IRC logs of #shogun for Tuesday, 2018-05-08

« Monday, 2018-05-07 Index Wednesday, 2018-05-09 »
--- Log opened Tue May 08 00:00:50 2018
--- Day changed Tue May 08 2018
-!- iglesiasg [~iglesias@f119189.upc-f.chello.nl] has quit [Quit: Leaving]00:00
-shogun-buildbot:#shogun- Build nightly_bsd_license #149 is complete: Failure [failed update shogun (failure)] - http://buildbot.shogun-toolbox.org:8080/#builders/4/builds/14903:00
-shogun-buildbot:#shogun- Build clang - static analysis #135 is complete: Exception [exception update shogun (failure) uploading static_analysis (exception)] - http://buildbot.shogun-toolbox.org:8080/#builders/29/builds/13503:00
-shogun-buildbot:#shogun- Build cookbook - nightly #137 is complete: Failure [failed update shogun (failure)] - http://buildbot.shogun-toolbox.org:8080/#builders/14/builds/13703:00
-shogun-buildbot:#shogun- Build doxygen - nightly #135 is complete: Failure [failed update shogun (failure)] - http://buildbot.shogun-toolbox.org:8080/#builders/12/builds/13503:00
-!- HeikoS [~heiko@bmi-airport-1.inf.ethz.ch] has joined #shogun10:43
-!- mode/#shogun [+o HeikoS] by ChanServ10:43
-!- witness [uid10044@gateway/web/irccloud.com/x-mcszsykrhrjmoqoy] has quit [Quit: Connection closed for inactivity]10:50
@wikinggeektoni here?11:15
sukey1[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4275 synchronized by karlnapf11:18
@HeikoSwuwei: hi11:21
@HeikoSwuwei: you still have that problem with the unsigned parameter?11:21
wuweiyeah I still don't get a way11:22
wuweiBesides some parameters are registered as CSGObject*, which also causes errors in python, for example in random forests11:24
@HeikoSwuwei: yes11:28
@HeikoSwuwei: so for sgobject11:28
@HeikoSsee my recent patches. We dont register as SGObject anymore11:29
@HeikoSyou want to register by the base type11:29
@HeikoSCMachine*11:29
@HeikoSCKernel* etc11:29
@HeikoSthen you can put the object11:29
@HeikoSfor the unsigned parameter, you will have to give me an example11:29
@HeikoSwuwei: but generally, we have to dispatch the numeric types (i.e. try-error)11:29
@HeikoSwuwei: check shogun.i11:29
@HeikoSthere we dispatch ints to floats and things like this11:30
@HeikoSif you write obj.put("number", 4)11:30
@HeikoSit will try to set it as float32,float64,int32,int64, etc11:30
wuweiFor the unsigned, see notebook for LMNN11:30
@HeikoSso might have to add the unsigned dispatcher there11:30
@HeikoSwhich parameter of LMNN?11:30
wuweiLet me check11:31
wuweimax iter , IIRC11:32
@HeikoSwuwei: so for features11:33
@HeikoSSG_ADD((CSGObject**) &m_features, "features", "Training features",11:33
@HeikoSMS_NOT_AVAILABLE)11:33
@HeikoSyou have to remove the cast to CSGObject11:34
@HeikoSfor the unsigned int11:35
@HeikoSmmmh11:35
@HeikoSwiking: you think we should support uint as type?11:35
@HeikoSI tend to say no11:35
@wikingheheh11:35
@HeikoSwiking: what do you think?11:35
@wikingwhere is it unit?11:36
@HeikoSI mean there is very few places where this is used anyways11:36
@wikingwhich param?11:36
@HeikoSLMNN::m_maxiter11:36
@HeikoSmaxiter is just an int in most algorithms11:36
@HeikoSjust fernando decided to use uint11:36
@wikingehhe11:36
@wikingfernando makes things correct :P11:36
@wikingyeah i mean11:36
@wikingif this is the only usecase11:36
@wikingthen lets just convert it to int64 :)11:37
@wikingand tada done11:37
@HeikoSyep11:37
@HeikoSwuwei: so you can just change the type of that parameter11:37
wuweiokay I see11:38
@HeikoSwuwei: also check out the bottom of shogun.i11:38
@HeikoSthere are the dispatchers11:39
sukey1[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4275 merged by karlnapf11:39
sukey1[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/bce30238faf0bd5afb3cf9df9ab9d08076ad5bca by karlnapf11:39
-shogun-buildbot:#shogun- Build deb1 - libshogun #405 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/10/builds/40511:43
lisitsynHeikoS: the latest patches are like converging to something nice11:48
@HeikoSlisitsyn: yeah it is getting there11:54
@HeikoSlisitsyn: I managed to get rid of the dispatching in put/add as well11:54
@HeikoSjust cleaning up another PR11:55
@HeikoSlisitsyn: gotta run, be back in half an hour11:55
lisitsynalright :)11:56
@HeikoSlisitsyn: back12:20
@HeikoSlisitsyn: any news on your side?12:21
@HeikoSlisitsyn: there was this thing12:21
@HeikoSregister callback with get12:21
lisitsynnot yet sorry12:21
@HeikoSand what else? there was another thingi12:21
lisitsynprototype of callback12:21
lisitsynand merge plugins12:21
@HeikoSah yeah12:21
@HeikoSYou might specialize is_sg_base for specific T.12:23
@HeikoSlisitsyn: can you give me a hint on that?12:23
lisitsynHeikoS: ah yeah sure12:24
lisitsyndeclare template<class T> struct is_sg_base12:24
lisitsynand then declare12:24
lisitsynstruct is_sg_base<CMacihine> { ... }12:24
@HeikoSI see12:24
lisitsynI don't know if it is better12:24
@HeikoSso we then just define more template specialisations12:24
@HeikoSmmh12:25
@HeikoSand you know about the constexpr?12:25
lisitsyninside the is_sg_base you can declare static constexpr something12:25
lisitsynwell I don't think constexpr is very important in here12:25
@HeikoSok12:27
@HeikoSI think i will just leave it for now12:27
@HeikoSwe can always improve this12:27
@HeikoSjust wanted to avoid the dispatching code escalating12:27
sukey1[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4276 synchronized by karlnapf12:34
@HeikoSlisitsyn:  ^12:35
sukey1[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4276 synchronized by karlnapf12:35
sukey1[https://github.com/shogun-toolbox/shogun] karlnapf pushed 2 commits:12:56
sukey1https://github.com/shogun-toolbox/shogun/commit/e70c6d6010a77e6269b9e2a565acfe45411bcd9d12:56
sukey1https://github.com/shogun-toolbox/shogun/commit/5bede6e8603fe1d895649c19ce0fc56299282dda12:56
-!- durovo2 [~durovo@37.8f.559e.ip4.static.sl-reverse.com] has quit [Remote host closed the connection]12:59
-!- durovo [~durovo@37.8f.559e.ip4.static.sl-reverse.com] has joined #shogun12:59
@HeikoSwuwei: yo13:18
@HeikoSwuwei: about your notebook PR13:18
@HeikoSdid you run all notebooks with the changes you made?13:18
wuweihi heiko13:18
wuweiyes i did13:18
@HeikoSgreat13:18
@HeikoSgreat patch! :)13:19
sukey1[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4269 merged by karlnapf13:19
sukey1[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/cceeb315e9f75b41364eddd53a5b7e874acf9d88 by karlnapf13:19
@HeikoSwuwei: I guess there are still some left?13:19
wuweiyeah i'm still woking on it13:19
@HeikoSwuwei: cool!13:19
@HeikoSwuwei: I sent out an email saying that you are working on the notebooks, so nobody starts working on the same thing13:23
@HeikoSwuwei: next step is to port all the legacy python examples to meta examples13:23
wuweipython in undocumented?13:25
wuweiall right!13:25
-!- HeikoS [~heiko@bmi-airport-1.inf.ethz.ch] has quit [Ping timeout: 264 seconds]13:28
-!- HeikoS [~heiko@inf-ise-etx-dock-1-111.ethz.ch] has joined #shogun13:48
-!- mode/#shogun [+o HeikoS] by ChanServ13:48
-!- HeikoS [~heiko@inf-ise-etx-dock-1-111.ethz.ch] has quit [Read error: Connection reset by peer]13:51
-!- HeikoS [~heiko@inf-ise-etx-dock-1-111.ethz.ch] has joined #shogun13:51
-!- mode/#shogun [+o HeikoS] by ChanServ13:51
-!- HeikoS [~heiko@inf-ise-etx-dock-1-111.ethz.ch] has quit [Client Quit]13:51
-!- HeikoS [~heiko@inf-ise-etx-dock-1-111.ethz.ch] has joined #shogun13:59
-!- mode/#shogun [+o HeikoS] by ChanServ13:59
-!- HeikoS [~heiko@inf-ise-etx-dock-1-111.ethz.ch] has quit [Ping timeout: 256 seconds]14:04
-!- HeikoS [~heiko@inf-ise-etx-dock-1-111.ethz.ch] has joined #shogun14:21
-!- mode/#shogun [+o HeikoS] by ChanServ14:21
-!- HeikoS [~heiko@inf-ise-etx-dock-1-111.ethz.ch] has quit [Ping timeout: 260 seconds]14:34
-!- travis-ci [~travis-ci@ec2-54-160-194-98.compute-1.amazonaws.com] has joined #shogun14:40
travis-ciit's Heiko Strathmann'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/37631059614:40
-!- travis-ci [~travis-ci@ec2-54-160-194-98.compute-1.amazonaws.com] has left #shogun []14:40
sukey1[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4267 closed by vigsterkr14:42
@wikingwuwei, we have our video conference at 1500 UTC right?14:46
@wikingthat's 10pm beijing time?14:47
wuwei1500 UTC, 11pm in beijing :)14:47
@wikingaaah14:47
@wikingokok i've got it14:47
@wikingit's 5pm CET14:47
-!- travis-ci [~travis-ci@ec2-54-211-20-75.compute-1.amazonaws.com] has joined #shogun15:29
travis-ciit's Heiko Strathmann'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/37631059615:29
-!- travis-ci [~travis-ci@ec2-54-211-20-75.compute-1.amazonaws.com] has left #shogun []15:29
@wikinglisitsyn, https://stackoverflow.com/questions/4131091/dynamic-cast-from-void15:45
@wikingideas? :)15:45
@wikingcoz storage is void*15:45
lisitsynwiking: I guess you need reinterpret_cast<SGObject*> then dynamic_cast<T>16:01
shogitter(vigsterkr) lisitsyn craaaaazyyyyzuzi16:01
shogitter(vigsterkr) :)16:02
@wikingmmm16:02
lisitsyn?\_(?)_/?16:02
lisitsynyolo16:02
shogitter(vigsterkr) YOL0!16:02
@wikingmmm shogitter is one way :)16:04
@wikinglisitsyn, http://matthias.vallentin.net/slides/berkeley-cpp-summit-caf.pdf16:04
@wikinglisitsyn, check page 3716:06
@wikingbetter than erlang :O16:06
lisitsynah yeah16:07
lisitsynerlang is not fast indeed16:07
@wikingif we ever wanna do distrib shit :)16:08
@wikinglisitsyn, ok so do you know why would clone fail on an obj where i dynamically add a param that is owned :)16:16
@wikingmeaning it's not a member var :)16:16
lisitsynwiking: what is 'fail'?16:17
@wikingC++ exception with description "[ERROR] In file ../src/shogun/base/SGObject.cpp line 730: There is no parameter called "seed" in RandomCARTree16:17
@wikingso i dynamically add the seed tag to the RandomCARTree16:17
lisitsynoh16:17
@wikingwith create_param16:17
lisitsynwhere do you add it?16:17
@wikingbasically it's a c->create_parameter(kSeedTag, AnyParameter(make_any(seed)));16:17
@wikingit's totally dynamic :) i.e. not registered or anythign16:18
lisitsynclone creates empty object16:18
@wikingthat's fine :)16:18
@wikingbut i mean shouldn't it fill up the parameters map16:19
@wikingfrom src?16:19
lisitsynno, it tries to access the parameter on an empty object16:19
lisitsynbecause we call clone_from16:19
@wikingmiju16:20
@wikingyeah just checked SGO::clone16:20
@wikingso we actually do not support these type of runtime crazyness :)16:21
@wikingwe need std::optional :)16:22
@wikingmaybe? :>16:22
@wikingwould Any(Maybe<int>) work?16:24
@wikingok lets try... :)16:24
lisitsynhaha16:27
lisitsynwiking: I think maybe is to be dropped16:35
lisitsynC++14 has optional already16:35
lisitsynah sorry it is 1716:35
lisitsyndo we use 17 already?16:36
@wikingnoup16:36
@wiking:(16:36
@wikingyeah i've checked16:36
@wiking17 is not really supported yet16:36
@wikingso we do maybe for the time being16:36
lisitsynyeah jsut year more16:36
@wiking:>16:45
@wikinglisitsyn, btw since seed is part of self->map... essentially we could just do .put("seed", Maybe<int32_t>)16:46
@wikingdo we want to have a specialized set_seed in sgo?16:47
@wikingthat wraps the whole story of maybe<int32_t> ?16:47
@wiking(maybe only in swig?)16:47
@wikingi mean there has to be a specially handler for put("seed",...) because of traversing the children...16:47
@wikingso unfortunately i'll have to do a shitty type traited catch for the put("seed")16:48
@wiking:(16:48
@wikingor should we just check if you do a put16:48
@wiking"seed" then throw an exception that you should set that via16:49
@wiking.set_seed ? :)16:49
@wikingshitty16:49
@wikingno matter how we do it16:49
@wikingor?16:49
-!- travis-ci [~travis-ci@ec2-54-91-120-31.compute-1.amazonaws.com] has joined #shogun20:00
travis-ciit's Shubham Shukla's turn to pay the next round of drinks for the massacre he caused in shubham808/shogun: https://travis-ci.org/shubham808/shogun/builds/37646017520:00
-!- travis-ci [~travis-ci@ec2-54-91-120-31.compute-1.amazonaws.com] has left #shogun []20:00
-!- travis-ci [~travis-ci@ec2-54-81-80-255.compute-1.amazonaws.com] has joined #shogun20:20
travis-ciit's Shubham Shukla's turn to pay the next round of drinks for the massacre he caused in shubham808/shogun: https://travis-ci.org/shubham808/shogun/builds/37648561920:20
-!- travis-ci [~travis-ci@ec2-54-81-80-255.compute-1.amazonaws.com] has left #shogun []20:20
-!- travis-ci [~travis-ci@ec2-54-198-47-146.compute-1.amazonaws.com] has joined #shogun20:46
travis-ciit's Shubham Shukla's turn to pay the next round of drinks for the massacre he caused in shubham808/shogun: https://travis-ci.org/shubham808/shogun/builds/37646440420:46
-!- travis-ci [~travis-ci@ec2-54-198-47-146.compute-1.amazonaws.com] has left #shogun []20:46
sukey1[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4277 opened by shubham80821:17
-!- travis-ci [~travis-ci@ec2-54-81-80-255.compute-1.amazonaws.com] has joined #shogun22:11
travis-ciit's Shubham Shukla's turn to pay the next round of drinks for the massacre he caused in shubham808/shogun: https://travis-ci.org/shubham808/shogun/builds/37653318322:11
-!- travis-ci [~travis-ci@ec2-54-81-80-255.compute-1.amazonaws.com] has left #shogun []22:11
-!- travis-ci [~travis-ci@ec2-54-91-120-31.compute-1.amazonaws.com] has joined #shogun22:20
travis-ciit's Shubham Shukla's turn to pay the next round of drinks for the massacre he caused in shubham808/shogun: https://travis-ci.org/shubham808/shogun/builds/37653612422:20
-!- travis-ci [~travis-ci@ec2-54-91-120-31.compute-1.amazonaws.com] has left #shogun []22:20
--- Log closed Wed May 09 00:00:58 2018
« Monday, 2018-05-07 Index Wednesday, 2018-05-09 »

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