Open in new window / Try shogun cloud
--- Log opened Sun Mar 23 00:00:50 2014
-!- khalednasr [~k.nasr92@41.69.203.191] has quit [Quit: Leaving]00:15
-!- newlido [~quassel@41.237.151.196] has quit [Ping timeout: 240 seconds]01:08
-!- kislay [~kislay@210.212.61.185] has quit [Ping timeout: 240 seconds]01:54
-!- kislay [~kislay@14.139.41.150] has joined #shogun02:00
-!- thoralf [~thoralf@91-65-136-194-dynip.superkabel.de] has quit [Quit: Konversation terminated!]02:02
-!- kislay [~kislay@14.139.41.150] has quit [Ping timeout: 268 seconds]02:20
-!- PirosB3 [~pirosb3@cpc4-newt31-2-0-cust631.newt.cable.virginm.net] has quit [Quit: PirosB3]02:41
-!- nikolisgal [~nikolisga@ppp141237211150.access.hol.gr] has quit [Remote host closed the connection]02:41
-!- mode/#shogun [+o sonney2k] by ChanServ03:23
shogun-buildbotbuild #755 of nightly_default is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/75503:42
-!- vperic [~vperic@194.228.13.178] has quit [Ping timeout: 246 seconds]04:56
-!- DarkRoom_ [~darkroom@14.139.82.6] has quit [Read error: Operation timed out]04:59
-!- kislay [~kislay@14.139.41.150] has joined #shogun05:54
-!- newlido [~quassel@41.237.151.196] has joined #shogun06:59
-!- newlido [~quassel@41.237.151.196] has quit [Remote host closed the connection]07:00
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun08:44
-!- mode/#shogun [+o iglesiasg] by ChanServ08:44
-!- jiaolong_ [553a5634@gateway/web/freenode/ip.85.58.86.52] has quit [Quit: Page closed]09:15
-!- kislay [~kislay@14.139.41.150] has quit [Read error: Connection reset by peer]10:02
-!- kislay [~kislay@14.139.41.150] has joined #shogun10:05
-!- andrew [~andrew@c-50-184-110-141.hsd1.ca.comcast.net] has joined #shogun10:10
-!- nikolisgal [~nikolisga@ppp141237211150.access.hol.gr] has joined #shogun10:15
-!- andrew [~andrew@c-50-184-110-141.hsd1.ca.comcast.net] has quit [Ping timeout: 240 seconds]11:15
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Quit: Leaving]11:55
-!- Netsplit *.net <-> *.split quits: pfm_, @wiking12:30
-!- Netsplit over, joins: pfm_12:41
-!- PirosB3 [~pirosb3@cpc4-newt31-2-0-cust631.newt.cable.virginm.net] has joined #shogun12:45
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun12:46
-!- ServerMode/#shogun [+o wiking] by kornbluth.freenode.net12:46
-!- vperic [~vperic@194.228.13.178] has joined #shogun13:07
-!- nikolisgal [~nikolisga@ppp141237211150.access.hol.gr] has quit [Quit: Leaving]13:11
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun13:25
-!- mode/#shogun [+o iglesiasg] by ChanServ13:26
-!- HeikoS [~heiko@0546d087.skybroadband.com] has joined #shogun14:14
HeikoS.14:15
-!- PirosB3 [~pirosb3@cpc4-newt31-2-0-cust631.newt.cable.virginm.net] has quit [Quit: PirosB3]14:32
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Quit: Leaving]14:43
-!- pl8787 [9fe22b60@gateway/web/freenode/ip.159.226.43.96] has joined #shogun14:43
pl8787heikoS : Is there any io class in shogun can read string feature ?14:44
HeikoSpl8787: in the sense of?14:44
pl8787the format such as : 1|name|cate|0|014:45
pl8787I use csv reader, but only numerate feature can read14:47
HeikoSI see14:47
HeikoSpl8787: I dont know, you should ask sonney2k, andor lisitsyn, wiking14:48
lisitsynI don't speak english sorry14:48
lisitsyn:D14:48
HeikoSif its currently not possible, it might be agood idea to extend the existing class to be able to read categorical andor string data14:48
HeikoSlisitsyn: ?14:48
lisitsynjk14:48
-!- bumblebee [~quassel@202.78.175.199] has joined #shogun14:49
lisitsynI don't know actually14:50
pl8787lisitsyn: how to solve this problem?14:50
lisitsynlet me go through the code14:50
pl8787lisitsy: ok thanks14:50
lisitsynshould be possible14:51
lisitsynah no14:53
lisitsynSG_NOTIMPLEMENTED14:53
lisitsyn:D14:53
pl8787ok, maybe I could implement myself, or use other language(python) preprocess the data14:55
-!- vperic [~vperic@194.228.13.178] has quit [Quit: Konversation terminated!]14:55
-!- HeikoS [~heiko@0546d087.skybroadband.com] has quit [Ping timeout: 264 seconds]14:57
-!- HeikoS [~heiko@ra-vpn-212.ra-vpn.ucl.ac.uk] has joined #shogun15:14
-!- HeikoS [~heiko@ra-vpn-212.ra-vpn.ucl.ac.uk] has quit [Read error: Connection reset by peer]15:25
-!- nikolisgal [~nikolisga@ppp141237211150.access.hol.gr] has joined #shogun15:28
-!- HeikoS [~heiko@0546d087.skybroadband.com] has joined #shogun15:44
-!- HeikoS [~heiko@0546d087.skybroadband.com] has quit [Ping timeout: 264 seconds]15:52
-!- kprah [67157f4c@gateway/web/freenode/ip.103.21.127.76] has joined #shogun17:04
kprahwiking: hey, you there?17:04
kprahsonney2k: I had a question on present bmrm solvers17:06
kprahif we already have a bmrm solver (libbmrm.cpp/h) , then what's the reason of having an proximal bundle solver (libp3bm.cpp)?17:07
kprah\msg17:08
-!- kprah_ [67157f4c@gateway/web/freenode/ip.103.21.127.76] has joined #shogun17:43
-!- kprah [67157f4c@gateway/web/freenode/ip.103.21.127.76] has quit [Ping timeout: 245 seconds]17:45
-!- kprah [67157d4c@gateway/web/freenode/ip.103.21.125.76] has joined #shogun17:51
-!- kprah_ [67157f4c@gateway/web/freenode/ip.103.21.127.76] has quit [Ping timeout: 245 seconds]17:55
-!- kprah [67157d4c@gateway/web/freenode/ip.103.21.125.76] has quit [Quit: Page closed]17:59
-!- nikolisgal [~nikolisga@ppp141237211150.access.hol.gr] has quit [Quit: Leaving]18:17
-!- kprah [67157f4c@gateway/web/freenode/ip.103.21.127.76] has joined #shogun18:27
-!- kprah [67157f4c@gateway/web/freenode/ip.103.21.127.76] has quit [Quit: Page closed]18:39
-!- bumblebee [~quassel@202.78.175.199] has quit [Read error: Operation timed out]18:51
-!- bumblebee [~quassel@202.78.175.199] has joined #shogun18:51
-!- jiaolong [9e6d09e4@gateway/web/freenode/ip.158.109.9.228] has quit [Quit: Page closed]18:59
-!- bumblebee_ [~quassel@202.78.175.199] has joined #shogun19:04
-!- bumblebee [~quassel@202.78.175.199] has quit [Ping timeout: 240 seconds]19:04
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun19:19
-!- mode/#shogun [+o iglesiasg] by ChanServ19:19
-!- sonne|osx [~sonne@e178251106.adsl.alicedsl.de] has joined #shogun19:37
-!- besser82 [quassel@fedora/besser82] has quit [Remote host closed the connection]19:37
-!- dhruv13j [dhruv13j@gateway/shell/bnc4free/x-argwiipfafaabqkc] has quit [Remote host closed the connection]19:37
-!- sonne|osx [~sonne@e178251106.adsl.alicedsl.de] has quit [Client Quit]19:39
-!- sonne|osx [~sonne@e178251106.adsl.alicedsl.de] has joined #shogun19:41
-!- bumblebee_ [~quassel@202.78.175.199] has quit [Ping timeout: 252 seconds]19:45
-!- HeikoS [~heiko@0546d087.skybroadband.com] has joined #shogun19:58
-!- besser82 [quassel@fedora/besser82] has joined #shogun20:00
-!- mode/#shogun [+o besser82] by ChanServ20:00
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun20:02
shogun-notifier-shogun: pl8787 :develop * 3afb4cf / tests/unit/statistics/HSIC_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/3afb4cf1bddf5c55d9cc9a3f2b4f1e79f551941620:02
shogun-notifier-shogun: Update BSD license at the top of the file.20:02
shogun-notifier-shogun: Heiko Strathmann :develop * adc9500 / tests/unit/statistics/HSIC_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/adc950001671e2af852629828e80708bcef8a43e20:02
shogun-notifier-shogun: Merge pull request #2066 from pl8787/hsic20:02
shogun-notifier-shogun:20:02
shogun-notifier-shogun: #1983 Update BSD license at the top of the file.20:02
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Quit: Leaving]20:26
-!- dhruv13j [dhruv13j@gateway/shell/bnc4free/x-dojotckqqmfmtwvw] has joined #shogun20:30
-!- PirosB3 [~pirosb3@cpc4-newt31-2-0-cust631.newt.cable.virginm.net] has joined #shogun20:31
kislayHeikoS, hey.20:42
HeikoSkislay: hey!20:42
kislayHeikoS, please tell me some pointers to get started with the opencv integration issue.20:43
lisitsynHeikoS: !20:49
lisitsynhttps://github.com/lisitsyn/aer/blob/master/test_lib_src/library.cpp20:49
HeikoSkislay: I dont know this very well20:49
lisitsynsee last three lines20:49
lisitsynfour20:49
HeikoSkislay: best is to write a mail to kevin about this, or just explore on your own a bit20:49
HeikoSlisitsyn: aer?20:50
lisitsynHeikoS: don't tell me it is the first time you see it20:50
lisitsyn:D20:50
-!- travis-ci [~travis-ci@ec2-54-81-190-108.compute-1.amazonaws.com] has joined #shogun20:50
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/2137720720:50
-!- travis-ci [~travis-ci@ec2-54-81-190-108.compute-1.amazonaws.com] has left #shogun []20:50
HeikoSlisitsyn: I am sorry, so many things going on :)20:50
lisitsynHeikoS: yes that's what I develop to replace all the stuff20:51
lisitsynwell20:51
HeikoSwhat do you want to tell me with this gist?20:51
lisitsynthat's how you define plugin20:51
HeikoSis this how to import modules?20:51
lisitsynrather export20:52
HeikoSI see20:52
kislayHeikoS, okay.20:52
HeikoSI think thats sweet!20:52
lisitsynHeikoS: what I want is some feedback because I am developing it mostly for shogun20:52
HeikoSok I see20:52
lisitsynHeikoS: so basically you'd have to say20:52
lisitsynI want to export class blabla20:53
-!- zxtx [~zv@cpe-98-148-41-3.socal.res.rr.com] has joined #shogun20:53
HeikoSlisitsyn: ok, so what about multiple classes?20:53
lisitsynwhich is inherited from one of our base classes20:53
HeikoSlisitsyn: and how are modules organised?20:53
HeikoSlisitsyn: do we keep the current folder structure?20:53
lisitsynI don't know20:53
lisitsynI don't force anything on that20:53
HeikoSok,20:53
lisitsynwhat I have now is library/metaclass/object infrastructure20:53
HeikoSI mean for single classes this is perfect20:54
HeikoSlisitsyn: okay, and everyone is too busy to give feedback? :D20:54
lisitsynsorry don't get it20:54
lisitsynHeikoS: no I only ask you :D20:54
lisitsynbecause you are some kind of fan of 'proper' things20:54
HeikoSlisitsyn: can a class be registered in multiple libs?20:54
lisitsynone class in multiple libs??20:54
HeikoSlisitsyn: haha, yeah but unable to do that myfelf :D20:54
HeikoSmyself20:54
lisitsynwhat's the usecase20:55
HeikoSlisitsyn: so it would be cool if we could register the classes like within multiple modules20:55
HeikoSlisitsyn: say, GP module20:55
HeikoSneeds GaussianKernel20:55
HeikoSbut LibSVM module also needs20:55
lisitsynyou're speaking about dependencies?20:55
HeikoSlisitsyn: yeah20:55
lisitsynwell that's to be done yet20:55
HeikoSlisitsyn: so it would be cool in the end if I just said: I only want GPs20:55
HeikoSor only LibSVM20:56
lisitsynthat's actually a great problem20:56
lisitsynHeikoS: a lot of things are going to be done with strings20:56
lisitsynthat's a drawback20:57
HeikoSmmmh20:57
HeikoSok20:57
lisitsynHeikoS: if you need gaussian kernel20:57
lisitsynyou probably would ask some system20:57
lisitsynto spawn "GaussianKernel" for you20:57
lisitsynit would be As<Kernel> for you though20:58
lisitsynHeikoS: As is my shared_ptr ;)20:58
lisitsynI found it a good naming but I can get wrong20:58
HeikoSehm, I am using track there20:58
HeikoSwhat?20:58
lisitsynHeikoS: remember we need shared_ptrs?20:58
HeikoSyeah20:59
lisitsynbasically I am hiding shared_ptr20:59
lisitsynI called it As20:59
HeikoSgive somm overview of those again? :)20:59
lisitsynso As<Kernel>20:59
lisitsynis a smart pointer for kernel20:59
lisitsynyou can use it as kernel whatever way you want20:59
lisitsynHeikoS: overview of shared_ptr?21:00
lisitsynit is basically what we did with all these ref*21:01
HeikoSyeah and why we want them21:01
lisitsynHeikoS: it is in C++1121:01
lisitsynit is just the same thing as SGRefObject or whatever21:01
lisitsynrefcount21:02
lisitsynthat's all DIY things that are already in std21:02
lisitsynHeikoS: so basically why to have probably buggy implementation of something from the standard library ;)21:04
HeikoSok21:04
HeikoSI agree21:04
HeikoSbut thats different to the modules framwork no?21:04
lisitsynno not really21:04
lisitsynthey are coming together unfortunately :)21:05
-!- bumblebee [~quassel@202.78.172.162] has joined #shogun21:05
lisitsynsay you want to spawn something21:05
lisitsynsome kernel may be21:05
lisitsynwhat you get is either raw pointer21:06
lisitsynor some smart ptr already21:06
lisitsynsee what I mean?21:07
lisitsynHeikoS: well and serialization problem can be solved here as well21:09
lisitsynif you don't mind :)21:09
lisitsynHeikoS: basically I store everything in some type-erasure type (any)21:09
HeikoSthats cool21:09
lisitsynHeikoS: e.g. there is a type21:10
lisitsynthat allows21:10
lisitsynAny x(3.0);21:10
HeikoSlisitsyn: I have problems keeping the overview on this. Could you maybe write a little manifest on what is done how and link to the gists?21:10
lisitsynHeikoS: uhm yeah but it is still work in progress21:10
lisitsynI am just keeping you updated ;)21:10
lisitsynHeikoS: ah yeah and I have something to make dptrs easy21:13
lisitsynok 'easier'21:13
lisitsynthat's obvious but still21:13
lisitsynthere is a class DPtr21:13
lisitsynso what you do is21:13
lisitsynprivate:21:14
lisitsynclass Self;21:14
lisitsynDPtr<Self> self;21:14
lisitsynHeikoS: then you define this private class and use it in .cpp21:14
lisitsyndptr's default constructor creates an instance for you so you don't care about that at all21:14
lisitsynokay I already overloaded you ;)21:15
HeikoSlisitsyn: sorry21:17
HeikoSlisitsyn: what about a motivation document or something?21:18
HeikoSwhere you explain why we want to do this, how you want to do this, and give examples, etc21:18
lisitsynas I said I want to work out some issues still21:18
HeikoSI think that would greatly simplify the discussion21:18
HeikoSlisitsyn: if we structured that a bit21:18
HeikoSeasier to identify issues21:18
lisitsynok will do later once I resolve some things21:18
HeikoSlisitsyn: sorry about that, but its too hard to understand all this from chat21:24
lisitsynhah okie21:25
HeikoSlisitsyn: I also think this will make it easier for the others to participate. It might be that noone really understands what you are doing with that currently ;)21:25
lisitsynHeikoS: that's exactly true21:25
lisitsynI am pretty lonely here :D21:25
HeikoSlisitsyn: hehe, well then do this, will make it much easier. Also, then people will know about you doing this21:30
HeikoSlisitsyn: and since its hard to convince anyone in shogun to be excited about framework changes, this is also necessary to get manpower ;)21:31
HeikoSlisitsyn: I bet thoralf will be excited21:31
lisitsynheh okay21:33
lisitsynHeikoS: have you known about C++'s name mangling?21:36
HeikoSlisitsyn: no whats that?21:37
lisitsynHeikoS: e.g. if you have a function21:37
lisitsynint f(int x)21:37
lisitsynyou can't just reference it with 'f'21:37
lisitsynif you are using C++21:37
lisitsynit would be exported as21:37
lisitsynintfdsfdsffPSdintxdfdsfsd21:37
lisitsynI mean it's name is pretty much ruined21:38
HeikoShuh?21:38
lisitsynHeikoS: have you seen 'extern "C"' in code?21:38
HeikoSwhere ?21:38
HeikoSyeah21:38
lisitsynwell in some libraries may be21:38
lisitsynthat's to tell compiler21:38
lisitsynthat you don't want to mangle it21:38
lisitsynit uses C rules of externalization21:38
HeikoSI see21:38
HeikoSah weird21:39
lisitsynhaha21:39
lisitsynHeikoS: funny thing is that type mangling is not in the standard of C++21:39
lisitsynif you have compiler X you can do that any way you want21:39
HeikoSlisitsyn: yeaaaah, read and responded to *all* shogun emails finally ;)21:39
lisitsynthey follow same rules nowadays21:39
lisitsynbut still you can't be guaranteed21:39
lisitsynHeikoS: heh I don't have that much emails thanks to deity21:40
HeikoSlisitsyn: I will do some cooking now21:40
lisitsynokay have fun21:40
HeikoSlisitsyn: with vincent btw. And then I will kick him to write a schedule ;)21:40
HeikoSlisitsyn: looking forward for your description of the module stuff21:40
HeikoSsee you21:40
lisitsynheh okay hello to vincent21:40
lisitsyndon't expect it earlier than in a week :D21:40
-!- soumyaC_ [uid15286@gateway/web/irccloud.com/x-ywfpriynusavedzn] has joined #shogun21:41
dhruv13jlisitsyn: hey! just curious, why is the name mangling thing done?21:41
lisitsyndhruv13j: because of C++ things like const and etc21:42
lisitsyne.g.21:42
lisitsynf() const21:42
lisitsynand21:42
lisitsynf()21:42
lisitsynthey are basically different functions21:42
lisitsynand you can have member functions21:42
lisitsynfriend functions, templates21:42
lisitsynnamespaces21:42
dhruv13jso, to differentiate between these?21:43
lisitsynyeah21:43
lisitsynyou don't have anything like that in C21:43
dhruv13jis it possible to overload on the basis of constness?21:43
dhruv13jlike f()21:43
dhruv13jand f() const21:43
lisitsynthey are different functions21:43
dhruv13jhmm...21:43
lisitsynyou can have totally different implementations of these21:43
dhruv13jand these are compiler dependent?21:45
lisitsynmangling or?21:45
dhruv13jlike clang's way is different from g++'s way to magle?21:45
-!- HeikoS [~heiko@0546d087.skybroadband.com] has quit [Ping timeout: 268 seconds]21:45
lisitsynclang and g++ probably do it the same way now21:45
dhruv13jif it was different, would it be a problem while loading modules etc?21:46
lisitsynbut older compilers don't21:46
lisitsyndhruv13j: that's why if you do anything with modules21:46
lisitsynyou need to do 'extern "C"'21:46
dhruv13jhmm... cool21:46
lisitsynyou can't do extern "C" on templates21:47
lisitsynor things like that21:47
dhruv13jdoesn't that limit designing modules?21:48
lisitsynwell that limits you to do some bottleneck21:48
lisitsynlike hide everything into one function21:49
lisitsynwhich returns all the data you need as an object21:49
dhruv13ji don't understand... so templates are out if you design modules?21:50
lisitsyndhruv13j: yeah actually yes21:51
lisitsynyou can't export templates21:51
lisitsyneven with C++21:51
lisitsynyou can only export specialized templates21:51
dhruv13junless you can assure all modules are compiled with the same compiler?21:53
lisitsyndhruv13j: no even with such an assumptin21:54
lisitsynI mean templates are compile-time21:54
dhruv13ji mean don't use extern "C"21:54
lisitsynso you always have to put them to headers21:55
dhruv13jokay...21:57
lisitsynso if you depend on headers of some module21:58
lisitsynit becomes not a module for you but the part of the library21:59
dhruv13jdoesn't this make it tough to use factory classes?22:05
lisitsyndhruv13j: well a bit22:05
lisitsynyou usually don't export classes22:05
lisitsynbut some functions that can create objects22:05
dhruv13jlike global functions, which are friends of classes?22:06
lisitsynextern "C" X* createX() { return new X(); }22:06
lisitsynthat's ok ^22:06
dhruv13jokay...22:07
dhruv13jfriend functions will only be needed if some other generic behaviour is required22:08
-!- thoralf [~thoralf@91-66-33-156-dynip.superkabel.de] has joined #shogun22:08
thoralfHey.22:08
lisitsynhey22:08
dhruv13jthoralf: hello!22:09
thoralfdhruv13j, lisitsyn: Hey hackers.22:09
lisitsyndhruv13j: _ZN9__gnu_cxx13new_allocatorIN3aer13LibraryHandleEE9constructIS2_IRKSsEEEvPT_DpOT0_22:09
lisitsyndhruv13j: ^ example of mangled name22:09
dhruv13jlisitsyn: http://pastebin.instantbird.com/64255022:11
dhruv13jlisitsyn: The compiler always seems to choose the const function...22:12
lisitsynhow did you check that?22:12
lisitsynI mean you always get 'x'22:12
dhruv13jthe cout22:12
dhruv13jyeah22:12
dhruv13jno the non-const returns x++22:13
lisitsynx++ or ++x?22:13
lisitsyn:D22:13
dhruv13jlol22:13
lisitsynit should use non-const function22:13
dhruv13jyeah22:13
lisitsynit probably uses22:13
dhruv13jnon-const22:13
dhruv13jlol22:13
lisitsynbut if you make a function22:14
lisitsynthat takes const reference22:14
lisitsynit should call the const thing22:14
dhruv13jin the this argument22:14
dhruv13jgot it!22:14
lisitsynyou can also use22:14
lisitsynconst &22:14
lisitsynand const &&22:14
lisitsynfunctions22:14
lisitsynbut that's like dangerous zone :D22:14
dhruv13jyeah...22:15
dhruv13jlisitsyn: thanks!22:19
-!- HeikoS [~heiko@0546d087.skybroadband.com] has joined #shogun22:20
thoralfMr. S22:20
HeikoSthoralf: Mr. K22:27
-!- PirosB3 [~pirosb3@cpc4-newt31-2-0-cust631.newt.cable.virginm.net] has quit [Quit: PirosB3]22:28
-!- sonne|osx [~sonne@e178251106.adsl.alicedsl.de] has quit [Quit: sonne|osx]22:36
-!- HeikoS [~heiko@0546d087.skybroadband.com] has quit [Ping timeout: 240 seconds]22:40
-!- khalednasr [~k.nasr92@41.69.156.170] has joined #shogun22:46
lisitsynkhalednasr: hey22:47
khalednasrlisitsyn, hey22:47
lisitsynkhalednasr: sorry I wasn't following your PR lately22:47
lisitsynreviewing it now22:47
-!- PirosB3 [~pirosb3@cpc4-newt31-2-0-cust631.newt.cable.virginm.net] has joined #shogun22:48
khalednasrlisitsyn, take your time :)22:49
lisitsynkhalednasr: get_layer thing is pretty dangerous22:51
thoralflisitsyn: WOW, I just benchmarked my import cleanups.  45m34s before, 39m56s after.22:51
thoralfSaving of over 10%. :)22:51
lisitsynwhat's that?22:51
-!- PirosB3 [~pirosb3@cpc4-newt31-2-0-cust631.newt.cable.virginm.net] has quit [Ping timeout: 240 seconds]22:53
thoralfI don't understand.22:53
lisitsynkhalednasr: okay I can tolerate that :D I think that's ok now22:53
khalednasrlisitsyn, any alternatives for the get_layer thing?22:53
lisitsynkhalednasr: well unique_ptr may be22:53
lisitsynwith custom deleter22:53
lisitsynthat just SG_UNREFs it22:54
khalednasrlisitsyn, seems too much for such a simple function22:56
lisitsynkhalednasr: too much overhead you mean?22:57
khalednasrlisitsyn, yeah22:57
lisitsynkhalednasr: I'll merge your PR once it passes travis build22:58
lisitsynit is in progress22:58
khalednasrlisitsyn, there are a few bugs that i'm still working on22:58
lisitsynkhalednasr: oh really? :)22:58
lisitsynI wanted to merge it already :D22:58
lisitsynwhat are the bugs?22:59
khalednasrlisitsyn, next commit will be better :)22:59
khalednasrlisitsyn,  one of the unit tests is failing  on travis for some reason22:59
lisitsynalright22:59
lisitsynkhalednasr: check my comment for your proposal23:01
khalednasrlisitsyn, alright, thanks23:02
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]23:02
-!- jiaolong [553a5634@gateway/web/freenode/ip.85.58.86.52] has joined #shogun23:07
thoralfHey jiaolong23:13
lisitsynkhalednasr: and also23:16
lisitsynkhalednasr: could you please be more active communication-wise?23:16
lisitsynkhalednasr: theo simply doesn't know you yet I guess :)23:17
lisitsynyour PR thing is pretty huge amount of job but we shouldn't lose that point23:17
khalednasrlisitsyn, I'll try :)23:19
-!- travis-ci [~travis-ci@ec2-54-80-179-187.compute-1.amazonaws.com] has joined #shogun23:19
travis-ci[travis-ci] it's khalednasr's turn to pay the next round of drinks for the massacre he caused in khalednasr/shogun: http://travis-ci.org/khalednasr/shogun/builds/2138424723:19
-!- travis-ci [~travis-ci@ec2-54-80-179-187.compute-1.amazonaws.com] has left #shogun []23:19
lisitsyncool23:20
-!- HeikoS [~heiko@0546d087.skybroadband.com] has joined #shogun23:20
khalednasrlisitsyn, TBH I'm not very good at communication, any advice? :)23:20
lisitsynkhalednasr: well just write letters ;)23:21
lisitsynkhalednasr: did you already presented yourself to theo?23:21
khalednasrlisitsyn, just the proposal23:21
lisitsynoh23:21
lisitsynthen please write a mail23:22
lisitsynexplain something about yourself and let him know what you would like to work on23:22
lisitsynkhalednasr: that's really important as you'd have to negotiate project details23:25
khalednasrlisitsyn, ok, thanks23:26
jiaolongthoralf: Hi23:28
thoralfjiaolong: Just sent you a small shell script: https://gist.github.com/tklein23/973073723:29
jiaolongthoralf: Thank you so much!23:31
thoralfjiaolong: Everything alright so far?23:31
jiaolongthoralf: I didnn't do unit test today. Sorry.23:31
jiaolongthoralf: The Travis error is from a unit test and I am fixing it.23:31
jiaolongthoralf: -1, +1 vs 0, +1 issue23:31
jiaolongthoralf: yes , I think no big problems.23:32
thoralfjiaolong: No problem, happens from time to time. :)23:32
jiaolongthoralf: OK, I will test it throughly23:33
jiaolongthoralf: I didn't know how to run the testing before23:34
jiaolongthoralf: I thought cmake -ENABLE_TESTING=on .. && make will do everything :(23:34
jiaolongthoralf: Regarding to the refactoring MulticlassMultipleOutputLabels, which  label type should we use? int16_t or index_t?23:36
thoralfjiaolong: In short: "make ARGS='--output-on-failure' -jXXX test"23:36
thoralfjiaolong: int32_t should be fine.  I'll fix it in MultilabelLabels.23:36
jiaolongthoralf: OK, cool!23:36
jiaolongthoralf: so the work is just replace MulticlassMultipleOutputLabels by MultilabelLabels?23:37
thoralfjiaolong: Yes.23:38
thoralfjiaolong: You may want to secure your changes by a few unit tests.23:39
thoralfDon't know what already exists.23:39
jiaolongthoralf: Yes, I will find a time to do it23:39
jiaolongthoralf: I checked the set_label, it seems MulticlassMultipleOutputLabels is quite different23:40
thoralfjiaolong: How different?  Will it be a problem?23:40
jiaolongthoralf: well, I just took a quick look, I need to read and understand the code better23:40
jiaolongthoralf: the real label index is not the same as the indx in set_labe(indx, vector)23:42
thoralfjiaolong: Then this will be part of the refactoring.23:43
jiaolongthoralf: is there any examples related to these two types of labels?23:43
jiaolongthoralf: OK23:43
thoralfjiaolong: I guess yes.  Do a grep for MulticlassMultipleOutputLabels over unit-tests or examples.23:43
jiaolongthoralf: OK, MultilabelLabels is not in the LabelsFactory23:44
jiaolongthoralf: why?23:44
thoralfjiaolong: Hmm.  I never needed it.  Add it if you need to.23:45
jiaolongthoralf: OK, just curious, not a big problem23:46
jiaolongthoralf: OK, thank you very much! I am checking the code. See you!23:46
thoralfBye23:47
-!- thoralf [~thoralf@91-66-33-156-dynip.superkabel.de] has quit [Quit: Konversation terminated!]23:56
--- Log closed Mon Mar 24 00:00:51 2014