IRC logs of #shogun for Thursday, 2018-02-08

« Wednesday, 2018-02-07 Index Friday, 2018-02-09 »
--- Log opened Thu Feb 08 00:00:53 2018
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 248 seconds]00:25
-!- travis-ci [~travis-ci@ec2-23-20-242-232.compute-1.amazonaws.com] has joined #shogun00:28
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/33873029100:28
-!- travis-ci [~travis-ci@ec2-23-20-242-232.compute-1.amazonaws.com] has left #shogun []00:28
-!- travis-ci [~travis-ci@ec2-54-204-229-205.compute-1.amazonaws.com] has joined #shogun00:47
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/33873029100:47
-!- travis-ci [~travis-ci@ec2-54-204-229-205.compute-1.amazonaws.com] has left #shogun []00:47
-!- travis-ci [~travis-ci@ec2-54-87-77-174.compute-1.amazonaws.com] has joined #shogun01:37
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/33874795501:37
-!- travis-ci [~travis-ci@ec2-54-87-77-174.compute-1.amazonaws.com] has left #shogun []01:37
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/d404710d9e5ee392259651b8075c8d178bbcdb07 by vigsterkr02:00
-!- travis-ci [~travis-ci@ec2-54-227-67-178.compute-1.amazonaws.com] has joined #shogun02:02
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/33874795502:02
-!- travis-ci [~travis-ci@ec2-54-227-67-178.compute-1.amazonaws.com] has left #shogun []02:02
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/a3acc968c2b0d2c4adde3b03743b6e67d8df09e5 by vigsterkr02:10
-!- travis-ci [~travis-ci@ec2-54-87-77-174.compute-1.amazonaws.com] has joined #shogun03:27
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/33873029103:27
-!- travis-ci [~travis-ci@ec2-54-87-77-174.compute-1.amazonaws.com] has left #shogun []03:27
-!- travis-ci [~travis-ci@ec2-54-160-166-164.compute-1.amazonaws.com] has joined #shogun03:46
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/33878865203:46
-!- travis-ci [~travis-ci@ec2-54-160-166-164.compute-1.amazonaws.com] has left #shogun []03:46
-!- travis-ci [~travis-ci@ec2-54-87-77-174.compute-1.amazonaws.com] has joined #shogun03:53
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/33878865203:53
-!- travis-ci [~travis-ci@ec2-54-87-77-174.compute-1.amazonaws.com] has left #shogun []03:53
-!- wuwei [ca781378@gateway/web/freenode/ip.202.120.19.120] has joined #shogun07:28
wuweihello, there are some methods in linalg that use non-const reference for input, e.g SGVector<T> matrix_prod(SGMatrix<T>& A, SGVector<T>& b, bool transpose = false), is there particular reasons for this, or may I add const qualifiers to them ?07:30
@wikingwuwei, hey08:23
@wikingwuwei, if they are input vars08:24
@wikingthey should be const08:24
wuweiall right :D08:37
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4159 opened by vinx1308:46
-!- wuwei [ca781378@gateway/web/freenode/ip.202.120.19.120] has quit [Ping timeout: 260 seconds]09:41
-!- travis-ci [~travis-ci@ec2-54-160-166-164.compute-1.amazonaws.com] has joined #shogun10:36
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/33873029110:36
-!- travis-ci [~travis-ci@ec2-54-160-166-164.compute-1.amazonaws.com] has left #shogun []10:36
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has joined #shogun10:38
-!- mode/#shogun [+o HeikoS] by ChanServ10:38
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4159 merged by karlnapf10:42
@sukey[https://github.com/shogun-toolbox/shogun] karlnapf pushed 2 commits:10:42
@sukeyhttps://github.com/shogun-toolbox/shogun/commit/dbe5fdd4fa6424de9ca5a1f0187992b69711247610:42
@sukeyhttps://github.com/shogun-toolbox/shogun/commit/15dcbe4fa39e18b922e4e9b9810d5006961ffa6f10:42
-shogun-buildbot:#shogun- Build trusty - libshogun - viennacl #316 is complete: Failure [failed test (failure)] - http://buildbot.shogun-toolbox.org:8080/#builders/6/builds/31610:47
@HeikoSwiking: jo!12:04
@HeikoSwiking: I am going to check out the typedo branch a bit now, is everything you did pushed in there?12:05
@wikinghave fun12:14
@wikingits pretty simple12:14
@HeikoSthx12:15
@HeikoSbtw you'd like a bread on Monday? :D12:15
@wikingno12:16
@wikingi think i have to12:16
@wikingcheck myself out12:16
@wikingi have some symthoms that probably i have gluten alergy12:16
@wiking:)(12:16
@HeikoSuh12:16
@HeikoSpitty12:16
@wikingso no bread for me for weeks now12:16
@HeikoSsourdough has more broken down gluten actually, but doesnt help if you have an allergy I guess12:17
@wikingyeah i know12:17
@wikingbut something is weird for the last 6 months12:17
@wikingso now i'm eliminating the particles12:17
@wikingand this gluten thing seems to be doing something12:17
@HeikoSsure12:17
@wikingi have my bread drying there for this week already12:19
@HeikoSlisitsyn: you there?12:21
@HeikoSwiking: there is a problem with registering as CKernel*12:21
@HeikoSany doesnt like abstract classes12:22
lisitsynHeikoS: hey12:22
@HeikoSlisitsyn: have a moment?12:22
lisitsynyes12:22
@HeikoSlisitsyn: ok so12:22
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4160 opened by vinx1312:22
@HeikoSthere is an alternative approach to the put thing12:22
@HeikoSwhich is templating the put method12:22
@HeikoSremember the problem with the kernel being registered as CSGObject?12:22
@HeikoSand then put("kernel", CKernel*) doesnt work12:22
@wikingHeikoS, that's too bad... but that should be somehow what is enforcing type ... other than global OBJ12:22
@wikingsee my comment on github12:23
@HeikoSlisitsyn: so now I though, let's just register as CKernel* in tags12:23
@HeikoSbut any doesnt seem to like abstract classes ..12:23
@HeikoScomments?12:23
-!- ironstark [uid221607@gateway/web/irccloud.com/x-jleyurzuaqjcmawz] has quit [Ping timeout: 256 seconds]12:23
@HeikoSlisitsyn: it is the decltype stuff12:24
@HeikoS/home/heiko/git/shogun/src/shogun/lib/any.h:302:27: error: invalid abstract parameter type 'shogun::CKernel'12:24
@HeikoS       [](T lhs, T rhs) -> bool { return any_detail::compare(lhs, rhs); });12:24
lisitsynHeikoS: it shouldn't care about abstractness12:24
lisitsynas soon as it is pointer12:24
@HeikoSwatch_param("kernel", &kernel, "");12:24
@HeikoS^ that's all I did12:25
lisitsynHeikoS: why don't put("kernel", CSGObject*) work for kernel?12:26
@HeikoSlisitsyn: because then12:26
@HeikoSput("kernel", CKernel*) will not work12:26
lisitsynI mean I did it and it worked12:26
lisitsyn%)12:27
@wikinglisitsyn, because then you say in has<CSGObj>12:27
@wikingthat it should be an sgobj12:27
@HeikoSand also I can put any class yes12:27
@wikingwhich is *always* true12:27
@wikingfor KernelMachine12:27
@wikingand btw12:27
@wikingif you say that its fine12:27
lisitsynyes but it fails on cast?12:27
@wikingthen why do we return12:27
lisitsynahh12:27
@wikingCKernel* with kernel("")12:27
@wiking:)12:27
lisitsynyou mean you can put everything here12:27
lisitsynso it works but wrong way12:27
@wikingit's not really consequent12:27
@HeikoSy12:27
lisitsynok12:27
@wikinglisitsyn, i exaplained12:28
@wikinglisitsyn, ready github12:28
@wiking:)12:28
@wikingeven pinged both of u12:28
lisitsynok ok let me think12:28
@wikinganyhow HeikoS this works just that it give u a nice deprecation warning12:28
@wikingso that we do somethign with it12:28
@HeikoSlisitsyn: maybe I am just doing something stupid ...12:29
@HeikoSwait a sec12:29
@HeikoSwatch_param("kernel", (CSGObject**) &kernel, "");12:29
@HeikoSalso doenst work12:29
@HeikoSbut it should, since this is what SG_ADD does12:29
@HeikoSlisitsyn: in fact it works12:32
@HeikoSsorry :)12:32
@HeikoSmmh12:35
@HeikoS53: [ERROR] In file /home/heiko/git/shogun/src/shogun/base/SGObject.h line 357: Type for parameter with name "kernel" is not correct.12:35
@HeikoSstill!12:35
@HeikoSsvm->put("kernel", (CKernel*) kernel)12:39
@HeikoSsvm->put("kernel", (CSGObject*) kernel)12:39
@HeikoSboth fail12:39
@HeikoSeven GaussianKernel* fails :(12:40
@wiking?12:44
@wikingwhich case?12:44
@wikingthe templates?12:44
@wikingor you are testing something else12:44
@HeikoSwiking: I think that the put call that is matched is the one for CSGObject*12:45
@wikingbut i mean which case12:45
@HeikoSah mayb enot12:45
@wikingyou are giving half info12:45
@HeikoSsry12:45
@wikingi dont understand12:45
@HeikoSI changed12:45
@HeikoSwatch_param("kernel", &kernel, .. )12:45
@HeikoSso registered is CKernel12:45
@wikingyeah i mean hopefully then12:46
@wikingit'll fail12:46
@wiking:)12:46
@wikingif u use the template12:46
@wikingif you managed to compile with registering a CKernel type12:47
@wikingthen the template mosdef should fail12:47
@HeikoSbut kernel is of type CKernel*12:47
@HeikoSyeah I also get this12:47
@HeikoS/home/heiko/git/shogun/build/examples/meta/cpp/meta_api/kwargs.cpp:70:21: warning: 'void shogun::CSGObject::put(const string&, T) [with T = shogun::CKernel*; std::enable_if_t<std::is_convertible<T, shogun::CSGObject*>::value>* <anonymous> = 0u; std::__cxx11::string = std::__cxx11::basic_string<char>]' is deprecated: SGObject parameters should be registered as a base class. [-Wdeprecated-declarations]12:47
@HeikoS svm->put("kernel", k);12:47
@HeikoS                     ^12:47
@wikingyeah12:47
@wikingi mean if you managed to REALLY register12:47
@wikingCKernel*12:47
@wikingthen obviously12:47
@wikingyou should drop that deprecated template func12:47
@wikingbecause this way of course it fails12:47
@wikingit waits for a CKernel*12:48
@wikingbut that upcasts it to SGO12:48
@wikingand then it is like12:48
@HeikoSbut that's what I pass12:48
@wikingSGO* != CKernel*12:48
@wikingso fucku12:48
@wikingthat's what the error saying12:48
@HeikoSno12:48
@wikingcan u12:48
@HeikoSk is of type CKernel*12:48
@wikingyes12:48
@wikingand you registered kernel as?12:48
@HeikoSCKernel*12:48
@wikingok cool12:48
@wikingthen drop the whole function deprecation12:48
@wikingi mean the deprecated one12:49
@HeikoSok will try12:49
@wikingthat is in the ifdef12:49
@wikingand drop the enable_if_t thingy12:49
@wikingfrom the 2nd template param12:49
@HeikoSok12:49
@wikingof the other put and put typename U = void12:49
@wikingof course it gonna fail12:49
@wikingif you managed to put CKernel*12:49
@wikingas this one is now upcasting everything to be SGO*12:49
@HeikoSjust one template parameter?12:49
@wikingand you expect CKernel*12:49
@HeikoSyeah I get it now12:49
@wikingwhat do you mean12:50
@wikingno12:50
@HeikoSor second one empty?12:50
@wikingthe second one is the old one12:50
@wikingtypename U= void12:50
@HeikoSkk12:50
@HeikoStemplate <typename T, typename U=void>12:50
@HeikoSvoid put(const std::string& name, const T value)12:50
@wikingyes12:50
@wikingand the other put should not be in it12:50
@wikingbut how did u register12:50
@wikingCKernel12:50
@HeikoSsee above12:51
@wikingto the param fw?12:51
@wikingcan you copypaste12:51
@wikingi mean lol12:51
@wiking:)12:51
@HeikoSatch_param("kernel", &kernel, .. )12:51
@HeikoSw12:51
@wikingyeah12:51
@wikingbut ->add?12:51
@wikingor no ->add?12:51
@HeikoSthe add is still sgo12:51
@HeikoSbut nothing changes if I remove this12:51
@wikingimo that wont work12:51
@HeikoSas tags dont care about ->add12:51
@wikingas you are still registering12:51
@wikingand has is checking12:52
@HeikoSwe have two parameter frameworks in place12:52
@wikingfor SGO*12:52
@wikingyes12:52
@wikingi know12:52
@wikingi've tried that yesterday12:52
@wikingit still checks for SGO12:52
@wikingafaik12:52
@HeikoSthe frameworks never interact12:52
@HeikoSso if I register in old or not should not have an impact on has12:53
@HeikoSi in fact can just comment it out for now12:53
@wikingafaik when i tried what you are writing now12:53
@wikingit still checked for SGO12:53
@HeikoSmaybe12:53
@HeikoSbut not because of ->add12:53
@HeikoSthat's any/tags then12:53
@HeikoSbut yeah checking12:53
@wikingi'm just saying thats what it did12:54
@wikingwhen u said that yesterday12:54
@wikingand i've added that12:54
@HeikoSthis things a beast12:55
@HeikoSquite interesting, trying to put all these langs under one API12:56
@HeikoSwith type safety and generic put12:56
@HeikoSok code's compiling12:56
@HeikoScoffeeeee12:56
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/20c432a4893b6fc6d1b5f8e1f38346918dad9208 by karlnapf13:13
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4160 merged by karlnapf13:14
@sukey[https://github.com/shogun-toolbox/shogun] karlnapf pushed 2 commits:13:14
@sukeyhttps://github.com/shogun-toolbox/shogun/commit/1a68c421ddf35051d544340774a3d58ccdc3ad2d13:14
@sukeyhttps://github.com/shogun-toolbox/shogun/commit/f76b556aa85ad9df0850dfb02ff031dcdb87afff13:14
@HeikoS378:   File "/home/heiko/git/shogun/build/examples/meta/python/meta_api/kwargs.py", line 60, in <module>13:29
@HeikoS378:     svm.put("kernel", k)13:29
@HeikoS378: RuntimeError: [ERROR] In file /home/heiko/git/shogun/src/shogun/base/SGObject.h line 357: Type for parameter with name "kernel" is not correct.13:29
@HeikoSah so need to make put in SWIG only accept sgo13:29
@HeikoSand in c++ all13:29
@HeikoSjust realised this was kinda there13:29
@HeikoSdamn ;)13:31
-!- rcurtin [~ryan@knife.lugatgt.org] has quit [Ping timeout: 260 seconds]13:34
-!- rcurtin [~ryan@knife.lugatgt.org] has joined #shogun13:34
@HeikoSnope it is actually different13:47
@HeikoSlisitsyn: so we can only have a put<CSGObject*> in swig13:47
lisitsynHeikoS: just by changing how we register?13:47
@HeikoSnono13:48
@HeikoSso I mean I can register as CKernel*13:48
lisitsynyes13:48
@HeikoSand that works with cpp13:48
lisitsynand keep put(SGObject*) *only*13:48
lisitsynok13:48
@HeikoSyes for SWIG13:48
@HeikoSthats what is in the feautre branch now13:48
@HeikoSBUT13:48
@HeikoS378: RuntimeError: [ERROR] In file /home/heiko/git/shogun/src/shogun/base/SGObject.h line 357: Type for parameter with name "kernel" is not correct.13:48
@HeikoSand that is since13:48
@HeikoS %template(put) shogun::CSGObject::put<type, void>;13:48
@HeikoSSUPPORT_TAG(object, CSGObject*)13:49
lisitsynhah so we still need casting in any13:49
@HeikoSin there, the argument is SGObject13:49
@HeikoSyes :)13:49
lisitsynI see13:49
@HeikoSso need to be able to pass SGO to CKernel*13:49
@HeikoSand if it is correct, then it works, otherwise runtime error13:50
@HeikoSkinda logical13:50
lisitsynyes13:50
@HeikoSbut the other way around from what we had earlier (registering everything as SGObject)13:50
@HeikoSin swig we can only have put for SGObject due to overloading13:50
lisitsynHeikoS: wait13:50
lisitsyn:)13:50
@HeikoSbut internally we want to store the type13:50
lisitsynHeikoS: the type is actually not lost13:50
@HeikoSso be able to avoid put("kernel", features)13:50
lisitsynso we can avoid patching all the SG_ADDs13:51
lisitsynwhich I want to avoid13:51
lisitsynHeikoS: the trick is that the type is still here13:51
lisitsynso it might worth a patch in any13:51
lisitsynthat just checks if they are castable to each other13:51
@HeikoSand then we still downcast everything to CSGObject?13:52
lisitsynyes13:52
@HeikoSok, that would be better13:52
lisitsynyes because touching every SG_ADD13:52
lisitsynis meh13:52
@HeikoSi did this for now13:52
@HeikoShttps://github.com/shogun-toolbox/shogun/commit/20c432a4893b6fc6d1b5f8e1f38346918dad9208#diff-389ad46d2d6e6e20f6d97b6943e2783cR63813:52
@HeikoSjust to test13:52
@HeikoSbut pls pls fix it :)13:52
lisitsynok let me spend some time after lunch on that13:52
lisitsynI think I've got some time13:52
@HeikoSand we still register using SG_ADD, expose only put<CSGObject*> to swig, and all that13:53
@HeikoSlet me know13:53
@HeikoSI will be off now13:53
@HeikoSits a feature branch13:53
@HeikoStypedo13:53
@HeikoSsee you13:53
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4161 opened by vinx1313:58
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 255 seconds]13:58
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/416113:58
-!- ironstark [uid221607@gateway/web/irccloud.com/x-qmdiqxzgjjyivdos] has joined #shogun14:19
-!- travis-ci [~travis-ci@ec2-54-160-166-164.compute-1.amazonaws.com] has joined #shogun14:36
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/33894853814:36
-!- travis-ci [~travis-ci@ec2-54-160-166-164.compute-1.amazonaws.com] has left #shogun []14:36
-!- travis-ci [~travis-ci@ec2-54-87-77-174.compute-1.amazonaws.com] has joined #shogun14:56
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/33894853814:56
-!- travis-ci [~travis-ci@ec2-54-87-77-174.compute-1.amazonaws.com] has left #shogun []14:56
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4162 opened by lisitsyn17:37
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/416217:37
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4161 merged by vigsterkr18:57
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/5993912e48e6c2cd5251aacfa1dc5416b9a892a4 by vigsterkr18:57
-shogun-buildbot:#shogun- Build trusty - libshogun - viennacl #320 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/6/builds/32019:01
-!- iglesias [503877bd@gateway/web/freenode/ip.80.56.119.189] has joined #shogun20:30
-!- iglesias [503877bd@gateway/web/freenode/ip.80.56.119.189] has quit [Quit: Page closed]20:37
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4162 synchronized by lisitsyn21:53
-!- Netsplit *.net <-> *.split quits: @ChanServ21:57
-!- Netsplit over, joins: @ChanServ22:31
--- Log closed Fri Feb 09 00:00:54 2018
« Wednesday, 2018-02-07 Index Friday, 2018-02-09 »

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