IRC logs of #shogun for Saturday, 2016-04-02

« Friday, 2016-04-01 Index Sunday, 2016-04-03 »
--- Log opened Sat Apr 02 00:00:56 2016
-!- besser82 [~besser82@fedora/besser82] has joined #shogun00:30
-!- mode/#shogun [+o besser82] by ChanServ00:30
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 248 seconds]01:00
shogun-buildbotbuild #1113 of nightly_default is complete: Failure [failed notebooks]  Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/1113  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Ariane Paola Gomes <arianepaola@users.noreply.github.com>04:08
-!- BurningBear [~smuxi@121.237.190.164] has joined #shogun07:47
-!- arianepaola [~ariane@unaffiliated/arianepaola] has quit [Ping timeout: 246 seconds]09:13
-!- BurningBear [~smuxi@121.237.190.164] has quit [Ping timeout: 248 seconds]09:43
-!- besser82 [~besser82@fedora/besser82] has joined #shogun11:25
-!- mode/#shogun [+o besser82] by ChanServ11:25
-!- besser82 [~besser82@fedora/besser82] has quit [Remote host closed the connection]11:33
-!- besser82 [~besser82@fedora/besser82] has joined #shogun11:34
-!- mode/#shogun [+o besser82] by ChanServ11:34
-!- c4goldsw [8a33717d@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.113.125] has joined #shogun14:33
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun14:54
-!- mode/#shogun [+o HeikoS] by ChanServ14:54
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit]14:56
c4goldswAnyone around?  I'm getting a build error that being caused by a bug.  I believe I now how to fix it, but I want to clarify I'm taking the right approach.15:17
c4goldswwiking are you around?15:39
lisitsync4goldsw: /me at your service15:41
c4goldswGreat.15:42
c4goldswI believe I've encountered a bug that simply requires using an #ifdef.  That aside, whenever I try to build, the source file is overwritten - why is this happening?15:42
lisitsynwhat source file?15:43
c4goldswHold on.15:43
c4goldswexamples/meta/cpp/gaussian_process/gaussian_process_regression.cpp15:44
lisitsynah I see15:44
lisitsynyes everything in meta/ is being autogenerated15:44
c4goldswFrom what?15:44
lisitsync4goldsw: examples/meta/src15:45
c4goldswGreat, I'll try doing that then.  I'll probably have a few questions about the bug in a few minutes.15:45
c4goldswI just opened the corresponding .sg file and realized that I can't place an #ifdef, correct?15:47
c4goldswlisitsyn I need to ensure that the code is only compiled if HAVE_NLOPT is true.  Otherwise, an empty main should be provided.  gaussian_process_regression.cpp depends on CGradientModelSelection, which is only defined if #ifdef HAVE_NLOPT - I'm getting errors saying that CGradientModelSelection is undefined.  I believe that having the #ifdef will so15:53
c4goldswlve the issue.15:53
lisitsync4goldsw: yeah it is a bit more tough15:54
lisitsynwe can't place ifdef there15:54
c4goldswWhat can we do then?15:55
lisitsynwe can patch our examples generator15:55
lisitsynto support some Requires(HAVE_NLOPT)15:55
c4goldswHow would this be done?15:56
lisitsync4goldsw: there is examples/meta/generator15:56
c4goldswI have it open.15:57
lisitsyngenerate.py is called by cmake15:57
lisitsynyou can find it being called in examples/meta/CMakeLists.txt15:57
c4goldswSo, I simply have to edit generate.py to support Requires(HAVE_NLOPT) , correct?15:58
lisitsync4goldsw: yes kind of15:58
lisitsync4goldsw: just a second doing some small research on that16:00
c4goldswlisitsyn I'm just poking around examples/meta/CMakeLists.txt16:00
lisitsync4goldsw: to speed up your development you should call make meta_examples16:01
lisitsynto not build anything but meta_examples16:01
lisitsynwhat I am currently curious of is whether HAVE_XXX is passed in the environment16:02
lisitsynwhen calling generate.py16:02
c4goldswThat is, if it can be interpreted?16:03
c4goldswAnd I'm able to build meta_examples without any issue.16:04
lisitsync4goldsw: yeah that's ok16:04
lisitsync4goldsw: check what generate.py does16:05
lisitsyntl;dr it reads all sg files and then parses them with some another module16:05
c4goldswStill going over it.16:06
c4goldswThe translate module, yes?16:07
lisitsynyeap16:08
c4goldswWhere is it located?16:09
c4goldswNevermind16:09
c4goldswI see it.16:09
c4goldswlisitsyn I'm looking through it now to see if it actually supports #ifdef or some meta-language equivalent.16:11
-!- BurningBear [~smuxi@223.3.169.194] has joined #shogun16:11
lisitsync4goldsw: ok I have a plan16:11
lisitsync4goldsw: first, you need to make generate.py know of these HAVE_SOMETHING variables16:12
lisitsynit involves some cmake hacking but shouldn't be really hard16:12
c4goldswThen what?16:12
lisitsync4goldsw: then in generate.py you could pass them into parse.py16:14
lisitsyn(when parsing a file)16:14
lisitsynand make it able to parse some requirement sting16:14
lisitsynstring*16:15
lisitsynit could look like '# requires: HAVE_SOMETHING' or anything like that16:15
lisitsynonce parsed this line it should check whether all required variables are here16:15
lisitsynthen simply ignore if not16:16
c4goldswAlright, thanks for the help!  As for the logistical side of things: do I create an issue for it, then submit a corresponding PR?16:16
-!- curiousguy13 [~curiousgu@120.59.100.246] has joined #shogun16:17
c4goldswlisitsyn I have to go now, so I'll just check the chatlogs.  Thank you!16:18
lisitsync4goldsw: you can go with PR directly I think16:19
-!- c4goldsw [8a33717d@gateway/web/cgi-irc/kiwiirc.com/ip.138.51.113.125] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]16:20
BurningBearhello, I am new user, I meet a program when I run  "CDenseFeatures<float64_t>* features= new CDenseFeatures<float64_t>(matrix);  int num_vectors = features->get_num_vectors();     features->get_feature_matrix(num_feat,num_vec);" the result of num_feat and num_vec are correct, but the result of num_vectors is wrong (a meaningless value like -869546368), is there something wrong?16:28
-!- curiousguy13 [~curiousgu@120.59.100.246] has quit [Ping timeout: 244 seconds]16:49
-!- BurningBear [~smuxi@223.3.169.194] has quit [Remote host closed the connection]17:09
-!- BurningBear [~smuxi@223.3.169.194] has joined #shogun17:10
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun17:18
-!- mode/#shogun [+o HeikoS] by ChanServ17:18
-!- curiousguy13 [~curiousgu@59.177.204.27] has joined #shogun17:19
shogun-buildbotbuild #585 of trusty - libshogun - viennacl is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun%20-%20viennacl/builds/585  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>17:28
@HeikoSshogun-buildbot: force build --brach=develop 'trusty - libshogun - viennacl'17:30
shogun-buildbotSomething bad happened (see logs)17:30
@HeikoSforce build --branch=develop 'trusty - libshogun - viennacl'17:30
shogun-buildbotbuild #586 of trusty - libshogun - viennacl is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun%20-%20viennacl/builds/58617:31
shogun-buildbotbuild #2833 of bsd1 - libshogun is complete: Failure [failed configure]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2833  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>17:34
shogun-buildbotbuild #1195 of FCRH - libshogun is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/FCRH%20-%20libshogun/builds/1195  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>17:37
@HeikoSwiking: ping17:37
@HeikoSlisitsyn: ping17:37
@HeikoSbesser82: ping17:37
@HeikoSneed some input on the BSD build.17:38
shogun-buildbotbuild #183 of FC22 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/FC22%20-%20libshogun/builds/183  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>17:38
-!- BurningBear [~smuxi@223.3.169.194] has quit [Ping timeout: 244 seconds]17:42
-!- BurningBear [~smuxi@223.3.169.194] has joined #shogun17:50
-!- travis-ci [~travis-ci@ec2-54-205-222-0.compute-1.amazonaws.com] has joined #shogun17:52
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/12029125817:52
-!- travis-ci [~travis-ci@ec2-54-205-222-0.compute-1.amazonaws.com] has left #shogun []17:52
-!- BurningBear [~smuxi@223.3.169.194] has quit [Remote host closed the connection]18:56
-!- travis-ci [~travis-ci@ec2-23-20-20-92.compute-1.amazonaws.com] has joined #shogun19:02
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/12029201219:02
-!- travis-ci [~travis-ci@ec2-23-20-20-92.compute-1.amazonaws.com] has left #shogun []19:02
-!- travis-ci [~travis-ci@ec2-54-159-167-2.compute-1.amazonaws.com] has joined #shogun19:33
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/12029225219:33
-!- travis-ci [~travis-ci@ec2-54-159-167-2.compute-1.amazonaws.com] has left #shogun []19:33
@besser82HeikoS, yes?19:43
-!- travis-ci [~travis-ci@ec2-54-205-222-0.compute-1.amazonaws.com] has joined #shogun20:12
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/12029767120:12
-!- travis-ci [~travis-ci@ec2-54-205-222-0.compute-1.amazonaws.com] has left #shogun []20:12
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.]20:53
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun20:53
-!- mode/#shogun [+o HeikoS] by ChanServ20:53
-!- travis-ci [~travis-ci@ec2-54-159-167-2.compute-1.amazonaws.com] has joined #shogun20:57
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/12030337520:57
-!- travis-ci [~travis-ci@ec2-54-159-167-2.compute-1.amazonaws.com] has left #shogun []20:57
@HeikoSbesser82: what is the best way to NOT break the build when guarded classes are not available in example? not meta, just in general.21:18
@HeikoSI can do21:18
@HeikoStry:21:18
@HeikoS    import XXX21:18
@HeikoSexcept ImportError:21:18
@HeikoS    exit()21:18
@HeikoSin python21:18
@HeikoSbut I dont want to do this in every example21:19
@HeikoSmaybe should I do it in Cmake?21:19
@HeikoSi.e. not execute test/example21:19
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.]21:19
-!- travis-ci [~travis-ci@ec2-54-205-222-0.compute-1.amazonaws.com] has joined #shogun21:24
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/12030794621:24
-!- travis-ci [~travis-ci@ec2-54-205-222-0.compute-1.amazonaws.com] has left #shogun []21:24
shogun-buildbotbuild #635 of debian wheezy - memcheck is complete: Failure [failed memory check]  Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/635  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>21:42
@besser82HeikoS, well…  one could load such a file before and probe it's exitcode…22:00
@besser82HeikoS, so you'd need to use exit(-127) or exit(1) instead of plain exit()22:00
@besser82HeikoS, and then eval in CMake to decide about running the test or not…22:01
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 250 seconds]22:07
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun22:24
-!- mode/#shogun [+o HeikoS] by ChanServ22:24
@HeikoSwiking: jo22:24
@HeikoSaround?22:24
@HeikoSlisitsyn: hi22:25
@HeikoSaround?22:25
lisitsynyes22:25
@HeikoShave a question22:25
@HeikoSso I have a guard that hides GPL code22:25
lisitsynyes22:26
@HeikoSand I tried in on OCAS22:26
@HeikoSand now some of the modular example fail since SVMOCAC is not defined anymore22:26
@HeikoShow to handle this case best?22:26
@HeikoSI can try-except in python22:26
@HeikoSbut that requires a manual solution for every target language22:26
@HeikoSwhat would be better is if these files would not be part of the test build22:26
@HeikoSbut that means I have to exclude them in CMake itself22:27
lisitsynok I get the problem22:27
@HeikoSI am leaning towards that22:27
@HeikoSbut dont know22:27
@HeikoSthis might also solve meta example problem22:27
@HeikoSlisitsyn: thoughts on it?22:29
lisitsynno instant solution22:29
lisitsynit is a bit tricky22:29
@HeikoSwhat about putting something in CMakeLists22:30
@HeikoSlike22:30
@HeikoSIF (USE_GPL_SHOGUN)22:30
@HeikoSexclude from list22:30
@HeikoSENDIF()22:30
@HeikoSbecause cmake globs for filenames22:30
lisitsynwhat about putting these requirements directly to the file22:30
@HeikoSwhich file?22:31
@HeikoSexample?22:31
@HeikoSdont know, I would prefer if example code is free of such technical mambo jambo....just confuses people22:31
lisitsynHeikoS: something.py22:31
lisitsynfirst line22:31
lisitsyn# requires HAVE_NLOPT22:31
@HeikoSand then read the with cmake you mean?22:31
lisitsynyeap22:31
@HeikoSthat might be an option22:32
@HeikoSBUt22:32
@HeikoSit requires opening the file and checking22:32
@HeikoSwhich is a bit of a mess22:32
@HeikoSespecially since we have like 10 target languages with different cmakelists22:32
lisitsynyeah22:32
@HeikoSsame for meta22:32
@HeikoSbut it seems there is no way around22:32
lisitsynHeikoS: I talked to a guy c4goldsw today22:33
lisitsynhe wanted to patch meta to support these checks22:34
lisitsynI suggested to add this thing to parser22:34
@HeikoSwhy the parser?22:34
lisitsynor generate.py22:34
@HeikoSI would prefer to avoid that22:34
@HeikoSthe parse should not care about the build in my eyes22:34
@HeikoSand not about semantics22:34
@HeikoSjust parse and translate22:35
@HeikoSthen have a separate module for the build22:35
@HeikoSor dont you agree?22:35
lisitsynyes true22:35
lisitsynthen generate.py22:35
lisitsyngenerate.py should care22:35
@HeikoSmmh22:35
@HeikoSnot so sure22:36
@HeikoSI would just generate these meta example listings and then not execute them22:36
lisitsynyeah probably it could be better..22:36
@HeikoSbecause otherwise they are again full of weird checks and all that22:36
@HeikoSit scares people22:36
@HeikoSand the meta examples were meant to not do that ;)22:36
@HeikoSall we want is the test build to be green22:37
@HeikoSI will try to modify the CmakeLists and then push to feature branch to discuss22:37
lisitsynHeikoS: but why generate examples that won't work (if they are for people)22:37
@HeikoSlisitsyn: it is just a listing22:38
@HeikoS"this is how it would look like"22:38
@HeikoSno good?22:38
@HeikoSit is the same as for the modular example that we have right now22:38
@HeikoSdont work if say LAPACK is not installed22:39
lisitsynyeah but it is because we don't generate them22:39
@HeikoSsure22:39
@HeikoSok then22:39
lisitsynI find some sense in exposing only supported examples22:39
lisitsynhowever users could ask why22:40
lisitsyn:D22:40
@HeikoSmaybe we can do that, and for normal examples I will just do by hand22:40
@HeikoS:D22:40
@HeikoS"WHERE IS MY EXAMPLE?!?!?!?!"22:40
lisitsynI can imagine someone asking where it is22:40
lisitsynyeah22:40
@HeikoSyeah22:40
@HeikoSmight be better if it was there22:40
@HeikoSand then user sees it, wants it, and then we tell them to install22:41
@HeikoSboth ways, user will ask us22:41
@HeikoSI mean all examples will be in cookbook22:41
@HeikoSso website in some way shows what is possible22:41
@HeikoSlisitsyn: so how would the generator realise when to not build an example?22:43
@HeikoSlisitsyn: ah man it's bad....mixing the generator with shogun semantics22:43
lisitsynyeah22:43
@HeikoSwont work on its own anymore22:43
@HeikoSthough it already doesnt due to the ctags22:44
lisitsynbad22:44
@HeikoSbut ctags is just a file22:44
@HeikoSlisitsyn: HOW TO SOLVE IT?22:44
@HeikoSTELL ME :D22:44
lisitsynwell ctags are optional now22:44
@HeikoSlisitsyn: yeah22:47
@HeikoSlisitsyn: how to to it then?22:47
@HeikoSgenerator.py gets say a list of "HAVE_FOO", "USE_BAR", etc22:47
@HeikoSlike class list?22:47
lisitsynyeah22:47
lisitsynI was thinking about that way22:47
@HeikoSand then scans the file for the macro?22:48
lisitsynyeah some 'requires'22:48
@HeikoSso you want to add that to the files?22:48
lisitsynyes, if it needs something special22:48
lisitsynwe put a comment22:48
lisitsyn# requires HAVE_NLOPT22:48
@HeikoSok sounds good22:48
@HeikoSand that guy was confident to do that?22:48
lisitsynHeikoS: yeah but lets see22:48
@HeikoSok then22:49
lisitsynHeikoS: I am thinking whether22:49
lisitsynit is ok to fail22:49
@HeikoSi will do things by hand for the existing modular examples22:49
@HeikoSok to fail?22:49
lisitsynI mean why not to fail more gracefully22:49
lisitsyncurrently we say no such class22:49
lisitsynbut maybe we can tell them why there is no such class22:49
@HeikoSlisitsyn: well22:50
@HeikoSlisitsyn: exceptions can do that22:50
@HeikoSbut thats all crap in shogun22:50
@HeikoSlisitsyn: if we macro guard them, then the class is not defined22:51
@HeikoSif one puts in the effort to write an alternative version of the class that throws runtime error in constructor, thats better22:51
@HeikoSBUT thats a lot of work22:51
@HeikoSlisitsyn: you know meta examples could in fact do something very simple22:52
@HeikoSlisitsyn: every language has a construct to check whether all used classes are available22:53
@HeikoSso we can just put a check after the import22:53
@HeikoSbut that only works for runtime languages22:53
@HeikoSlisitsyn: the cpp examples have to be compiled22:53
@HeikoSjava and csharp as well22:53
@HeikoSlisitsyn: lets think about it ... i will go now22:54
lisitsynok will thinki about it22:55
@HeikoSsee you!22:56
@HeikoSlisitsyn: once thing22:56
@HeikoSlisitsyn: did you see my feature branch for the USE_GPL_SHOGUN option?22:56
lisitsynnot yet22:56
@HeikoSlisitsyn: please tell me if there is anything you would do different22:56
@HeikoSotherwise, I will go ahead22:56
@HeikoSsince it needs to be done soon, before gsoc22:57
@HeikoSalso, if you know which classes are GPL only, send an email22:57
@HeikoSI think I will go for NLOPT next22:57
@HeikoSafter ocas is done22:57
lisitsynHeikoS: yeah some of my classes22:57
lisitsynwill send a list22:57
@HeikoSlisitsyn: only need the lowest of them....compile error will tell the rest22:58
@HeikoSthanks!22:58
@HeikoSsee you!22:58
lisitsynsee you22:58
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.]22:58
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun23:03
-!- mode/#shogun [+o HeikoS] by ChanServ23:03
-!- travis-ci [~travis-ci@ec2-54-159-167-2.compute-1.amazonaws.com] has joined #shogun23:36
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/12033384023:36
-!- travis-ci [~travis-ci@ec2-54-159-167-2.compute-1.amazonaws.com] has left #shogun []23:36
--- Log closed Sun Apr 03 00:00:57 2016
« Friday, 2016-04-01 Index Sunday, 2016-04-03 »

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