IRC logs of #shogun for Saturday, 2018-01-20

« Friday, 2018-01-19 Index Sunday, 2018-01-21 »
--- Log opened Sat Jan 20 00:00:27 2018
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 264 seconds]00:16
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has joined #shogun00:31
-!- mode/#shogun [+o HeikoS] by ChanServ00:31
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 256 seconds]01:01
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has joined #shogun01:19
-!- mode/#shogun [+o HeikoS] by ChanServ01:19
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/408901:23
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4089 karlnapf added label: "good first issue"01:23
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4089 karlnapf added label: "BUG"01:23
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4089 karlnapf removed label: "good first issue"01:24
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 248 seconds]01:31
-shogun-buildbot:#shogun- Build nightly trusty deb #94 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/26/builds/9403:01
-shogun-buildbot:#shogun- Build clang - undefined behaviour analysis #75 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/18/builds/7503:27
-!- witness [uid10044@gateway/web/irccloud.com/x-gmirmsbchqyhcwfw] has joined #shogun05:23
-!- saurabh7 [~saurabh@106.51.19.168] has joined #shogun05:35
-shogun-buildbot:#shogun- Build doxygen - nightly #75 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/12/builds/7507:06
-!- rrahul [~rahul@49.206.127.43] has joined #shogun07:46
-!- saurabh7 [~saurabh@106.51.19.168] has quit [Ping timeout: 264 seconds]08:17
-!- saurabh7 [~saurabh@106.51.19.168] has joined #shogun08:20
-!- rrahul [~rahul@49.206.127.43] has quit [Ping timeout: 268 seconds]11:16
-!- rrahul [~rahul@49.206.127.43] has joined #shogun11:23
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has joined #shogun11:57
-!- mode/#shogun [+o HeikoS] by ChanServ11:57
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 248 seconds]12:09
-!- saurabh7 [~saurabh@106.51.19.168] has quit [Quit: Ex-Chat]12:33
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4092 opened by shubham80814:16
-!- rrahul [~rahul@49.206.127.43] has left #shogun []15:11
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4087 synchronized by vinx1316:12
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4092 synchronized by shubham80816:13
-!- witness [uid10044@gateway/web/irccloud.com/x-gmirmsbchqyhcwfw] has quit [Quit: Connection closed for inactivity]16:22
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has joined #shogun17:50
-!- mode/#shogun [+o HeikoS] by ChanServ17:50
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4087 merged by karlnapf17:55
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/c5f2733dd73b1aa82e96dedbcb9e90223ebeb6d9 by karlnapf17:55
@HeikoSlisitsyn: you around?18:25
lisitsynHeikoS: partially :)18:26
@HeikoSlisitsyn: I would like to discuss the API stuff for a few mins18:26
@HeikoSsince I am writing up the GSoC project18:26
lisitsynok18:26
@HeikoSso all this API that we cam up with18:26
@HeikoShttps://github.com/shogun-toolbox/shogun/wiki/Hackathon-2017-base-api18:26
@HeikoSwhere does that live?18:26
@HeikoSI had in mind to write a new set of classes for that18:27
@HeikoSbut now I am wondering whether that is the best idea18:27
lisitsynI don't get18:27
lisitsynwhat do you mean?18:27
@HeikoSwhen we say we want to re-design the API18:27
@HeikoSwhat does that actually mean technically?18:27
@HeikoSI mean I have some picture in mind how it should look from interfaces18:28
lisitsynwell one task is to rename it to fit/transform, then the other task is ..18:28
@HeikoSbut how to we get from CMachine.h to that simple thing we discussed that just has fit/predict18:28
lisitsynlike redesign labels18:28
lisitsynI think we gradually change CMachine to work this way18:28
lisitsynno?18:28
@HeikoSokok18:28
@HeikoSand then merge the preprocessor and transformer18:29
@HeikoSbut we work with the existing classes18:29
lisitsynyes18:29
lisitsynI think so18:29
@HeikoScool18:29
@HeikoSand then are the common base classes?18:29
@HeikoSlike for "fit"18:29
@HeikoS?18:29
@HeikoSI guess yes18:29
lisitsynyes18:29
lisitsynwell it is just renaming18:30
@HeikoSok18:30
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4093 opened by vigsterkr18:30
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4093 vigsterkr added label: "SWIG"18:30
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4093 vigsterkr added label: "BUG"18:30
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/409318:30
lisitsynnow we call it train :)18:30
@HeikoSand what about stuff like "apply_locked"18:30
@HeikoSwe hide that from SWIG?18:30
lisitsynI hope so18:30
lisitsynI mean apply_locked is some kind of workaround iirc18:31
@HeikoSyes18:32
@HeikoSokok18:32
@HeikoSso this is renaming, re-designing/re-factoring, and cleaning18:32
@HeikoSand we start from some cute user stories18:32
Trixislol18:32
lisitsynuser stories is like what we had discussed?18:33
lisitsynlike how the code looks like18:33
lisitsynthen probably yes18:33
Trixisok so if i have stringcharfeatures, what exactly does get_feature_vector return, in python18:33
lisitsynwe can describe something in code and then make it possible18:33
lisitsynTrixis: what do you get?18:34
Trixiswell, id like to get it as a python string18:34
Trixisthe feature vector at the corresponding index18:34
lisitsynbut you get ['s', 'o', 'm', 'e', 't', 'h', 'i', 'n', 'g']?18:35
Trixis(i have a director kernel which essentially performs a look up for the precomputed result)18:35
Trixisi assume that's what im getting18:35
Trixishmm18:35
Trixis['' '' '' '' '' '' '' '' '' ''] even better18:38
lisitsyn:)18:39
Trixisfeatures = StringCharFeatures(id_data, ALPHANUM) is how i create the features18:39
Trixiswhere id_data is a python list of strings18:40
Trixisnot really sure why it's printing that18:50
@HeikoSTrixis: hey18:52
TrixisHeikoS: hi18:52
@HeikoSsorry I wasnt following ealier, whats the prob?18:52
Trixisi'm trying to work with StringFeatures, and i'd like to get the ft. returned by get_feature_vector() as a python string18:53
Trixishowever instead what gets returned is ['' '' '' '' '' '' '' '' '' ''] which doesnt look even remotely write (or thats what it prints as)18:54
@HeikoSI see18:54
@HeikoSthere must be some example somewhere ...18:54
@HeikoSlet me dig18:54
Trixisthx18:56
@HeikoSthere is a problem19:05
@HeikoSTrixis: at least with Python 319:05
Trixisoh19:05
Trixisany workarounds?19:05
@HeikoS.get_features() gives me a python list of strings19:06
@HeikoS(all of them)19:06
Trixisright19:06
Trixisyep19:07
Trixisthat works19:07
Trixisthx19:07
@HeikoSnw19:08
@HeikoSIll put up an issue19:08
@HeikoSwhat problem do you get if you ask for a vector?19:08
@HeikoSI get this: https://gist.github.com/karlnapf/b9122a880c9954cf795f20471b12363c19:08
@HeikoSon cloud.shogun.ml which runs on python319:08
Trixisi dont even get an error19:09
Trixisjust something malformed19:09
Trixisthe nb would print ['' '' '' '' '' '' '' '' '' '']19:10
Trixiswelp and shogun just crashed after i reran a few things, and in a bad way... guess ill have to restart interpreter19:10
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4094 karlnapf added label: "good first issue"19:12
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4094 karlnapf added label: "BUG"19:12
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4094 opened by karlnapf19:12
@HeikoSTrixis: could you run a script that does this with valgrind and see whether it complains?19:12
@HeikoSsee above for the bug report19:12
Trixisnah the crash is related to director errors19:13
Trixisthe notebook doesnt like if a director paragraph gets rerun19:13
Trixisso yeah, ill see if alls working once the whole thing reruns, which means recomputing all kernels. probably in two hrs19:16
@HeikoScan't you store intermediate results?19:21
@HeikoSlisitsyn: another question19:22
@HeikoS"help on parameters" ... what is that?19:22
TrixisHeikoS: i should yeah...19:22
lisitsynHeikoS: well we have some descriptions for parameters19:34
@HeikoSlisitsyn: where would those live?19:34
@HeikoSdoxygen?19:34
lisitsynbut I think it would be nice to have a way to read them19:34
@HeikoSor runtime?19:34
lisitsynprobably runtime19:34
@HeikoS@brief?19:34
@HeikoSlisitsyn: do tags have a "description" field?19:34
lisitsynnot yet but they should19:35
@HeikoSah, we should somehow integrate that with doxygen19:35
@HeikoSotherwise we have two19:35
@HeikoSruntime and doxygen19:35
@HeikoSwhich would suck19:35
lisitsynyeah probably (if possible)19:35
lisitsynyes19:35
@HeikoSwe already have @brief and cookbook which overlaps too much19:35
lisitsynso that's another task19:35
lisitsynintegrate this runtime thingy with doxygen19:35
lisitsynwell it is some .py script that enriches stuff with parameter descriptions19:35
lisitsyndoable19:35
@HeikoSkk19:36
@HeikoSlisitsyn: the python script modifies the c++ code?19:37
@HeikoSmmmmmmh19:37
lisitsynHeikoS: no, reads it19:37
@HeikoSand then?19:37
lisitsynputs into doxygen for interfaces19:37
@HeikoSit needs to be in the .so no?19:37
@HeikoSah19:37
@HeikoSyou mean doxygen is generated from tags19:37
lisitsynyes19:37
@HeikoSkk19:37
lisitsynthat should be good enough19:37
@HeikoSlisitsyn: soo19:38
@HeikoSshall we move all docs to the tags?19:38
@HeikoS@brief as well?19:38
@HeikoSah19:38
@HeikoSactually19:38
@HeikoSin python you can do already19:38
@HeikoShelp(CSVM)19:39
@HeikoSI think19:39
lisitsynyes19:39
lisitsynHeikoS: we already had some scrub_docstrings thingy19:40
@sukey[https://github.com/shogun-toolbox/shogun] Wiki page: GSoC_2018_project_usability edited by karlnapf19:40
@HeikoSkk19:40
lisitsynno idea what it does19:40
@HeikoSlisitsyn: could you read https://github.com/shogun-toolbox/shogun/wiki/GSoC_2018_project_usability19:40
@HeikoSand edit19:40
@HeikoStell me what I should edit19:40
@HeikoSscrape what you dont like19:40
@HeikoSI think plugins should maybe not be part of it19:41
@HeikoSaready enough19:41
lisitsynno, we don't need plugins yet19:41
lisitsynplugins would be trivial once we have all the required changes in19:41
lisitsynHeikoS: let me glance over it tomorrow19:42
lisitsynfirst look looks good but needs thorough reading :)19:43
@HeikoSill take care of that19:43
@HeikoSjust tell me if all points make sense, if you want to remove or add19:43
lisitsynHeikoS: yeah it looks good, but let me think and let you know19:45
lisitsyn:)19:45
lisitsynthinking fast and slow you know haha19:45
@HeikoSokok :)19:45
@HeikoSlisitsyn: and the array?19:45
@HeikoSlisitsyn: that is worrying me19:45
@HeikoSthat our integration tests are inactive19:46
lisitsynHeikoS: I stopped at the point of implementing clone19:46
@HeikoSlisitsyn: no need19:46
@HeikoSI can do19:46
@HeikoSand also clone is not yet merged19:46
@HeikoSonly needs equals19:46
@HeikoSbut in fact you can just put a stub19:46
@HeikoSI can implement the thing, just wanted your structure in there19:46
lisitsynuh then I can do it faster19:47
lisitsynbut clone is a bit funny there19:47
@HeikoSlet's start with equals19:47
@HeikoSwe have a working clone19:47
@HeikoSbut equals is broken19:47
lisitsynHeikoS: ok so let me find add_vector19:47
lisitsynand also register array using the new functions19:48
@HeikoSStringFeatures19:48
@HeikoSor DynamicObjectArray19:48
lisitsynHeikoS: what test should I run?19:48
@HeikoSthere is no test that fails19:48
@HeikoSneed to write one19:48
lisitsynfsck19:48
@HeikoSwhere you make two objects with different arrays and then expect_false on the equals19:48
@HeikoSactually19:48
@HeikoSwait a sec19:48
@HeikoSTEST(SGObject, DISABLED_equals_DynamicObjectArray_different)19:49
@HeikoSif you register the array in there this guy will fail19:49
lisitsynok sounds good19:50
@sukey[https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4071 closed by karlnapf19:54
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4075 merged by karlnapf19:54
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/de366755a53b4852f82c134896b10cbc99dd0cf4 by karlnapf19:54
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4095 opened by lisitsyn20:37
lisitsynHeikoS: we should replace SG_REF and SG_UNREF with functions21:12
lisitsynjust declare ref()/unref() somewhere21:12
shogitter(geektoni) lisitsyn: Some? ;)21:13
lisitsynoh geektoni from somewhere :)21:13
lisitsynI mean we have to unroll this macro a lot of times21:13
lisitsynwhile we can just use a global function21:13
shogitter(geektoni) yeah, I'm always here, though.21:14
shogitter(geektoni) watching21:14
shogitter(geektoni) but yes, I get your idea21:15
shogitter(geektoni) lisitsyn: does it improve compile time (or readability in general)? Or it is just to get rid of the macros?21:16
lisitsynI think it might improve compile time as well21:16
lisitsynjust a bit tho21:16
lisitsynoh it is so funny how that gitter works21:17
lisitsyndoes anybody know if we can also bridge slack in there?21:17
shogitter(geektoni) lisitsyn: no idea21:18
@wikingyes21:40
@wikingwe can21:40
@sukey[https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4095 merged by lisitsyn21:47
@sukey[https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/a66e13f0e10cd2e4645142c0ae87d93c992f211c by lisitsyn21:47
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 248 seconds]23:29
--- Log closed Sun Jan 21 00:00:28 2018
« Friday, 2018-01-19 Index Sunday, 2018-01-21 »

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