IRC logs of #shogun for Monday, 2018-02-12

« Sunday, 2018-02-11 Index Tuesday, 2018-02-13 »
--- Log opened Mon Feb 12 00:00:58 2018
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 240 seconds]00:24
-!- Prashanth [uid170915@gateway/web/irccloud.com/x-irvwgopjkkmucmsq] has joined #shogun03:20
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4169 opened by naoa08:00
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun08:10
-!- Dhruvt97 [~remedy441@14.139.122.120] has left #shogun []08:11
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun08:12
-!- Dhruvt97 [~remedy441@14.139.122.120] has left #shogun []08:12
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun08:15
-!- Positron_ [~textual@73.93.142.231] has joined #shogun08:32
-!- Positron_ [~textual@73.93.142.231] has quit [Client Quit]08:33
@wikingsukey, flip08:54
@sukey(._.) ~ ? ???08:54
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/5ee18e1ef37de6a8cc24cb90c4c975d98dfb736c by vigsterkr08:55
@sukey[https://github.com/shogun-toolbox/shogun] New branch feature/golang created08:55
@wikingthere... go support for all :)08:55
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4170 opened by vigsterkr09:01
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4170 vigsterkr added label: "SWIG"09:01
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4170 vigsterkr added label: "meta examples"09:01
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has joined #shogun09:02
travis-ciit's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34037407209:02
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has left #shogun []09:02
-!- travis-ci [~travis-ci@ec2-54-234-205-100.compute-1.amazonaws.com] has joined #shogun09:03
travis-ciit's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34037407209:03
-!- travis-ci [~travis-ci@ec2-54-234-205-100.compute-1.amazonaws.com] has left #shogun []09:03
-!- Dhruvt97 [~remedy441@14.139.122.120] has quit [Read error: Connection reset by peer]09:04
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun09:06
-!- sheng_ [8c777906@gateway/web/freenode/ip.140.119.121.6] has joined #shogun09:09
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/03158aea6df4e40d9fba0758a815fe7c8cc1f69c by vigsterkr09:27
-!- Dhruvt97 [~remedy441@14.139.122.120] has quit [Read error: Connection reset by peer]09:59
-!- travis-ci [~travis-ci@ec2-54-234-205-100.compute-1.amazonaws.com] has joined #shogun10:17
travis-ciit's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34038148410:17
-!- travis-ci [~travis-ci@ec2-54-234-205-100.compute-1.amazonaws.com] has left #shogun []10:17
-!- Prashanth [uid170915@gateway/web/irccloud.com/x-irvwgopjkkmucmsq] has quit [Quit: Connection closed for inactivity]10:19
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has joined #shogun10:34
travis-ciit's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34038148410:34
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has left #shogun []10:34
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has joined #shogun10:48
-!- mode/#shogun [+o HeikoS] by ChanServ10:48
-!- king [5862c914@gateway/web/freenode/ip.88.98.201.20] has quit [Ping timeout: 260 seconds]10:55
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun11:04
-!- sheng_ [8c777906@gateway/web/freenode/ip.140.119.121.6] has quit [Ping timeout: 260 seconds]11:16
@HeikoSlisitsyn: jojo11:25
@HeikoSwiking:11:27
-!- Dhruvt97 [~remedy441@14.139.122.120] has quit [Read error: Connection reset by peer]11:38
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4164 karlnapf added label: "Cleanups"11:44
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4164 karlnapf added label: "BUG"11:44
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4164 karlnapf added label: "bugfixing"11:44
@wikingHeikoS, where r u? :D11:47
@HeikoSwiking: still uk11:47
@HeikoSflight is tonight11:47
@wikingah ok11:47
@wikingso when r u landing/coming11:47
@HeikoShaha11:47
@wiking?11:47
@wikingbtw you guys broke develop11:47
@wikingplz fix11:48
@HeikoSthe segfault?11:48
@wikinghttps://github.com/shogun-toolbox/shogun/commit/d1763b83fbc029bae1487201ef56977efdfe43a011:49
@wikingthis broke it11:49
@wikingbased on buildbot11:49
@wikingbtw have u seen ?:P11:49
@wikingi had a bored moment and didnt dare to start with the llvm project that needs for webassembly => golang support :P11:49
@HeikoSyep thats the one11:50
@HeikoSyeah man cool!11:51
@wikingtypemap for sgvect and matrix should work11:51
@wikingneed to do the meta example extension11:51
@HeikoSah yeah11:51
@HeikoSsaw11:51
@HeikoSthink adding new langs is easy11:51
@HeikoSeasy (tm)11:52
@wikingi'll now check on python typemaps11:52
@wikingfor sg*11:52
@wikingas there seems to be some shit11:52
@wikingwith python 3611:52
@wikingand actually will move to numpy 1.7+ api11:52
@wikingthat was released in 201311:52
@wikingany objections?11:52
@wikingi.e. stop using deprecated api of numpy11:53
@HeikoSah11:53
@HeikoSno11:53
@HeikoSall good11:53
@HeikoSthink the examples we have don't care11:53
@wiking?11:53
@HeikoSnot sure about notebooks/deprecated examples11:53
@wikingwhich examples?11:53
@HeikoSI mean they will still work11:53
@HeikoSmeta11:53
@wikingnon oit's all about11:53
@wikingtypemaps11:53
@wikingi think we can assume11:53
@wikingthat anybody who'd like to use shogun11:53
@wikingwould have a numpy older than 1.711:54
@wiking(2013)11:54
@HeikoSyes11:54
@wikingso all good then11:54
@HeikoSI mean python deps is the last thing we should keep backwards compatible :D11:54
@wikingif you could peep on that bug11:54
@wikingthat causes errors11:55
@wikingwould be great11:55
@wikingas that blocks pr CIs11:55
lisitsynHeikoS: hey11:55
@HeikoSlisitsyn: hey11:55
lisitsynhave you seen my messages from yesterday?11:55
@HeikoSyes I saw11:55
@HeikoSwanted to discuss11:55
lisitsynok so we have a problem11:55
lisitsyn:)11:55
@HeikoSlisitsyn: what do you think about actually enforcing the correctly registered type11:55
lisitsynI guess we have to11:56
@HeikoSrequires changing all the SG_ADD11:56
lisitsynyeah but it is regexp11:56
@HeikoSbut the other things are a bit nicer11:56
lisitsynshould be ok11:56
lisitsynI basically can't patch any to support that11:56
@HeikoSlisitsyn: ok will update the draft then11:56
lisitsyn:(11:56
@HeikoSand do it for a single class11:56
lisitsynI mean it is runtime vs compile-time11:56
lisitsynno marriage :)11:56
@HeikoSlisitsyn: yeah sure11:56
@HeikoSwe are exploring :D11:56
@wikingbtw11:57
@HeikoSlisitsyn: btw, your error msg segfaults11:57
@wikingcould i have a request for both of you plz11:57
lisitsynlean on steroids11:57
lisitsyn:D11:57
@HeikoShehe11:57
@wikingcan we stop merging bs/tests into develop11:57
@wiking?11:57
@wikingplz11:57
lisitsynHeikoS: segfaults??11:57
@wikingand now i'm not referring to the last breaking stuff11:57
@wikingbut in general11:57
@wikingchanges that you HeikoS did with the macro etc11:57
lisitsynuhmm we can but it will become develop-211:57
@wikingbecause if they keep changing11:57
@wikingthen any other development11:57
@wikingis super slowed down11:58
@wikingbecause of ccache11:58
@wikingcontamination11:58
lisitsynyeah your concern is understandable11:58
@wikingas we know SGObject.h fucks evertyhing11:58
@HeikoSsure11:58
@wikingso now even small changes in develop triggers full recompile11:58
@wikingof the library11:58
lisitsynwe just need to iterate, I don't mind if we iterate in a branch11:58
@wikingwithout ccache11:58
@wikinglisitsyn, yeah sure11:58
@wikingfeature branch is for that11:58
@wikingno? :)11:58
lisitsynthat's true but we were afraid of it becoming unmergeable11:59
@HeikoSme too11:59
-!- Prashanth [uid170915@gateway/web/irccloud.com/x-geaudwezzpnojtxi] has joined #shogun11:59
@wikingit wont happen11:59
@HeikoSbut for the current11:59
lisitsynthat's the only point11:59
@HeikoSit is ok11:59
@wikingwe are not pivoting so much11:59
@HeikoSnobody else touches this11:59
@wikingin develop lately11:59
lisitsynwe did a bit in early stages11:59
lisitsynbut now with gsoc students sure11:59
@wikinglisitsyn, yeah but it's only you HeikoS and maybe me11:59
lisitsynwe need to jump in to the branch11:59
@wikingi mean 3 devs can do shit in a featurebranch11:59
@wikingand keep rebasing imo12:00
@wikingi'm using only hedo for compilation12:00
@wikingas that's 'fast' when i dont have ccache as well12:00
@wikingon my own machine this is a killer12:00
@wikingand yes12:00
@wikingit'd be great that at least the base class would be registered (see the type in the class and not SGO)12:01
@wikingbut i've already raised this in a github comment12:01
lisitsynwiking: I think we have to merge that feature branch often12:01
@HeikoSwiking: I think we should do that yws12:01
@wikinglisitsyn, sure we can12:01
lisitsynI mean we can switch to that branch12:01
@wikinglisitsyn, but i mean if you want i can take care of that12:02
lisitsynbut we need to merge it like a few times12:02
@wikinglisitsyn, how to merge it or rebase it12:02
@wikingif this helps you in any ways12:02
@wikingi'm happy to take that responsibility12:02
@wikingi know that develop indeed should be for things12:02
@wikingliek this12:02
@wikingbut with this codebase12:02
@wikingit's really a killer if you keep touching SGO.h12:02
lisitsynyeah I know12:02
lisitsynthat's why we're in hurry to finish things12:02
PrashanthHello12:04
PrashanthI was trying to build the Random Forest Cookbook and I found this issue while running Random Forest example12:04
Prashanth[WARN] In file /home/prashanth/shogun/src/shogun/multiclass/tree/CARTree.cpp line 273: Feature types are not specified. All features are considered as continuous in training12:04
@HeikoSPrashanth: thats a quite informative error msg no?12:04
durovowiking: Hello,  it seems that CDynamicArray cannot store struct objects such as CStatistics::SigmoidParamters, I am getting this error: error: no matching function for call to 'shogun::Parameter::add_vector(shogun::CStatistics::SigmoidParamters**, int32_t*, const char [6], const char [26])'. Can you please suggest a way to store an array of SigmoidParamters such that they can be SG_ADDed?12:05
@wikingdurovo, hello12:05
@wikingheheh12:05
@wikingso people12:05
@wikingHeikoS, lisitsyn12:05
@wikingdurovo, i'll get back to you in a sec12:06
durovowiking: Okay12:06
@wikingHeikoS, lisitsyn so this is actually we should still talk about a bit12:06
@HeikoSPrashanth: you will need to set feature types12:06
@wikingbeing able to serialize/clone objs with std::vector/map/set12:06
@wiking:)12:06
@wikingi mean PLEASE12:06
@wikingthis of course the blocker atm is12:07
@wikingm_parameter->add()12:07
@wikingi mean if we switch to put(param, )12:07
@wikingand std::vector/map/set12:08
@wikingis not possible12:08
@wikingit's gonna be a bit of a WTF moment :)12:08
@HeikoSwiking: sure we want that12:08
@HeikoScan probably work that out with the tags12:08
@HeikoSneeds an extension for std:: maybe12:09
@wikingyeah12:09
@wikingthat's what i mean12:09
@wikingthat we need that12:09
@HeikoSwiking: but I fear there will be lots of special cases coming with that12:09
@wikingyeah12:10
@wikingi wanna say i dont care12:11
@wiking:)12:11
@wikingit's just that we need that12:11
@wikingreally12:11
@HeikoSwiking: lisitsyn: if we register objects by their base class (like CKernel)12:11
@wikingwe wanna get rid of DynArray12:11
@wiking:)12:11
@HeikoSthen how does the swig method look like?12:11
lisitsynwe can switch to std:: once we stop using fields12:12
@wikingthere's typemap for std::vector/map12:12
@wiking:)12:12
@wikingindeed we already rely on std::vector typemaps12:12
lisitsynI am not sure it is very soon12:12
@wikingnote i didn't say now12:13
@wikingi just said that it'd be great to have it on the roadmap12:13
@wikingas that's really something that would make a lot of things easier12:13
@wikingand we could get rid of a lot of custom hacks12:13
@wikingbtw when's the announcment?12:24
lisitsyntoday12:24
@wikingyeah that i know12:24
@wikingbut within that ? :)12:24
lisitsynI guess evening for us12:25
@wikingtypical 1900 utc?12:25
@wikingdurovo, so here i am12:31
@wikinglemme look into it12:31
durovowiking: ok12:32
@wikingdurovo, i'm not so sure if i understand why CDynamicArray wouldn't be able to store any type?12:32
@wikingCDynamicArray is a wrapper around DynArray12:32
@wikingwhere DynArray has T* array;12:32
durovoCDynamicArray seems to work for primitives like index_t, float64_t etc12:34
durovodoesn't seem to work for a struct12:34
durovomy terminal is full of errors like shogun/src/shogun/base/Parameter.h:551:7: note:   no known conversion for argument 1 from 'shogun::CStatistics::SigmoidParamters**' to 'int8_t** {aka signed char**}'12:35
@wikingdurovo, ? wait... how do you SG_ADD the CDynamicArray?12:36
@wikingcoz that error seems to be from something of SG_ADD12:36
@wikingnot CDynamicArray<WhateverClass>12:36
durovoThis is without SG_ADD, here is the full error <https://pastebin.com/2rCCajf7>12:39
@wikingit's impossible that you get12:40
@wiking  m_parameters->add_vector(&m_array.array,12:40
@wikingwithout SGADD12:40
@wikingoh wait12:41
@wikingm_parameters->add_vector(&m_array.array,12:41
@wiking&m_array.current_num_elements, "array",12:41
@wiking"Memory for dynamic array.");12:41
@wiking:)))))))))))))))))))))))12:41
@wikingpure beauty]12:41
@wiking:)12:41
@wikingdurovo, mea culpa... for making you work with this interface12:42
@wiking:)))))12:42
durovo:smile:12:43
lisitsynwiking: it seems we have to have only put(string, CSGObject*)12:44
lisitsynand do *manual* dispatching of types12:44
lisitsyn:D12:44
@wiking:>>>>>>>>>>>>>>>>>>>>>>>12:45
@wikingnooooooooooooo12:45
@wikingda PAin12:45
lisitsynthe thing is that12:45
lisitsynwe don't need type-safety here12:45
lisitsynbecause you put it by string anyways12:45
@wikinglisitsyn, yeah i mean12:45
lisitsynit is always runtime12:45
lisitsynwiking: but this approach should work in all interfaces without troubles12:45
@wikingit'd would be nice to have this level of checking as well12:46
@wikingbut i mean12:46
@wikingif no then no12:46
@wikingbut need a proper checking runtime12:46
@wikingright? :)12:46
@wikingas atm checking for SGO* is like12:46
@wikingyeah man it's gonna be SGO as eveyrthing that is not primitive is SGO*12:46
@wiking:P12:46
lisitsynwiking: yeah we'd have to implement some fast way to dispatch12:47
lisitsynnot a big deal but it is a hack to make it work in all interfaces12:47
@wiking:<12:47
lisitsynwiking: so it is like if object is kernel then put it as kernel12:47
lisitsynwiking: it is actually fast using map with typeid12:48
lisitsynit is hashable12:48
@wiking:D12:48
@wikingtypeindex12:48
lisitsynwiking: yes12:48
@wikingftw12:48
lisitsynwe don't care about performance here because it is done just a few times12:48
@wikingman numpy docs are uselesss12:48
@wikinglisitsyn, well12:48
@wikingyes and no12:48
@wiking:)12:48
@wikingi mean there was a constant concern12:48
lisitsynyou never set a kernel millions of times12:48
@wikingabout how to update12:48
@wikingbut anyhow we settled12:48
@wikingwith using temp vars12:49
lisitsynwiking: I'd expect we have much more terrible bottlenecks :)12:49
@wikingyeye i mean it's even less in case of SGOderived12:49
@wikinglisitsyn, as always12:49
@wiking;)12:49
lisitsynmuuch more12:49
@wikingi mean there's always bigger shit12:49
@wiking:))))12:49
@wikingbut thats' not an excuse12:49
@wikingto add shit12:49
@wiking:P12:49
@wikingthis is not in reference what you suggested12:49
@wikingjust in general12:50
lisitsynwell that's the only solution we have now12:50
@wikingyeah12:50
@wikingi gotcha12:50
@wikingdo what you have to do. :)12:50
@wikingi'm digging numpy array shitz12:50
@wikingmagical ansi c api12:50
@wiking#define NPY_NO_DEPRECATED_API NPY_1_7_API_VERSION12:50
@wikingftw12:50
lisitsynoh they have some things here12:50
@wikingbtw12:52
@wikingshouldn't we start using alignments12:52
@wikingfor memory/ :)12:52
@wikingwhile we are at talking about perfomance12:53
@wiking:P12:53
@wikingand that we OVERRIDE every possible malloc in this lib12:53
lisitsynwiking: in algorithms maybe12:53
@wikingyes12:53
@wikingi mean i'm just following up12:53
@wikingthat how much shit we have12:53
lisitsynwiking: we can also use stack-allocated vectors and shit12:53
@wikingye12:53
@wikingmost of the time12:53
@wikingthat should be possible12:53
lisitsynlinear model w can usually be on stack12:53
@wikingindeed12:54
@wikingthe problem with that12:54
@wikingmmmm12:54
@wikingwanted to say linalg::12:54
@wikingbut we can fake it right12:54
@wiking? :)12:54
lisitsynwiking: we can implement vector that switches storage depending on it's size12:54
@wikinglisitsyn, i more meant like12:55
@wikingSGVector(&stacker, size, false)12:55
@wiking:D12:55
lisitsynwiking: kind of12:55
@wikingbut yeah12:55
@wikingwe can do that later12:55
@wikingbut yeah12:55
@wikingour memory shitz12:55
lisitsynwiking: I believe we need to split vectors used in algorithms and vectors used for storage/interfaces12:55
@wikingfor vec/matrix12:55
@wikingis not the nicesest12:55
@wikinglisitsyn, mmmm probablyu12:56
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun13:00
-!- Dhruvt97 [~remedy441@14.139.122.120] has left #shogun []13:00
@wikinglisitsyn, so do you have a better way to discover public api than llvm?13:01
lisitsynwiking: public api of what?13:01
@wikinga class13:01
@wikingor many class13:01
lisitsynwiking: need context13:01
@wikingsay we have a bunch of classes13:01
@wiking(shogun)13:01
lisitsynaha13:02
@wikingand we need an automated way to generate a some source13:02
@wikingthat is based on the classes ctor public api13:02
@wikingyou need the correct typing etc13:02
@wikingimo llvm woudl be the best for this13:03
lisitsynwiking: I expect it to be runtime13:03
@wikingnono compile time13:03
@wikingi mean this is happening once13:03
@wiking(the context is webassembly13:03
@wiking)13:03
@wikingfor being able to generate the right bindings13:04
lisitsynwiking: I think we can have some help() that works like to_string() but much more detailed13:04
lisitsynahh13:04
lisitsynyou want programmable thing?13:04
@wikingwe need to be able to generate either a webidl13:04
@wikingor some .cpp file13:04
@wikingthat does the binding between the c++ class and some shit13:04
@wikingunfortuantely because of name mangling in c++13:04
@wikingthere's no way aroudn this13:04
lisitsynaha I see13:04
@wikingwe need to do some binding13:04
@wikingand i dont wanna do this manually13:05
@wikingfor each and every class :)13:05
@wikingyou need to generate stuff like this13:05
@wikingEMSCRIPTEN_BINDINGS(my_class_example) {13:05
@wiking  class_<MyClass>("MyClass")13:05
@wiking    .constructor<int, std::string>()13:05
@wiking    .function("incrementX", &MyClass::incrementX)13:05
@wiking    .property("x", &MyClass::getX, &MyClass::setX)13:05
@wiking    .class_function("getStringFromInstance", &MyClass::getStringFromInstance)13:05
@wiking    ;13:05
lisitsynI see13:05
@wikingand the easiest for me it seems like13:05
@wikingto take llvm13:05
@wikingand get me the ast13:05
lisitsynwiking: do you want it to generate non-generic interface for javascript?13:06
lisitsynI mean not generic put/get but fields-like?13:06
@wikinglisitsyn, i mean for the put/get13:07
@wikingwe can get back13:07
@wikinglater13:07
@wikingthat's whatever :)13:07
@wikingi first just wanna be able13:07
lisitsynah it is not about that13:07
lisitsynok13:07
@wikingto create classes :))))13:07
lisitsynok so it does no magic13:07
@wikingand call train()13:07
@wikingor apply()13:07
@wikingand for that13:07
lisitsynand you need to manually wrap everything13:07
@wikingi still need to do these bindings13:07
@wikingyes13:08
lisitsynyeah then it is llvm-based reflection I guess13:08
@wikingk13:08
@wikingbtw once we would have that13:09
@wikingwe coudl do anything :P13:09
@wikingwithout swig13:10
@wiking:)13:10
@wikingthis is what actually google people do with clif13:10
lisitsynwiking: yay our own swig ftw13:10
@wikinghttps://github.com/google/clif13:10
lisitsynwiking: we actually need C interface13:10
@wikinglisitsyn, if we woudl have ansi c interface13:10
lisitsynC interface = any language interface possible13:11
@wikingi wouldn't need to do anything like this13:11
@wikingyes13:11
lisitsynwiking: it's 30 years and counting and still no abi in C++13:11
lisitsyn:D13:11
@wiking:D13:11
lisitsynbecause it is so hard to come to any agreement13:11
@wikingwell highly advanced languages dont need abi13:11
@wiking:D13:11
PrashanthCould I know more about what features types Random Forest requires?13:27
@HeikoSPrashanth: you can set a boolean mask for each feature13:32
@HeikoSyou will get a warning if the mask is not set13:32
@HeikoSthats what you saw above13:32
@HeikoSlisitsyn: would you mind fixing that segfault in CSGOject unittest?13:35
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4166 synchronized by durovo14:07
lisitsynHeikoS: can you point me to the segfault?14:08
@HeikoSlisitsyn: http://buildbot.shogun-toolbox.org:8080/#/builders/10/builds/29914:18
@sukey[https://github.com/shogun-toolbox/shogun] New branch feature/numpy1.7 created14:26
@sukey[https://github.com/shogun-toolbox/shogun] vigsterkr pushed 2 commits:14:26
@sukeyhttps://github.com/shogun-toolbox/shogun/commit/996b17f6ad735261c150621472897c40fe60162d14:26
@sukeyhttps://github.com/shogun-toolbox/shogun/commit/2cfde497799cc02f8f32aa5dcf06f0022bb6880b14:26
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun14:52
-!- Dhruvt97 [~remedy441@14.139.122.120] has quit [Read error: Connection reset by peer]14:59
lisitsynHeikoS: is it due to string formatting?15:00
lisitsynbut you had some patch don't you?15:00
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun15:01
@HeikoSlisitsyn: yes, but it is in feature branch convoluted15:04
@HeikoSI can put it in develop later15:04
@HeikoSlisitsyn: I dont know what causes the segfault, but it happened since your patch iirc15:05
lisitsynHeikoS: probably it is string formatting15:05
lisitsynbut I can't reproduce it on hedonism15:05
@HeikoSah I know15:05
@HeikoSthe missing argument15:05
@HeikoS%s15:05
lisitsynyes15:05
lisitsynHeikoS: the problem is that you gonna have some merging conflict15:05
lisitsynif I fix that15:05
@HeikoSi am not on typedo anymore15:06
@HeikoSI think it is a deadend15:06
@HeikoSI am local now :)15:06
@HeikoSjust a sec15:06
@HeikoShttps://github.com/shogun-toolbox/shogun/blob/typedo/src/shogun/base/SGObject.h#L35515:07
@HeikoShttps://github.com/shogun-toolbox/shogun/blob/typedo/src/shogun/base/SGObject.h#L43615:07
-!- lol_ [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has joined #shogun15:07
@HeikoSjust take those guys....15:07
@HeikoSi didnt put them in a sep commit sadly15:07
lol_I am having trouble installing shogun. I cmake it, and it gives the issue "Shogun can only be built with GPL codes if the source files are in   /Users/roopalvaid/shogun/src/gpl.  Please download or disable with   LICENSE_GPL_SHOGUN=OFF."15:08
@HeikoSlol_: as said in the readmes15:08
@HeikoSgit submodule init15:08
@HeikoSgit submodule update15:08
@HeikoSto download them15:08
@HeikoSor disable the GPL when you run cmake15:09
lol_I have tried both seem to not work.15:11
lol_I tried disabling it15:11
@HeikoScmake -DLICENSE_GPL_SHOGUN=Off15:12
@HeikoSlisitsyn: you still around?15:13
lisitsynHeikoS: yes15:14
@HeikoSok so swig question15:14
@HeikoStemplate <typename T>15:14
@HeikoSvoid heiko(SGMatrix<T> m);15:14
@HeikoSI declare this15:14
@HeikoSand then swig I tell this: %template(heiko) shogun::CSGObject::heiko<float64_t>;15:14
@HeikoSand then: 338: TypeError: in method 'SGObject_heiko', argument 2 of type 'SGMatrix< double >'15:15
lol_When I used DLICENSE_GPL_SHOGUN=OFF it reported parse error and when I passed "LICENSE_GPL_SHOGUN=OFF " argument same error15:15
@HeikoSlol_: what is the exact cmd you run?15:15
@HeikoSand what is the exact error msg?15:16
lol_"cmake -LICENSE _GPL_SHOGUN=OFF .."15:16
-!- travis-ci [~travis-ci@ec2-54-234-205-100.compute-1.amazonaws.com] has joined #shogun15:16
travis-ciit's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34048280015:16
-!- travis-ci [~travis-ci@ec2-54-234-205-100.compute-1.amazonaws.com] has left #shogun []15:16
@HeikoSlol_: check your spelling15:17
@HeikoSlol_:  it is "cmake -DLICENSE ...."15:17
lol_"Please download or disable with   LICENSE_GPL_SHOGUN=OFF." this is part of the error message .     "Parse error in command line argument: -DLICENSE Should be: VAR:type=value CMake Error: No cmake script provided. CMake Error: Problem processing arguments. Aborting." error when I used -DLICENSE15:19
@HeikoSlol_: dude :)15:20
@HeikoSlol_: "cmake -D LICENSE _GPL_SHOGUN=OFF .."15:20
@HeikoSsry15:20
@HeikoScmake -DLICENSE _GPL_SHOGUN=OFF ..15:20
@HeikoSlol_: but even better download the gpl codes15:21
@HeikoSlol_: because you will have to disable SVMLIGHT as well15:21
@HeikoSif you go without GPl15:21
@HeikoS-DLICENSE_GPL_SHOGUN=Off -DUSE_SVMLIGHT=Off15:21
lol_Alright I will try installing gpl or else use this.15:22
lol_Thank you15:22
@wikinglisitsyn, WTF IS THIS :)15:27
@wikinglisitsyn, top - 15:26:49 up 207 days, 14:02, 14 users,  load average: 39.20, 29.38, 19.4215:27
@wiking:D15:27
@wikingwhen i opened top15:27
-!- lol_ [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has quit [Ping timeout: 260 seconds]15:27
@wikingi saw your processes15:27
@wiking:>15:27
@wikingcc17553 lisitsyn  20   0  678484 623032  15876 R  17.5  1.9   0:11.80 cc1plus15:27
@wikingplz use -j<REASONABLE NUMBER>15:27
@wiking:>15:27
lisitsynwiking: :P15:28
lisitsynwiking: I am mostly done sorry15:28
@wikingheheh15:28
@wikingyeah i see your processes disappearing15:28
lisitsynwiking: make -jyolo15:28
@wikingyeah i see you went -j015:29
@wiking:)15:29
* wiking wonders whether now all the distance functions are failing because of the typemap changes or lisitsyn's stuff? :D15:36
lisitsyneh?15:36
@wikinglisitsyn, testing my numpy typemap changes15:36
@wikingand all the distance functions are foobar15:36
@wikingsegfault15:36
@wikingand now i dont know why exactly15:37
@wiking:)15:37
lisitsynwiking: oh I guess error message segfaulting15:37
lisitsynwiking: fix coming15:37
@wikingjust compiling the develop15:37
@wikingchecking whether the error is the same15:37
@wikingmanual bisect ftw!15:37
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4171 opened by lisitsyn15:39
lisitsynwiking: https://github.com/shogun-toolbox/shogun/pull/4171/files should fix that15:40
@wikingwhy no develop directly? :D15:40
@wikingi mean develop is broken anyways15:40
@wiking:)))15:40
@wikingbut i think i have other problems atm :)15:41
@wikingindeed :)15:41
@wikingi have typemap troubles :)))15:41
-!- travis-ci [~travis-ci@ec2-54-234-205-100.compute-1.amazonaws.com] has joined #shogun15:41
travis-ciit's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34048280015:41
-!- travis-ci [~travis-ci@ec2-54-234-205-100.compute-1.amazonaws.com] has left #shogun []15:41
-!- le_ [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has joined #shogun15:54
le_hey I am having error when I run "git submodule update --init"15:54
lisitsynwiking: it's been a while since I last commit directly ;P15:58
lisitsynyolo15:58
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4171 merged by lisitsyn15:58
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/1a314a6ee6a060a8bb0c531a18732ea0104c541b by lisitsyn15:58
@wikingmmm15:59
@wikingjust yolo15:59
@wikingif u do -j015:59
@wikingthen you do push master15:59
@wiking:D15:59
-!- all_regrets [1b3be182@gateway/web/freenode/ip.27.59.225.130] has joined #shogun15:59
all_regretsgit submodule update --init  is giving me issues15:59
-!- le_ [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has quit [Ping timeout: 260 seconds]16:01
-!- all_regrets [1b3be182@gateway/web/freenode/ip.27.59.225.130] has quit [Ping timeout: 260 seconds]16:04
-!- gah [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has joined #shogun16:05
gahhey16:05
gahI am having an error with "git submodule update --init"16:05
@wikinggah, in case you are having troulbe with using git please read the git manual or google for your error16:10
@sukey[https://github.com/shogun-toolbox/shogun] vigsterkr pushed 2 commits:16:14
@sukeyhttps://github.com/shogun-toolbox/shogun/commit/72455e81858840637b122fad37a59fd8483e60d816:14
@sukeyhttps://github.com/shogun-toolbox/shogun/commit/ba819e6012a9657fedc43217666db50fea00e05516:14
gahevery other git command works16:17
-!- Dhruvt97 [~remedy441@14.139.122.120] has left #shogun []16:20
-shogun-buildbot:#shogun- Build trusty - libshogun - viennacl #323 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/6/builds/32316:28
-!- hc_ [01a1edb8@gateway/web/freenode/ip.1.161.237.184] has joined #shogun16:28
-!- hc_ [01a1edb8@gateway/web/freenode/ip.1.161.237.184] has quit [Client Quit]16:28
-shogun-buildbot:#shogun- Build deb1 - libshogun #301 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/10/builds/30116:35
-!- gah [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has quit [Ping timeout: 260 seconds]16:40
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun16:49
-!- Dhruvt97 [~remedy441@14.139.122.120] has left #shogun []16:49
-!- Prashanth [uid170915@gateway/web/irccloud.com/x-geaudwezzpnojtxi] has quit [Quit: Connection closed for inactivity]17:17
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 260 seconds]17:19
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has joined #shogun17:32
travis-ciit's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34052150317:32
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has left #shogun []17:32
-!- Prashanth [uid170915@gateway/web/irccloud.com/x-wceleuoysuyjvatv] has joined #shogun17:42
@wikingfuck18:10
lisitsynhah18:11
lisitsynoh ok18:11
lisitsynok then we can focus on finishing the tags transition lol18:12
@wiking:D18:12
@wikingi wonder wtf18:12
@wikingwoudl be good to know18:13
@wikingbut that's not possible :(18:13
lisitsynwiking: curious what was wrong this time18:13
lisitsyn:)18:13
-!- iglesias [50fe9643@gateway/web/freenode/ip.80.254.150.67] has joined #shogun18:27
iglesiasdo they give feedback on the gsoc decision this year?18:28
@wikingiglesias, afaik no18:28
@wiking:)18:28
iglesiasreading your e-mail atm18:29
iglesiasit sounds good to try the possibility to take 1-2 of those slots if possible18:31
iglesiasbut yeah, we should take a look at numfocus application18:31
-!- iglesias [50fe9643@gateway/web/freenode/ip.80.254.150.67] has quit [Quit: Page closed]18:39
-!- travis-ci [~travis-ci@ec2-23-20-52-66.compute-1.amazonaws.com] has joined #shogun18:41
travis-ciit's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34052150318:41
-!- travis-ci [~travis-ci@ec2-23-20-52-66.compute-1.amazonaws.com] has left #shogun []18:41
-!- Dhruvt97 [~remedy441@14.139.122.120] has joined #shogun18:51
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4142 synchronized by yashdusing19:11
-!- travis-ci [~travis-ci@ec2-23-20-52-66.compute-1.amazonaws.com] has joined #shogun19:15
travis-ciit's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34052911219:15
-!- travis-ci [~travis-ci@ec2-23-20-52-66.compute-1.amazonaws.com] has left #shogun []19:15
-!- yashdusing [6733e932@gateway/web/freenode/ip.103.51.233.50] has joined #shogun19:16
-!- yashdusing [6733e932@gateway/web/freenode/ip.103.51.233.50] has quit [Client Quit]19:18
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has joined #shogun19:26
travis-ciit's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34052911219:26
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has left #shogun []19:26
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4172 opened by naoa19:38
-!- Dhruvt97 [~remedy441@14.139.122.120] has left #shogun []19:46
-!- sj484 [0e8bf0fb@gateway/web/freenode/ip.14.139.240.251] has joined #shogun21:18
sj484is shogun participating in gsoc201821:20
-!- sj484 [0e8bf0fb@gateway/web/freenode/ip.14.139.240.251] has quit [Client Quit]21:21
-!- sj484 [0e8bf0fb@gateway/web/freenode/ip.14.139.240.251] has joined #shogun21:22
-!- sj484 [0e8bf0fb@gateway/web/freenode/ip.14.139.240.251] has quit [Client Quit]21:22
@wikinglisitsyn, ping21:35
lisitsynwiking: hey21:37
@wikingok so question pythonian unicode asdf21:37
@wikingso say we have the magical shogun::SGString<char>21:38
@wikingand i have a list[u'', ..., u'']21:38
lisitsynaha21:38
@wikingwould you mind tha tthe content of shogun::SGString is actually the utf8 encoded str?21:39
@wikingi.e. it's not the ascii encoded version21:39
@wikingthis is of course gets tricky21:39
@wikingwhen you start doing str operations on the content21:39
@wikingfor example CSubsequenceStringKernel21:40
@wikingK+=m_lambda*m_lambda*(avec[j]==bvec[k])21:40
@wiking*Kp[i][j][k];21:40
@wikingavec[j]==bvec[k] is on the byte21:41
@wikingnot on the char21:41
@wiking:)21:41
@wikingi mean on the character21:41
@wiking:P21:41
@wikingsee what i mean?21:41
@wikingi mean the other option is to try to somehow convert the utf8 str -> ascii str... but what happens when that fails? :)21:43
@wikingnot every utf8 str could be converted to utf8 :P21:43
@wikingi mean ascii21:43
@wiking:)21:43
-!- ChanServ [ChanServ@services.] has quit [shutting down]22:02
-!- Prashanth [uid170915@gateway/web/irccloud.com/x-wceleuoysuyjvatv] has quit [Quit: Connection closed for inactivity]22:07
-!- ChanServ [ChanServ@services.] has joined #shogun22:11
-!- ServerMode/#shogun [+o ChanServ] by barjavel.freenode.net22:11
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/ea14003539e2f0e409fe358f72e53532ca90ac92 by vigsterkr22:13
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has joined #shogun22:31
travis-ciit's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/34068441422:31
-!- travis-ci [~travis-ci@ec2-54-162-219-217.compute-1.amazonaws.com] has left #shogun []22:31
--- Log closed Tue Feb 13 00:00:00 2018
« Sunday, 2018-02-11 Index Tuesday, 2018-02-13 »

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