Open in new window / Try shogun cloud
--- Log opened Mon May 23 00:00:08 2016
-!- zxtx [~zxtx@107.170.86.177] has joined #shogun00:25
-!- mizari [~mizari@95-174-213-100.nts.su] has quit [Ping timeout: 260 seconds]00:32
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun []03:43
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun03:45
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun []03:47
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun03:49
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun []03:50
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun03:50
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun []03:51
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has joined #shogun03:56
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has left #shogun []03:58
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has joined #shogun03:58
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has left #shogun []03:58
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has joined #shogun03:58
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has left #shogun []03:59
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has joined #shogun04:00
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has left #shogun []04:00
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has joined #shogun04:01
-!- arianepaola_ [~ariane@unaffiliated/arianepaola] has quit [Quit: Leaving.]04:02
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun04:07
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun []04:17
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun04:17
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun []04:18
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun04:55
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun []04:56
-!- zxtx [~zxtx@107.170.86.177] has quit [Remote host closed the connection]05:02
-!- zxtx [~zxtx@sunnythinking.org] has joined #shogun05:03
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun05:21
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun []05:22
-!- mizari [~mizari@95-174-213-100.nts.su] has joined #shogun05:44
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun07:05
-!- mizari [~mizari@95-174-213-100.nts.su] has quit [Ping timeout: 240 seconds]07:11
shogun-buildbotbuild #239 of deb1 - libshogun - PR is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/23907:52
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting]09:00
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun09:00
@wikingshogun-buildbot: force build --branch=develop 'xenial - libshogun'09:00
shogun-buildbotThe build has been queued, I'll give a shout when it starts09:01
shogun-buildbotbuild #10 forced09:02
shogun-buildbotI'll give a shout when the build finishes09:02
shogun-buildbotbuild #10 of xenial - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/1009:10
@wikingshogun-buildbot: force build --branch=develop 'clang - thread analysis'09:10
shogun-buildbotbuild #4 forced09:10
shogun-buildbotI'll give a shout when the build finishes09:10
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun []09:10
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun09:22
shogun-buildbotbuild #4 of clang - thread analysis is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang%20-%20thread%20analysis/builds/409:31
@wikingshogun-buildbot: force build --branch=develop 'clang - undefined behaviour analysis'09:32
shogun-buildbotbuild #3 forced09:32
shogun-buildbotI'll give a shout when the build finishes09:32
shogun-buildbotbuild #3 of clang - undefined behaviour analysis is complete: Failure [failed compile]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang%20-%20undefined%20behaviour%20analysis/builds/309:55
-!- HeikoS [~heiko@82-132-246-6.dab.02.net] has joined #shogun10:07
-!- mode/#shogun [+o HeikoS] by ChanServ10:07
@HeikoSwiking: hi10:17
@HeikoSarianepaola: hi!10:18
@HeikoSGSoC begins! happy coding everyone :)10:18
-!- besser82 [~besser82@fedora/besser82] has joined #shogun10:18
-!- mode/#shogun [+o besser82] by ChanServ10:18
@HeikoSwiking: the cookbook travis preview fails10:19
@HeikoShttp://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/237/steps/cookbook/logs/stdio10:19
@HeikoSthink sphinx is missing10:19
@HeikoSI guess that is since it moved to the AWS10:19
@wikingHeikoS: not anymore10:19
@wikingnoup10:19
@wikingit has nothing to do with it10:19
@wikingHeikoS: check now: https://github.com/shogun-toolbox/shogun/pull/319710:23
@wikinghttp://buildbot.shogun-toolbox.org/cookbook_pr/b326aa9fad9860a184a1d4059d83fe032419f001/10:23
@HeikoSoh now worked10:23
@HeikoSno idea what that was10:23
@wikingyep fixed it10:23
@wikingthere was a little error10:23
@HeikoSok thanks10:23
@wikingi wanna setup gpu worker10:24
@HeikoScool10:24
@HeikoSany news from gpl friend?10:24
@wikingyeah10:24
@HeikoScool!10:24
@wikinghe told me that it's ok10:24
@wikingso we can keep the old repo10:24
@wikingbut we'll have to remove the code10:24
@wiking(gpl code)10:25
@wikingto a separate repo10:25
@wikingwhich should be fine10:25
@HeikoSok then cool10:25
@HeikoShey that is great10:25
@HeikoSshall we do before release?10:25
@HeikoShow would you approach?10:26
@HeikoSin terms of build managment10:26
@wikingbuild managment?10:34
@wikingwhat do you mean?10:34
@wikingit's just an extra submodule10:34
@wikingand that's it10:34
-!- HeikoS [~heiko@82-132-246-6.dab.02.net] has quit [Ping timeout: 250 seconds]10:35
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 240 seconds]10:39
-!- sanuj [~sanuj@117.203.19.196] has joined #shogun10:57
sanujlisitsyn, wazzup11:06
-!- sanuj [~sanuj@117.203.19.196] has quit [Ping timeout: 244 seconds]11:25
-!- sanuj [~sanuj@59.91.134.255] has joined #shogun11:31
lisitsynsanuj: hey11:59
sanujlisitsyn, i start with adding tags to shogun?11:59
sanujor plugins?11:59
-!- besser82 [~besser82@fedora/besser82] has joined #shogun12:02
-!- mode/#shogun [+o besser82] by ChanServ12:03
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 244 seconds]12:19
-!- besser82 [~besser82@fedora/besser82] has joined #shogun12:31
-!- mode/#shogun [+o besser82] by ChanServ12:31
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 276 seconds]12:55
-!- leagoetz [~leagoetz@nat-212-197.internal.eduroam.ucl.ac.uk] has joined #shogun12:55
leagoetzarianepaola: am I missing your cookbook contributions somewhere? can't find them13:04
lisitsynhey leagoetz13:05
sanujleagoetz, mine is here: https://github.com/shogun-toolbox/shogun/pull/319713:07
sanuj:D13:07
sanujlisitsyn, adding tags to shogun's SGObject in a different branc13:08
sanujbranch13:08
lisitsynsanuj: yeah ok :)13:08
leagoetzsanuj: yes, great, I've seen your email already13:12
leagoetzin the process of updating the Wiki13:12
sanujcool13:13
arianepaolaHello everyone :-)13:15
leagoetzhey Ariane13:17
arianepaolaHi leagoetz, I will add a cookbook contribution shortly13:18
leagoetzhow is it going?13:18
leagoetzah, cool!13:18
leagoetzbecause GSoC is starting ... and we need everything ready :)13:18
arianepaola:-)13:18
leagoetzSaurabh7: I also need some cooking from you ...13:20
leagoetzalso your timeline says 'cookbook' for the final week13:22
leagoetzbut I hope it should be obvious to everyone now (given the amount of hassling by me ;P) how important it is to us that you contribute cookbook pages THROUGHOUT GSoC13:22
leagoetzjust a reminder ^^13:23
sanujleagoetz, i was wondering if it is possible to link the important meetings irc logs in the gsoc wiki section13:25
sanujwill be helpful13:25
leagoetzsure that should be possible13:51
leagoetzcertainly from now on13:51
leagoetzlike for our meeting tomorrow13:51
sanujthanks :)13:51
sanujwiking, there?13:54
-!- sanuj [~sanuj@59.91.134.255] has quit [Ping timeout: 240 seconds]14:03
-!- leagoetz_ [~leagoetz@nat-212-197.internal.eduroam.ucl.ac.uk] has joined #shogun14:08
-!- leagoetz [~leagoetz@nat-212-197.internal.eduroam.ucl.ac.uk] has quit [Ping timeout: 260 seconds]14:11
-!- sanuj [~sanuj@59.91.134.255] has joined #shogun14:56
-!- leagoetz_ [~leagoetz@nat-212-197.internal.eduroam.ucl.ac.uk] has quit []15:06
-!- leagoetz [~leagoetz@nat-212-197.internal.eduroam.ucl.ac.uk] has joined #shogun15:07
-!- leagoetz [~leagoetz@nat-212-197.internal.eduroam.ucl.ac.uk] has quit []15:19
-!- travis-ci [~travis-ci@ec2-54-196-57-170.compute-1.amazonaws.com] has joined #shogun15:31
travis-ciit's lambday'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/13227764615:31
-!- travis-ci [~travis-ci@ec2-54-196-57-170.compute-1.amazonaws.com] has left #shogun []15:31
-!- Saurabh7_ [~Saurabh7@1.39.15.60] has joined #shogun15:44
-!- sanuj [~sanuj@59.91.134.255] has quit [Ping timeout: 276 seconds]15:53
-!- Saurabh7_ [~Saurabh7@1.39.15.60] has quit [Quit: Leaving]15:54
-!- HeikoS [~heiko@82-132-231-192.dab.02.net] has joined #shogun15:58
-!- mode/#shogun [+o HeikoS] by ChanServ15:58
@wikingHeikoS: oooooooolala15:58
@HeikoSwiking: jojo15:59
@wikingarianepaola: pingooooooooooo15:59
@HeikoSsorry for earlier15:59
@HeikoStrain wifi was crap15:59
@HeikoSand my phone died15:59
@HeikoSwiking: so where are all the hard coding students? :)15:59
@HeikoSwiking: i migh thave to leave soon15:59
@wikinghahaha yeah dunno15:59
@HeikoSwiking: just was asking about release and gpl branch15:59
@wikingyeah i'm gonna finish those16:00
@HeikoSwhat sthe best approach in your eyes?16:00
@wikingas well as the remaining16:00
@HeikoSand how can I help16:00
@wikingtickets for the release16:00
@wikingand cut a release asap16:00
@HeikoSkk16:00
@wikingbtw managed to get a gpu instance working with shogun16:00
@wikingonly sad part is16:00
@HeikoSI think the GPL thing is most work, no?16:00
@HeikoSwiking: cool16:00
@wikingthat we only utilise OPENCL backend for viennacl16:00
@wikingmmm that's just simple copy paste shit16:01
@wikingnothing big there16:01
@HeikoSwiking: yeah but time16:01
@HeikoSok meta examples failing on travis16:01
@HeikoSnot sure since when and why16:01
@wikingyeah do something about that plz16:01
@HeikoSwill do16:01
@wikingthnx16:01
@wikinghttp://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/2865/steps/test%20java%20modular/logs/stdio16:02
@wikingthey are not failing here right16:02
@wiking?16:02
@HeikoSyep16:02
@HeikoSseems travis16:02
@HeikoSand its only the compiled ones16:03
@HeikoSjava and csharp16:03
-!- travis-ci [~travis-ci@ec2-54-147-204-36.compute-1.amazonaws.com] has joined #shogun16:05
travis-ciit's Soumyajit De'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/13227859316:05
-!- travis-ci [~travis-ci@ec2-54-147-204-36.compute-1.amazonaws.com] has left #shogun []16:05
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has joined #shogun16:07
-!- OXPHOS [9d8b131c@gateway/web/freenode/ip.157.139.19.28] has joined #shogun16:14
@HeikoSOXPHOS: hihi16:14
OXPHOSHeikoS: Greetings!16:14
@HeikoSOXPHOS: how are things?16:14
@HeikoSwiking: whats the java version on buildbot?16:15
@HeikoStravis is 1.8.016:15
OXPHOSHeikoS: got something to discuss with you16:15
@HeikoSOXPHOS: good!16:15
@wikingHeikoS: it's openjdk16:15
@wikingand not oracle jdk16:15
@HeikoSI see16:15
@wiking1.7.016:16
@HeikoSwiking: ah it was sanuj16:16
@HeikoShttps://travis-ci.org/shogun-toolbox/shogun/jobs/132213321#L258516:16
@HeikoStype error16:16
@HeikoSmight need to change the grammar16:16
OXPHOSHeikoS: start from the smallest? cookbook - I have modshogun error from the very beginning when I try to run examples16:17
@HeikoSOXPHOS: locally?16:18
@HeikoSin all languages?16:18
OXPHOSyes. even with the python example in undocumented16:18
OXPHOSyes.16:18
@HeikoSkk16:18
@HeikoSso in order to run e.g. a python example16:18
@HeikoSyou need to set two environment variables:16:18
@HeikoSPYTHONPATH and LD_LIBRARY_PATH16:18
@HeikoS(or run make install)16:18
@HeikoSbut better set the variables16:18
@HeikoSso that python sees your shogun build16:18
@HeikoSOXPHOS: the same is true for all other modular targets16:19
@HeikoSOXPHOS: know what I am talking about?16:19
OXPHOSaha..just to show python where things are?16:19
@HeikoSexactly16:19
OXPHOSwhat is LD_LIBRARY_PATH (LD)?16:19
@HeikoSexample:16:20
@HeikoSecho "Adding shogun_develop to python and ld lib path"16:20
@HeikoSexport PYTHONPATH=/home/heiko/git/shogun/shogun_develop/shogun/build/src/interfaces/python_modular:$PYTHONPATH16:20
@HeikoSexport LD_LIBRARY_PATH=/home/heiko/git/shogun/shogun_develop/shogun/build/src/shogun:$LD_LIBRARY_PATH16:20
@HeikoSOXPHOS: this is where the system looks for shared libraries16:20
@HeikoSin this case: libshogun.so16:20
@HeikoSOXPHOS: otherwise, you will get an error when running things16:20
OXPHOSokay thx. lemme try.16:21
@HeikoSOXPHOS: should be quick to fix16:21
@HeikoSlet me know if you need help16:21
@HeikoSor ask lisitsyn or wiking16:21
OXPHOSshould these be in the readme or this is too basic XD16:21
@HeikoSthey are16:21
@HeikoSif you feel like they need to be more upfront, please send a patch, very useful! :)16:21
OXPHOSsure : )16:22
OXPHOSHeikoS: Serialization: https://gist.github.com/OXPHOS/c4a11c1d378aa3d9911744300feb79b516:22
@HeikoSOXPHOS: so what is your plan for this week?16:22
OXPHOSstart with serialization since I kinda started?16:22
@HeikoSOXPHOS: cereal uses smart pointers16:23
OXPHOSI tried integrate cereal in SGVector and KMeans.16:23
@HeikoSso might depend on that transition16:23
@HeikoScool16:23
@HeikoSa working prototype anywhere?16:23
@HeikoSotherwise that is first step16:23
@HeikoSbut fine to start with this if you feel like16:23
@HeikoSwiking: ^ ok with you as well?16:24
-!- Saurabh7_ [~Saurabh7@117.248.210.124] has joined #shogun16:24
OXPHOSIt worked. I added some codes in SGVecotr and KMeans. I don't know how to highlight them other than pull request. But I put the patch in gist16:24
@HeikoSOXPHOS: I see you use std::vector to wrap shogun SGVector16:24
@HeikoSOXPHOS: we can do this in a feature branch16:24
@HeikoSshall I create one?16:24
@HeikoSSaurabh7 hi!16:25
OXPHOSsure thx.16:25
Saurabh7_yo hello16:25
@HeikoSI gotta leave soon, but will be back from train a bit later16:25
@HeikoSSaurabh7_ shall we talk tomorrow before the GSOC meeting?16:26
@HeikoSOXPHOS: same for you, shall we talk a bit then?16:26
Saurabh7_HeikoS, ok i ll be around16:26
@HeikoSboth, can you plan the week a bit in terms of what you want to do exactly etc?16:26
OXPHOSHeikoS: Sure!16:26
@HeikoSthen we can dicsuss tomorrow16:26
@HeikoSalso maybe some first concrete issues that we need to solve16:26
OXPHOSkk16:26
@HeikoSOXPHOS: I created a feature branch fork from develop16:27
@HeikoSyou can send PRs against that now16:27
OXPHOSHeikoS: cool thanks!16:27
@HeikoShttps://github.com/shogun-toolbox/shogun/tree/feature/cereal16:28
@HeikoSOXPHOS: feature brnaches are a bit more easy to work with16:28
@HeikoStravis runs on the PRs16:28
@HeikoSbut the buildbot doesnt16:28
@HeikoSso one can prototype more quickly16:29
@HeikoSbut then merging is more work16:29
@HeikoSOXPHOS: can you send a PR for the working SGVector?16:29
@HeikoSthen we can discuss over that tomorrow16:29
@HeikoSSaurabh7_: so about the indexing16:29
@HeikoSI am not sure I get the problem16:29
@HeikoScan you send an email?=16:29
OXPHOSsure. but i currently have cereal at local. should i just send the whole library for now?16:29
@HeikoSgotta run now, see you16:30
@HeikoSOXPHOS: mmh16:30
@HeikoSask wiking16:30
-!- Saurabh7__ [~Saurabh7@1.39.98.181] has joined #shogun16:30
@HeikoShe will have an opinion16:30
@HeikoSbye and see you later16:30
-!- HeikoS [~heiko@82-132-231-192.dab.02.net] has quit [Quit: Leaving.]16:30
-!- Saurabh7__ [~Saurabh7@1.39.98.181] has quit [Client Quit]16:30
-!- Saurabh7_ [~Saurabh7@117.248.210.124] has quit [Ping timeout: 276 seconds]16:31
OXPHOSwiking: there?16:31
@wikingOXPHOS: here here16:31
-!- Saurabh7_ [~Saurabh7@1.39.98.181] has joined #shogun16:32
OXPHOSwiking: Hello : ) I have some cereal prototype [w/o smart pointer] working for SGVector and KMeans now16:32
OXPHOShttps://gist.github.com/OXPHOS/c4a11c1d378aa3d9911744300feb79b516:32
@wikingkk lemme se16:34
@wikinge16:34
@wikingwhy dont you create a pr? :D16:34
OXPHOSwiking: I'm gonna pull a PR in feature branch. Can I dump the whole external/cereal library in by now?16:35
OXPHOSNot sure how travis will find cereal16:35
@wikingyeah16:35
@wikingwe can work on that16:35
OXPHOSwiking: okay. Then lemme do it now.16:37
@wikingcool16:37
-!- c4goldsw [c1a99ae1@gateway/web/cgi-irc/kiwiirc.com/ip.193.169.154.225] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]16:42
-!- travis-ci [~travis-ci@ec2-54-147-204-36.compute-1.amazonaws.com] has joined #shogun16:47
travis-ciit's lambday'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/13229753416:47
-!- travis-ci [~travis-ci@ec2-54-147-204-36.compute-1.amazonaws.com] has left #shogun []16:47
OXPHOSIs github down or just me?17:10
Saurabh7_works here17:11
@wikingcheck status.github.com17:13
@wikingyes it's down17:14
-!- hessian [547a0c25@gateway/web/freenode/ip.84.122.12.37] has joined #shogun17:16
hessianhello, tried to install shogun-toolbox on mac os but cannot get all the dependencies right. Is there a way to install it without building it from scratch ? Thanks.17:17
OXPHOSaha17:17
-!- travis-ci [~travis-ci@ec2-54-196-57-170.compute-1.amazonaws.com] has joined #shogun17:20
travis-ciit's lambday'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/13229877317:20
-!- travis-ci [~travis-ci@ec2-54-196-57-170.compute-1.amazonaws.com] has left #shogun []17:20
-!- Saurabh7_ [~Saurabh7@1.39.98.181] has quit [Ping timeout: 244 seconds]17:24
-!- hessian [547a0c25@gateway/web/freenode/ip.84.122.12.37] has quit [Ping timeout: 250 seconds]17:36
-!- Saurabh7_ [~Saurabh7@59.88.174.17] has joined #shogun17:45
-!- HeikoS [~heiko@176.12.107.140] has joined #shogun17:55
-!- mode/#shogun [+o HeikoS] by ChanServ17:55
@HeikoSSaurabh7_:17:55
@HeikoShi17:56
@HeikoSin train now17:56
-!- HeikoS [~heiko@176.12.107.140] has quit [Remote host closed the connection]17:56
-!- HeikoS [~heiko@176.12.107.140] has joined #shogun17:56
-!- mode/#shogun [+o HeikoS] by ChanServ17:56
@HeikoSSaurabh717:57
@HeikoSSaurabh7_:17:57
@HeikoSso about the indexing17:57
@HeikoSwe definitely dont want to change method signatures everywhere17:57
@HeikoSwiking: hi17:58
@HeikoSbut it should not be necessary17:58
@HeikoSSaurabh7 around?17:58
@HeikoSOXPHOS: you still around?18:01
OXPHOSHeikoS: yes18:01
@HeikoSOXPHOS: so we can discuss a bit now18:02
@HeikoStwo points I would like to talk about18:02
@HeikoSplan for this week and cereal prototype18:02
@HeikoSplan for this week: cookbook pages and cereal plan.18:03
@HeikoScereal prototype: PR against feature branch18:04
@HeikoSthoughts about those things?18:04
OXPHOSsounds good18:04
@HeikoSOXPHOS: I mean I wanted to get your thoughts and ideas on what these points would be18:05
@HeikoSOXPHOS: not just tell you ;)18:05
OXPHOShaha lemme grab my notes18:05
Saurabh7_HeikoS, hi18:05
Saurabh7_ machine->add_subset(train_inds, str(fold_id))       machine->train()18:05
@HeikoSSaurabh7 hi18:05
Saurabh7_is this thread safe thought18:05
Saurabh7_though18:05
@HeikoSSaurabh7_: I think I dont get why the method signatures need to be changed18:06
@HeikoSwhen there is a default parameter18:06
@HeikoSSaurabh7_: so essentially each identifier manages a difference subset18:06
@HeikoSand putting a particular one in doesnt affect the others18:06
Saurabh7_yes but  how would train know which to access18:07
OXPHOS1. test save/load methods in kmeans, json, xml, binary; 2. add save/load methods in SGString, SGMatrix etc. 3. add cereal.cmake18:07
OXPHOSI have 4 5 6 7. But honestly so far I don't feel smart pointer necessary18:07
OXPHOSand I tried smart pointer, failed..18:07
Saurabh7_ machine->add_subset (in thread 1)  )18:07
Saurabh7_some compuataion in 118:07
Saurabh7_ machine->add_subset(train_inds, str(fold_id)) in thread 218:08
Saurabh7_back to 1 ?18:08
@HeikoSSaurabh7_: get it now you mean inside18:08
@HeikoSOXPHOS: ok that is good18:08
@HeikoSOXPHOS: so I fear that allocating empty SGObject instances will be hard18:08
@HeikoSwhat about going ahead with a working version of SGVector18:08
@HeikoSand then a CSGOBject18:09
@HeikoSnot KMeans, as it has complex sub-types18:09
@HeikoSan easier one for now18:09
@HeikoSSaurabh7_: good point, any ideas to resolve? we dont want to change all that code18:09
@HeikoSSaurabh7_: but it might be impossible without in fact18:09
Saurabh7_yes i thought about it18:10
Saurabh7_cant see other way18:10
Saurabh7_other than have different subset stack on copies18:10
@HeikoSneed to tell the model somehow in which thread it lives with this approach18:10
Saurabh7_yes for taht each acces pattern would need the identifier18:10
@HeikoSSaurabh7_: you know how other libs do this?18:11
@HeikoSstore copies of data in memory?18:11
@HeikoSis that the only way?18:11
Saurabh7_dont know in detail18:11
Saurabh7_scikit uses joblib18:11
@HeikoSsklearn has a new version18:11
@HeikoSyeah thats the one18:11
@wikingHeikoS: ?18:11
@HeikoSwiking: discussing thread safe subset18:12
@HeikoSwiking: one way is passing an identifier with each subset18:12
@HeikoSbut then we need this idenfitier when asking for data from the thread18:12
@HeikoSwiking: any thoughts on how to do it without changing existing code?18:12
@wikingmmmm18:13
@HeikoSwiking: without storing copies of memory18:13
@wikingHeikoS: well either pointers to the data18:13
@HeikoSwiking: but that also requires the thread to know which fraction of data is accessed18:13
@wiking?18:14
@wikingwhat do you mean fraction of the data18:14
@wikinganyhow there's another option18:14
@HeikoSfor xvalidation e.g.18:14
@wikingis to have thread safe18:14
@wikingiterators18:14
@wikingbut first18:14
@wikingwhat you wanna actually fix here18:14
@HeikoSwiking: yeah the iterator would need to be passed to the algorithm18:14
@wikingdo we wanna solve the problem of a r/w view18:14
@HeikoSwe want multicore xvalidation18:14
@wikingor a read-only view18:14
@wiking?18:14
@HeikoSread only atm18:14
@wikingah then i think18:14
@wikingintroducing a const_iterator for the feature18:15
@wikingwould be the best idea18:15
@HeikoSkk18:15
@wikingor?18:15
@HeikoSso lets think about it18:15
@HeikoShave svm18:15
@HeikoSthat accepts a CFeatures instance18:15
@HeikoSwe make this one an iterator (somehow)18:15
@HeikoSand then pass different ones to each machine18:15
@HeikoSin fact18:16
@HeikoSno iterator necessary18:16
@HeikoSjust have multiple CFeatures instances18:16
@HeikoSeach with different subset18:16
@HeikoSpointing to same data18:16
@HeikoSthis way, the CMachine instance doesnt just interacts with that18:16
@HeikoSright?18:16
@wikingwell yeah18:17
@wikingyou dont need an actual itertor18:17
@wikingi just wanted to give an idea with the concept18:17
-!- Saurabh7__ [~Saurabh7@1.39.98.181] has joined #shogun18:17
@HeikoSwiking: yeah, just copies of CFeatures18:17
@HeikoSsure sure18:17
@wikingyeah but not the data18:17
@HeikoSyes, same data18:17
@HeikoSall const methods18:17
@wikingyep yep18:17
@HeikoSlike CSubsetFeatures18:17
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has joined #shogun18:17
-!- mode/#shogun [+o lambday] by ChanServ18:17
@HeikoSlambday: jojo18:17
@lambdayHeikoS: hola18:18
@HeikoSSaurabh7__: so18:18
@HeikoSget the idea?18:18
OXPHOSHeikoS: [not sure when to interrupt] Currently we only serialize parameters [in class] individually. thinking of serializing one class as a whole?18:18
Saurabh7__yeah sorry i missed some part18:18
@HeikoSOXPHOS: anytime!18:18
@wikingOXPHOS: got a PR in email18:18
@wikingbut 404 on github18:18
@wiking:)18:18
OXPHOSNever mind it is not the latest version..18:18
-!- Saurabh7_ [~Saurabh7@59.88.174.17] has quit [Ping timeout: 240 seconds]18:18
@wikingok works now18:18
shogun-buildbotbuild #241 of deb1 - libshogun - PR is complete: Failure [failed git]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/241  blamelist: OXPHOS18:19
@HeikoSOXPHOS: not sure what you mean18:19
OXPHOSwiking: some commits are not there yet18:19
@wikingno worries18:19
@wikingchecking stuff18:19
@lambdayOXPHOS: how's it going?18:19
@HeikoSSaurabh7__: so, the x-validation class would, rather than adding subsets to features, create multiple CFeatures objects, each pointing to the same data, but with different subsets applied to them18:19
Saurabh7__HeikoS,  so sort of shallow copy ?18:20
@HeikoSwiking: same for your bagging machine18:20
@HeikoSrather than putting subsets, create new feature objects18:20
@HeikoSor does it write?18:20
OXPHOSHeikoS: im thinking: kmeans.parameters.load() loads everything. So SGObject is created by default constructor or whatever before loading18:20
OXPHOSlambday: Hey!18:20
@lambdaywiking: hey what was the flag to check whether we have openmp or not? HAVE_OPENMP or something? you remember?18:20
@wikinglambday: in code or in cmake?18:21
@lambdaywiking: in code18:21
@wikinglambday: no thing like this is checked afaik18:21
@lambdaybasically I want to include omp.h, if we have openmp..18:21
@wikingbut imo #pragma -s are ignored18:21
OXPHOSHeikoS: do you mean we can load(kmeans). Then we create an instance of kmeans with saved parameters?18:21
@lambdayyeah for pragmas it is alright18:21
@wikingmm that case introduce18:22
@wikinga define in config.h.in18:22
@HeikoSSaurabh7__: no18:22
@HeikoSSaurabh7__:  different subset on each of them18:22
@HeikoSonly the same data18:22
@lambdaywiking: alright.. yeah was checking for that HAVE_* thing but couldn't find it anywhere.. thought if it's there you'd know :)18:22
@HeikoSso shallow copy + subset18:22
@lambdayOXPHOS: working on cerealization?18:23
@wikingOXPHOS: ok so this is noob18:23
@wikingOXPHOS: close the pr18:23
@wikingbeause you've imported actually cereal to the codebase18:23
@wikingfuck github is broken18:23
@HeikoSOXPHOS: sure, but if you use default constructor, then the memory for its members is not allocated ... that can be tricky. But will see, just go ahead for now18:23
@HeikoSOXPHOS: any plan on which cookbook pages you want to add?18:24
@HeikoSOXPHOS: and any update on the modular bindings issue you had?18:24
OXPHOSlambday: yes: ) had some work on linalg but not posted yet18:24
@wikingOXPHOS: ok so lets talk about this commit a bit :)18:24
@wikingi mean PR18:24
OXPHOSwiking: sure yes there's still problem18:25
@wikingOXPHOS: ok so first of all this is wrong18:25
@wikingbecause you import the whole cereal codebase18:25
@wikinginto shogun codebase18:25
@wikingdo not do that18:25
OXPHOSHeikoS: not yet..was making cereal to make sure it works18:25
@wikingOXPHOS: second: you started using STL in headers... please avoid that18:25
@wikingmeaning why do you need std::vector18:25
@wikingto serialize an array of TYPENAME18:26
@wikingit's cool to the the thing with cereal18:26
@wikingthe way eigen is dont18:26
@wiking*done18:26
@wikingi.e. add it as an external dependency if there's not cereal on the system18:26
@wikingbut do not git add the cereal codebase to the shogun source18:27
@wiking(see your pr)18:27
OXPHOSwiking: yes i know.  1. for external dependency i need a cereal.cmake right? or there's other ways18:27
@wikingOXPHOS: you should actually name it FindCereal.cmake18:28
@wikingjust to be consistent with cmake modules18:28
OXPHOSsure. and the vector -18:28
@wikingdid you use this18:28
@wikinghttps://github.com/DURAARK/e57Extract/blob/master/cmake/modules/FindCereal.cmake18:28
@wiking?18:28
OXPHOSaha no. didnt' try18:29
@wikingor i guess you used this18:29
@wikinghttps://github.com/kc5nra/sporkel/blob/master/ext/cereal.cmake18:29
@wikingbeause this assumes18:29
@wikingthat you have ceraal in your source tree18:29
@wikinganyhow18:29
@wikingtry to look around for FindCereal18:29
@wikinguse stuff that is opensource18:29
OXPHOSokay18:30
@wikingcopying opensource stuff is not wrong18:30
@wikingit's actually encouraged18:30
@wiking:)18:30
@lambdayhaha ^18:30
@lambday:D18:30
@wikinglambday: ?18:30
OXPHOS: ) so the vector - it was a T* vector that needs to be serialized18:30
@wikinglambday: dont you agree?18:30
@lambdaywiking: of course.. unless it's GPL18:30
@wikingOXPHOS: yeah i know what needed to be done18:31
OXPHOSI tried to convert it to a smart pointer, which gave error18:31
@wikingyep18:31
@wikingbut we can 'teach' cereal18:31
@wikinghow to serialize a T*18:31
@wikingno need for copy ctrs18:31
@wikingso first close this pr18:31
@wikingstart a new one18:31
@wikingand add there as the beginning18:31
@wikingthe cmake related stuff18:32
@wikingonce we have that18:32
@wikingwe'll switch to PoC stuff for objects18:32
@wikingok?18:32
OXPHOSsure18:32
@wikingtnx18:32
OXPHOSwas thinking modifying from that json.cmake18:33
@wikingthe stuff you wrote in the md for cmake part18:33
@wikingwas good18:33
@wikingthat's fine18:33
@wikingOXPHOS: but again18:33
@wikingthere's that one i found on github18:33
@wiking(FindCereal.cmake)18:33
@wikingthat will serve you good18:33
OXPHOSsure I didn't realize we can find such stuff. Thx!18:34
@wikingyeah yeah18:34
@wikingeasy18:34
@wikinggithub is your friend18:34
@wiking(when it works)18:34
@wikingjust search for findCereal18:34
OXPHOShaha18:34
@wiking:)18:34
@wikingjust ping me when you have the pr read18:35
@wiking*ready18:35
OXPHOSwiking: np. thx!18:35
@wikingnw18:36
OXPHOSHeikoS: for the cookbook - distance_braycurtis, canberra and canberraword? random pick18:37
@HeikoSOXPHOS: try algoerithms first18:37
@HeikoSbefore these "tools" such as distance are covered18:37
@HeikoSall classifiers and regression methods18:37
@HeikoSall clustering18:37
@HeikoSall preprocessing18:37
@HeikoSthen distance and kernel18:37
@HeikoSwiking: about thread safety18:38
@HeikoSjust talking with Saurabh7__18:38
@wikingyep18:38
@HeikoSso the access methods of features18:38
@HeikoSwe should ensure read-only18:38
@HeikoSon some of them18:38
@wikingconst18:38
@HeikoSand hacking mutex for the ones that modify18:38
@wikingc++ is great for that18:38
@HeikoSyeah that is what I was thinking18:38
@wikingHeikoS: atomic is fine18:38
@HeikoSso need to clean up interface18:38
@wikingno need for mutex18:38
@HeikoSwiking: true18:38
@HeikoSwiking: so CFetaures needs const read methods18:39
@wikingyep18:39
@wikingbest would be18:39
@wikingoperators18:39
@wiking;)18:39
@wikingif you know what i mean18:39
@HeikoSso basically whenever we do things in parallel, we need to make sure all operators called from within are const18:39
OXPHOSHeikoS: qda, ssk, clustering_hierarchical_modular?18:39
@HeikoSwiking: yeah sure, I mean we can go on the fly -- all algos that are parallelised, we make sure to make all calls const18:40
@wikingoperator[](index_t idx) const;18:40
@HeikoSSaurabh7__:  saw that ^18:40
@HeikoSOXPHOS: perfect :)18:40
Saurabh7__HeikoS, yep18:40
@HeikoSSaurabh7__: cool18:40
@HeikoSso this way we clean up things a bit on the fly18:40
arianepaola:-)18:40
@HeikoSand allow the optimizer to be more efficient18:40
@wikingand to have cleaner code18:41
@wikinginstead of18:41
@wikingcfeatures->get_feature_vector(2)18:41
@wiking:P18:41
@wikingcfeatures[2]18:41
@HeikoSSaurabh7__: lets start with this x-validation then for week 118:41
@wiking;P18:41
@HeikoSwiking: yeah actually good point18:41
@wikingi mean for god sakes18:41
@wikingthis is c++18:41
@HeikoSSaurabh7__: ^18:41
@wikingnot java18:41
@HeikoSinstance->get_weirdthing<another_weird_thing>()::interface18:42
@HeikoShaha18:42
@HeikoSarianepaola: hi there!18:42
arianepaolahi HeikoS18:42
@HeikoSarianepaola: how are things?18:42
arianepaolaHeikoS: good18:43
@wikingHeikoS: actually good question: what does swig do with operators :DDDD18:43
@HeikoSwiking: screw them up properly18:43
@wikingah cool18:43
@wikinghttp://www.swig.org/Doc1.3/SWIGPlus.html#SWIGPlus_nn2818:43
@HeikoSwiking: but such access methods do not even need to be exposed18:43
@HeikoSwiking: feats->get_feature_matrix()[i]18:44
@HeikoSarianepaola: had a chance to check a cookbook entry?18:44
@wikinghttp://www.swig.org/Doc3.0/SWIGPlus.html#SWIGPlus_nn2818:44
@wikingHeikoS: ^ swig works nicely with such things18:44
@wiking;)18:44
@HeikoSwiking: well then18:44
@HeikoSwiking: had a chat to arianepaola about helping with release pip?18:45
arianepaolaHeikoS: have to add one18:45
@wikingHeikoS: not yet about that18:45
@wikinganybody used virt-install before?18:45
@wikingi need an aarch64 linux instance18:46
@HeikoSnope18:47
@wikingi mean = we need an aarch64 instance :P18:47
@HeikoS:P18:47
@wikingand aws is not good for such things :)18:47
@HeikoSwiking: btw what our current usage rate?18:48
@wikinghahaha18:48
@wiking0.5 USD/day18:48
@HeikoSwiking: btw2, shall we maybe announce aws donation?18:48
@HeikoShaha18:48
@HeikoSnice18:48
@wikingHeikoS: today it'll be more18:48
@wikingbecause i tested gpu instance18:48
@HeikoSi see18:48
@wikingbut on average it is 0.5 usd/day18:48
@wikingso that should be good for anohter18:49
@HeikoSthats 150 / year18:49
@wiking87 year18:49
@wiking16000/0.5/36518:49
@HeikoSwhich is 100 times less that we needed18:49
@wikingor?18:49
@HeikoSeven less haha18:49
@wikingso yeah but we have to use that within 1 year18:49
@HeikoSwe can do lots of binary builds soon18:49
@HeikoSand also the cloud18:49
@wikingre: btw2 lets wait until we dont get the shogun-cloud back online18:49
@HeikoSkk18:49
@HeikoSgood idea18:50
@HeikoSand then announce together18:50
@wikingyep18:50
-!- besser82 [~besser82@fedora/besser82] has joined #shogun19:15
-!- mode/#shogun [+o besser82] by ChanServ19:15
-!- HeikoS [~heiko@176.12.107.140] has quit [Read error: Connection reset by peer]19:15
-!- HeikoS1 [~heiko@176.12.107.140] has joined #shogun19:15
-!- Saurabh7_ [~Saurabh7@1.39.98.181] has joined #shogun19:17
-!- Saurabh7__ [~Saurabh7@1.39.98.181] has quit [Ping timeout: 260 seconds]19:21
-!- Saurabh7_ [~Saurabh7@1.39.98.181] has quit [Ping timeout: 260 seconds]19:22
-!- HeikoS1 [~heiko@176.12.107.140] has quit [Ping timeout: 244 seconds]19:22
-!- travis-ci [~travis-ci@ec2-54-147-204-36.compute-1.amazonaws.com] has joined #shogun19:27
travis-ciit's lambday'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/13231673219:27
-!- travis-ci [~travis-ci@ec2-54-147-204-36.compute-1.amazonaws.com] has left #shogun []19:27
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 244 seconds]19:31
-!- HeikoS [~heiko@176.12.107.140] has joined #shogun20:13
-!- mode/#shogun [+o HeikoS] by ChanServ20:13
-!- HeikoS [~heiko@176.12.107.140] has quit [Quit: Leaving.]20:33
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has quit [Quit: Page closed]20:59
-!- OXPHOS [9d8b131c@gateway/web/freenode/ip.157.139.19.28] has quit [Quit: Page closed]21:39
shogun-buildbotbuild #242 of deb1 - libshogun - PR is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/24222:21
arianepaolauntil tomorrow everyone :-)22:47
-!- Saurabh7 [Saurabh7@gateway/shell/panicbnc/x-nnnvlugepkedqrcd] has quit [Ping timeout: 260 seconds]22:58
-!- Saurabh7 [Saurabh7@gateway/shell/panicbnc/x-wnnuuobwbmuhgbpn] has joined #shogun23:19
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun23:57
-!- mode/#shogun [+o HeikoS] by ChanServ23:57
--- Log closed Tue May 24 00:00:09 2016