraid of fucking hell ;)19:51
@wikingi have to still fix some headerstuff19:51
@wikingbljak this shiatz19:51
HeikoS1lisitsyn: are you familiar with this class_lust stuff?19:52
@wikingmmmm19:52
@wikingand of course my internet suxor19:53
@wikingHeikoS1: clone19:53
@lisitsynHeikoS1: class lust?19:53
@lisitsynhah19:53
HeikoS1wiking: its done19:53
@lisitsynah class list19:53
@wikingHeikoS1: i making cmake because i still dont have clone19:53
HeikoS1just needs to be integrated now19:53
@lisitsynHeikoS1: well touched it once19:53
@wikingHeikoS1: pushed?19:53
HeikoS1no19:53
HeikoS1cannot yet19:53
@wikingHeikoS1: pushit fuck19:53
@wikingi dont care if it's a feature19:53
@wikinganythign19:53
HeikoS1no not yet there are some clone methods in the code base, they need to be removed19:53
@wikingokok19:53
HeikoS1otherwise confusion19:53
@wikingjust push it19:53
HeikoS1also maybe you can help19:53
@wiking*PUSHIT*19:54
@wikingwhaddya need19:54
@wiking?19:54
HeikoS1I want a new file in the unit-tests19:54
HeikoS1which does19:54
HeikoS1TEST(SGObject,clone_CGaussianKernel)19:54
HeikoS1{19:54
HeikoS1CSGObject* object=new CGaussianKernel();19:54
HeikoS1CSGObject* clone=object->clone();19:54
HeikoS1EXPECT_TRUE(clone->equals(object));19:54
HeikoS1SG_UNREF(object);19:54
HeikoS1SG_UNREF(clone);19:54
HeikoS1}19:54
@wikingHeikoS1: and?19:54
HeikoS1for *all* classes in class_list.h19:54
@wikingah19:54
@wikingwell htat's easy19:54
HeikoS1should be quickly possible via class_list.py19:54
HeikoS1could you prepare that?19:54
@wikingnaaah dont fucking do that19:55
HeikoS1why not?19:55
@wikingone more time i see python script in a build process19:55
@wikingi will kill somebody19:55
@wikingwtf is this that we use python for BUILDING the sw19:55
HeikoS1wiking: there is already class_list.h anyway19:55
HeikoS1I need the very same includes19:55
@wikingHeikoS1: yeah i hate it19:55
@wikingit's like hell19:55
HeikoS1and one additional file in the unit tests19:55
@wikingnono19:55
HeikoS1where all these includes are in19:55
@wikingdont do it19:55
HeikoS1+ the gtest include19:56
HeikoS1I dont really care how this is done19:56
HeikoS1but I refuse to do it by hand19:56
@wikingHeikoS1: well it's easy to create one BIG Test()19:56
HeikoS1no seperate tests for each class19:56
HeikoS1also for each class that is ever added to shogun19:56
@wikingwell that's taken care by class_list.cpp.py19:56
@wikingobviously19:56
HeikoS1so I need the test now19:57
@wikingmmm i'll just check now the docs of gtest/gmock19:57
@wikingmaybe there's a way19:57
@wikingfucking hell i'm pushing a banch for the last 10 mins19:57
@wikinginternetz in 201319:57
@lisitsynguake is not working in debian19:58
@lisitsyn:D19:58
@wikingguake?19:58
@wikingquake?19:58
@lisitsynwiking: guake19:58
@wiking:)19:58
@lisitsynwiking: quake-like terminal19:58
@wikingwtf is guake?19:58
HeikoS1wiking how should gtest know all non-abstract shogun classes?19:58
@lisitsynwiking: I mean they have package but it is not starting at all19:58
HeikoS1the include can easily be done via #include "class_list.h"19:58
@wikinglisitsyn: lol19:58
HeikoS1but how to create all these tests?19:59
@wikingHeikoS1: it's not about that19:59
@wikinghttp://code.google.com/p/googletest/wiki/PumpManual19:59
@wikingi'm just checking this19:59
HeikoS1wiking: ok I leave this to you19:59
HeikoS1once its done and the existing clones are gone, I will push19:59
@wikingHeikoS1: should i send u the code when it's done?20:00
HeikoS1wiking: no just prepare a PR20:03
@wikingwe could create teh TEST with bumump20:03
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun20:03
shogun-notifier-shogun: Viktor Gal :feature/CMake * fa25547 / src/ (16 files): https://github.com/shogun-toolbox/shogun/commit/fa2554741ae8b4e2d2d20c1743b1033ee411375a20:03
shogun-notifier-shogun: LibShogun with cmake20:03
shogun-notifier-shogun: Viktor Gal :feature/CMake * 76de957 / / (32 files): https://github.com/shogun-toolbox/shogun/commit/76de9576a3334708677661f7141a16461726d88d20:03
shogun-notifier-shogun: Move cmake files to the root of the project20:03
shogun-notifier-shogun: add unit test support with cmake20:03
shogun-notifier-shogun: Viktor Gal :feature/CMake * 4902996 / .travis.yml,CMakeLists.txt,src/shogun/CMakeLists.txt,tests/unit/CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/49029960b297e7bdd2e51e0b874c847422e7be7820:03
shogun-notifier-shogun: Fix GMOCK source setting.20:03
shogun-notifier-shogun: change travis config for cmake20:03
HeikoS1wiking: or when its just one file, send it to me20:03
HeikoS1(or two files)20:03
@wikingayessssss20:04
@wikingpushed20:04
HeikoS1wiking: but this push did not include it, did it? ;)20:05
@wikingnopu20:06
@wikingthi sis cmake shiatz20:06
HeikoS1I see20:06
HeikoS1lisitsyn, wiking I will go home now. See you tomorrow!20:07
@wikingi guess this is the only way to do it20:07
@wikingA tool for preprocessor meta programming.  Useful for generating20:07
@wikingrepetitive boilerplate code.  Especially useful for writing C++20:07
@wikingclasses, functions, macros, and templates that need to work with20:07
@wikingvarious number of arguments.20:07
HeikoS1wiking: however you like :)20:07
@wikinganother python script20:08
@wikingfuck20:08
HeikoS1maybe ask sören then and do a nicer solution20:08
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has quit [Remote host closed the connection]20:09
-!- travis-ci [~travis-ci@ec2-54-242-93-196.compute-1.amazonaws.com] has joined #shogun20:10
travis-ci[travis-ci] it's Viktor Gal'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/793014420:10
-!- travis-ci [~travis-ci@ec2-54-242-93-196.compute-1.amazonaws.com] has left #shogun []20:10
HeikoS1wiking: you should wait for travid before merging stuff20:10
HeikoS1travis20:10
@wikingyes20:10
@wikingbut this is not yet ready20:10
HeikoS1I see, sorry then ;) going, bye!20:11
@wikingi'm just hoping somebody will hop on the cmake train and help out to finish it up20:11
HeikoS1I am very bad and very unpatient with those things20:11
HeikoS1hopefully it will make cross-platform compilation easier20:11
@wikinglol  g++ version is too old20:11
@wikingi wonder how did this happen20:11
HeikoS1wiking: but I trust you totally with that20:11
-!- HeikoS1 [~heiko@nat-160-110.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]20:13
-!- HeikoS1 [~heiko@nat-160-110.internal.eduroam.ucl.ac.uk] has joined #shogun20:14
shogun-notifier-shogun: Heiko Strathmann :sgobject_clone * 512d94f / src/shogun/lib/DataType.cpp,src/shogun/lib/DataType.h: https://github.com/shogun-toolbox/shogun/commit/512d94f6a6abb1a1fe43eb22d4dbe65e2347c85520:15
shogun-notifier-shogun: refactored sizeof_ptype, sizeof_stype to make them available from static context20:15
shogun-notifier-shogun: Heiko Strathmann :sgobject_clone * a5435a5 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/a5435a543f2265fc4409ac487f33fc72d7e7b66320:15
shogun-notifier-shogun: first parts of copy_ptype with unit tests20:15
shogun-notifier-shogun: Heiko Strathmann :sgobject_clone * 883ac63 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/883ac6316c1401812621067b8890205715c6b68420:15
shogun-notifier-shogun: -refactored compare_stype a bit20:15
shogun-notifier-shogun: -added copy_stype method with first bits20:15
shogun-notifier-shogun: -added first unit test for copy_stype20:15
shogun-notifier-shogun: Heiko Strathmann :sgobject_clone * a055e92 / src/shogun/lib/DataType.cpp,src/shogun/lib/DataType.h: https://github.com/shogun-toolbox/shogun/commit/a055e9220670cbabefbd1906810ba3a3b884916320:15
shogun-notifier-shogun: some refactoring of data type equals to compare wihtout lengths20:15
shogun-notifier-shogun: Heiko Strathmann :sgobject_clone * 1a42959 / src/shogun/base/Parameter.cpp,src/shogun/base/Parameter.h: https://github.com/shogun-toolbox/shogun/commit/1a429590529f9f4ea5b0c4d4847f6bbe90b79f3920:15
shogun-notifier-shogun: initial skeleton for copy method20:15
shogun-notifier-shogun: Heiko Strathmann :sgobject_clone * 11f008e / / (3 files): https://github.com/shogun-toolbox/shogun/commit/11f008e287ec3e9a2099a23ae03b04aba31ac47220:15
shogun-notifier-shogun: first parts of TParameter::copy with first unit test20:15
shogun-notifier-shogun: Heiko Strathmann :sgobject_clone * 2fc23a9 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/2fc23a9341f8a9bf15ee5241f515f7c5d822075020:15
shogun-notifier-shogun: -added copy for CT_MATRIX, CT_VECTOR20:15
shogun-notifier-shogun: -added copy for ST_STRING20:15
shogun-notifier-shogun: -unit-tests for those20:15
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has joined #shogun20:16
HeikoS1ouch20:16
HeikoS1no20:16
HeikoS1!!!20:16
@wikingwhat was this? :)20:16
lisitsynmy gnome shell is constantly dying! fucking debian20:16
HeikoS1I was stupid :(20:16
HeikoS1man20:16
@wikingHeikoS1: u pushed? :d20:16
HeikoS1wrong push20:16
@wikingwell its in now20:16
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has quit [Client Quit]20:17
@wikingreverting will be really fuckshit20:17
@wikingwtf is the version of that cmake on travis20:17
HeikoS1have to understand what happened now20:17
HeikoS1did I ever say that I hate git?20:18
@wikingHeikoS1: not yet20:18
HeikoS1ah phew20:18
HeikoS1this was pushed to another branch20:18
HeikoS1deleted it20:19
HeikoS1sorry for that20:19
-!- HeikoS1 [~heiko@nat-160-110.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]20:20
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has joined #shogun20:24
-!- travis-ci [~travis-ci@ec2-50-17-17-135.compute-1.amazonaws.com] has joined #shogun20:29
travis-ci[travis-ci] it's Heiko Strathmann'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/793039020:29
-!- travis-ci [~travis-ci@ec2-50-17-17-135.compute-1.amazonaws.com] has left #shogun []20:29
shogun-notifier-shogun: Viktor Gal :feature/CMake * a6f1f16 / CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/a6f1f16de04e6d97f32114215b100acaf11237b320:33
shogun-notifier-shogun: Fix gcc version checking20:33
shogun-notifier-shogun: Require 2.8.8 cmake or newer20:33
shogun-notifier-shogun: Viktor Gal :feature/CMake * db1e974 / .travis.yml: https://github.com/shogun-toolbox/shogun/commit/db1e9741787ab50f3dad69430ee210f9eb7d847b20:33
shogun-notifier-shogun: Fix travis exclude lines for cmake20:33
-!- travis-ci [~travis-ci@ec2-50-17-17-135.compute-1.amazonaws.com] has joined #shogun20:37
travis-ci[travis-ci] it's Viktor Gal'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/793072720:37
-!- travis-ci [~travis-ci@ec2-50-17-17-135.compute-1.amazonaws.com] has left #shogun []20:37
@wikingbooo20:40
shogun-notifier-shogun: Viktor Gal :feature/CMake * 5ba076f / CMakeLists.txt,cmake/CheckCompiler.cmake: https://github.com/shogun-toolbox/shogun/commit/5ba076f82f7c61e66a279cd677ccfe0b5e83bed620:44
shogun-notifier-shogun: Add workaround for compiler version check to support 2.8.7 cmake20:44
-!- travis-ci [~travis-ci@ec2-54-242-93-196.compute-1.amazonaws.com] has joined #shogun20:49
travis-ci[travis-ci] it's Viktor Gal'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/793091120:49
-!- travis-ci [~travis-ci@ec2-54-242-93-196.compute-1.amazonaws.com] has left #shogun []20:49
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has quit [Read error: No route to host]20:50
-!- HeikoS [~androirc@176.248.212.166] has quit [Read error: No route to host]20:51
-!- HeikoS [~androirc@176.248.212.166] has joined #shogun20:52
shogun-notifier-shogun: Viktor Gal :feature/CMake * 54a02e1 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/54a02e1db9b5a23a6ba2e8ec106dd7a4bff52ef020:52
shogun-notifier-shogun: Fix Eigen3 package name and backport FindJava, FindLibLZMA and20:52
shogun-notifier-shogun: FindDoxygen from 2.8.820:52
@wikingbljak20:54
-!- travis-ci [~travis-ci@ec2-107-22-45-75.compute-1.amazonaws.com] has joined #shogun20:56
travis-ci[travis-ci] it's Viktor Gal'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/793103420:56
-!- travis-ci [~travis-ci@ec2-107-22-45-75.compute-1.amazonaws.com] has left #shogun []20:56
-!- HeikoS [~androirc@176.248.212.166] has quit [Read error: Connection reset by peer]20:56
-!- HeikoS [~androirc@176.248.212.166] has joined #shogun20:57
-!- mode/#shogun [+o HeikoS] by ChanServ20:57
-!- HeikoS [~androirc@176.248.212.166] has quit [Remote host closed the connection]20:58
-!- HeikoS [~androirc@176.248.212.166] has joined #shogun20:58
-!- mode/#shogun [+o HeikoS] by ChanServ20:58
-!- n4nd0 [~iglesiasg@213.143.50.92] has quit [Quit: Leaving]21:18
gsomixsonney2k_, hey21:26
gsomixsonney2k_, https://github.com/shogun-toolbox/shogun/pull/113021:26
gsomixtest for CircularBuffer will be tomorrow. after my scary (very scary) little pass-fail exam. :)21:27
gsomixok, time to prerare >:321:38
-!- vgorbati [~vgorbati@212.2.159.34] has quit [Quit: vgorbati]21:44
-!- lisitsyn [~lisitsyn@109-226-114-235.clients.tlt.100megabit.ru] has joined #shogun21:53
-!- FSCV [~FSCV@189.139.252.135] has joined #shogun21:59
-!- HeikoS [~androirc@176.248.212.166] has quit [Remote host closed the connection]22:04
-!- gsomix [~gsomix@185.6.239.170] has quit [Ping timeout: 248 seconds]22:11
-!- lambday [67157d36@gateway/web/cgi-irc/kiwiirc.com/ip.103.21.125.54] has joined #shogun22:14
-!- mode/#shogun [+o sonney2k_] by ChanServ22:23
-!- sonney2k_ is now known as sonney2k22:23
@sonney2kwiking, ohh hey there I see you are enjoying the python fixup scripts?22:25
@sonney2kvan51, how is it going with the hashing?22:29
-!- HeikoS [~androirc@176.248.212.166] has joined #shogun22:29
-!- mode/#shogun [+o HeikoS] by ChanServ22:29
@HeikoSsonney2k, currently booking flights22:30
@sonney2kHeikoS, what was that push you did?22:30
@sonney2kabout?22:30
@HeikoS10.7 15.722:30
@HeikoSthat OK right?22:30
@sonney2kwwwhat?22:30
@sonney2kwell WS in on 12th22:30
@HeikoSjust double checking22:30
@HeikoSOk good:)22:31
@HeikoSsonney2k, push was an accident22:31
@HeikoSbut was a new branch, deleted it, so no changes ma22:31
@HeikoSwanted to push to my fork only22:32
@sonney2klisitsyn, gnome-shell is crashing on you? you serious? could it be that it crashes whenever you type a key in the overview thing?22:32
lisitsynsonney2k: totally random22:32
lisitsynsonney2k: I turned over from that unstable debian to stable ubuntu :D22:32
@sonney2klisitsyn, as in you just work not moving windows around?22:32
@sonney2klisitsyn, ahh ok then never mind22:33
lisitsynsonney2k: well22:33
lisitsynsonney2k: it hangs22:33
* sonney2k is enjoying debians stability22:33
lisitsynsonney2k: I can move pointer and etc22:33
lisitsynbut I have to switch and kill gnome-shell22:33
@sonney2kHeikoS, so what will it be?22:33
lisitsynthis is not something that should happen >1 per day22:33
@HeikoSsonney2k, ?22:34
@HeikoSsonney2k, what will it be?22:36
van51sonney2k: hey22:37
van51sonney2k: I have uploaded some code, it's mostly the skeleton for now22:38
van51sonney2k: it's here https://github.com/van51/shogun/commit/cf1cb59e51f6c162762bf89260a92458dd43141922:38
van51sonney2k: if you could provide some feedback, like if you agree with the design so far, it would be nice22:38
@sonney2kHeikoS, your patch22:39
@HeikoSsonney2k, clone is done....22:40
@sonney2kHeikoS, based on parameter framework I guess?22:40
-!- gsomix [~gsomix@83.234.54.9] has joined #shogun22:52
gsomixsonney2k, already asleep?22:53
@sonney2kvan51, the extra tokenizer class might be a bit overkill but lets see. in any case I think tokenizing just SGString<char> is sufficient22:57
@sonney2kvan51, I am a bit scared about us not having proper refcounting for SGString's yet and the overhead (even though small) we have when returning an SGString during tokenizing22:59
@sonney2kvan51, apart from that yes it looks fine23:01
@sonney2kgsomix, wassup?23:01
gsomixsonney2k, PR is updated. btw, what are problems with refcount for SGString?23:02
gsomixsonney2k, maybe I can help with it?23:03
@HeikoSsonney2k, yes, will add unit tests that test done and then equals equfatsr all non-abstract classes, based on class list.h23:04
@sonney2kgsomix, we need to make SGString* to be derived from SGReferencedData23:04
@sonney2kthat will kill serialization (biggest issue) and lots of other stuff though23:05
@sonney2kHeikoS, that is a nice automagic way!23:05
van51sonney2k: I thought of the tokenizer in case someone wants to tokenize on some specific delimiters or wants to create n-grams23:05
van51so that it would still work23:05
@sonney2kHeikoS, indeed it is very cool to test basic serialization for each and every class23:06
@HeikoSsonney2k, also parallel computation stuff.23:06
@sonney2kHeikoS, not sure what you mean by that23:06
@HeikoSgotta go now, see you tomorrow23:07
@HeikoSI mean grid-search23:07
van51sonney2k: you are afraid of the overhead because it creates a new object every time,right?23:07
@sonney2kvan51, yeah it makes certainly sense to split up based on different criteria23:07
@HeikoScl one machines23:07
@sonney2kvan51, I am only afraid of the overhead that is all23:07
@sonney2kvan51, maybe you can assume that input is an SGString<char> and then you only ever return the next index instead of a whole string23:08
@sonney2kHeikoS, but these need sane parameters then23:09
gsomixsonney2k, but there is solution of this problems for SGVector, SGMatrix, etc, right? is it can be applied to SGString?23:09
@sonney2kgsomix, yes23:09
@sonney2kI am pretty sick of having SGString around though23:09
@sonney2kwoudl love to drop it23:09
@sonney2kand just use SGVector23:09
@sonney2k(which is exactly  the same thing)23:10
van51sonney2k: yea that could work23:10
@sonney2kbut with refcoutnting already23:10
@sonney2kvan51, problem is that you need lowlevel functions then23:10
@sonney2kvan51, that operate on char*, int start, int end23:11
@sonney2kvan51, and seeing the dotfeature class mess we probably should drop a few functions from that one23:11
van51sonney2k: yea I just got all the functions in the header and would sort them out later23:12
gsomixsonney2k, aha. are there big problems with avoiding SGString and using SGVector instead?23:12
van51sonney2k: so I'll make a basic parser for now23:12
van51sonney2k: and it can be optimized later?23:12
van51sonney2k: like the way you said, it would return indexes23:12
van51I know that gsomix worked on a LineReader class, could it be modified/extended to parse SGVectors/SGStrings ?23:13
@sonney2kgsomix, well the big problem is that SGString is currently not auto-dereferenced. so it is used as a pointer and it is not clear when /where it is freed.23:14
van51because I also make that assumption that a single document is also a single SGString/SGVector23:14
@sonney2kgsomix, so you would need to a) adjust all code using it b) fix all typemaps c) fix serialization23:14
@sonney2kgsomix, and c) is the most difficult. we have no idea yet how to do it23:14
gsomixok, it's clear now23:14
@sonney2kvan51, gsomix's line reader will return one sgstring/vector yes per line23:16
@sonney2kvan51, so yes one doc  per string23:16
van51sonney2k: yeah but I thought you could modify the delimiters?23:16
@sonney2kyes you could23:16
@sonney2kso for the streaming variant it could return sgstrings23:16
@sonney2kfor each n-gram or whatever23:17
van51sonney2k: btw, about the SGString, if it was just dropped like you said then that would only leave us with a) adjusting current code23:18
van51I don't know if it's used extensively or if it has a special use but I did see that it was pretty much the same to SGVector23:19
@sonney2kvan51, yes it is23:20
@sonney2kvan51, problem is only that we have SGStringList (which we should then rather call SGVectorList) that is  not properly supported23:21
@sonney2kit needs serialization etc too23:21
@sonney2kand then StringFeatures should be converted to SGVariableLengthFeatures23:23
gsomixcu guys, need to prepare more23:23
@sonney2kgsomix, good luck!23:23
gsomixvan51, btw, if there are some suggestions or questions for LineReader - welcome. :)23:23
gsomixsonney2k, tnx.23:23
van51gsomix: sure, if I can come up with something yous guys haven't already :)23:24
van51gsomix: gl for tomorrow !23:24
van51sonney2k: yea sure those changes would have to be introduced, but I think because the SGString and the SGVector use the same representation internally it wouldn't be that different23:26
@sonney2kvan51, true except for the missing refcounting now23:27
van51sonney2k: ofc I have never coded serialization etc so this is just an impression23:27
@sonney2kvan51, so you currently have to do sth. like free_*23:27
@sonney2kand later do nothing or x=SGVector() to free x23:28
lisitsynsonney2k: one example of pooling is pool of sgvectors23:29
lisitsynthey are constructed very often now23:29
van51ah so SGVector is also missing ref-counting?23:29
lisitsynwould be better to avoid any allocations23:29
@sonney2kvan51, no   it uses ref-counting23:29
-!- lambday [67157d36@gateway/web/cgi-irc/kiwiirc.com/ip.103.21.125.54] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]23:30
@sonney2klisitsyn, did you know how expensive this is?23:30
van51sonney2k: I misread then23:30
@sonney2knp23:31
lisitsynsonney2k: well it can't be optimized out I believe23:31
lisitsynso produces some slight overhead23:31
@sonney2kI am pretty sure  2d toy data would suffer but anything reasonably big should have close to 0 impact23:32
lisitsynsonney2k: suffer is a wrong word23:32
@sonney2klisitsyn, well to avoid copying we could use references of course23:34
lisitsynsonney2k: references of sgvector?23:34
lisitsyndangerous23:34
@sonney2kbut we should see some speed degradation before trying to do sth like that23:34
@sonney2klisitsyn, sure23:34
lisitsynsonney2k: no I would not like to do that23:34
lisitsynit breaks the meaning of refcounting here23:35
lisitsynwhat do I mean is to avoid multiple memory allocations23:35
@sonney2klisitsyn, like SGVector x = y; ?23:35
lisitsynlike < 40 bytes or so23:35
lisitsynsonney2k: no I mean we can hold a pool of ready sgvectors23:35
lisitsynso we just give it if one calls a constructor23:36
@sonney2kmake an example of what memory allocations you mean?23:36
lisitsynsonney2k: constructor23:36
@sonney2kvan51, I am still working on the sparse vector thing btw...23:38
@sonney2kalright23:38
@sonney2ktime for me to sleep23:39
van51sonney2k: hehe23:39
van51sonney2k: where do you thing you 're going? you have work to do..23:39
-!- gsomix [~gsomix@83.234.54.9] has quit [Ping timeout: 252 seconds]23:52
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]23:52
-!- gsomix [~gsomix@83.234.54.9] has joined #shogun23:58
--- Log closed Mon Jun 10 00:00:38 2013
« Saturday, 2013-06-08 Index Monday, 2013-06-10 »

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