Open in new window / Try shogun cloud
--- Log opened Wed Sep 25 00:00:13 2013
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has quit [Quit: Leaving.]00:39
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has joined #shogun00:42
-!- mode/#shogun [+o HeikoS] by ChanServ00:42
-!- Netsplit *.net <-> *.split quits: gsomix, @sonney2k, @iglesiasg01:47
-!- Netsplit over, joins: @iglesiasg, gsomix, @sonney2k01:54
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat]02:04
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun02:27
shogun-notifier-shogun: Heiko Strathmann :gp_notebook * a5a2951 / doc/ipython-notebooks/gaussian_process/gaussian_process_classification.ipynb: https://github.com/shogun-toolbox/shogun/commit/a5a295194d8e3e10a3968afb713c5526d0069b9e02:27
shogun-notifier-shogun: wrote part on toy regression02:27
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has quit [Quit: Leaving.]02:32
shogun-buildbotbuild #466 of nightly_all is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_all/builds/46603:22
-!- travis-ci [~travis-ci@ec2-107-22-24-84.compute-1.amazonaws.com] has joined #shogun03:47
travis-ci[travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/1175760603:47
-!- travis-ci [~travis-ci@ec2-107-22-24-84.compute-1.amazonaws.com] has left #shogun []03:47
-!- sonne|osx_ [~sonne@g225082005.adsl.alicedsl.de] has joined #shogun03:53
-!- sonne|osx [~sonne@f053045153.adsl.alicedsl.de] has quit [Ping timeout: 246 seconds]03:55
-!- sonne|osx_ is now known as sonne|osx03:55
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]05:27
-!- sonne|osx [~sonne@g225082005.adsl.alicedsl.de] has quit [Quit: sonne|osx]06:52
-!- new_lido [~walid@41.218.177.206] has quit [Ping timeout: 240 seconds]07:31
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has joined #shogun07:37
-!- sonne|osx [~sonne@89.204.138.93] has joined #shogun08:17
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun08:19
-!- mode/#shogun [+o iglesiasg] by ChanServ08:20
-!- sonne|osx [~sonne@89.204.138.93] has quit [Ping timeout: 246 seconds]08:24
-!- sonne|osx [~sonne@82.113.106.4] has joined #shogun08:24
gsomixsonne|osx, good morning08:31
sonne|osxgsomix: hey moin!08:31
-!- sonne|osx [~sonne@82.113.106.4] has quit [Quit: sonne|osx]08:36
-!- besser82 [~besser82@fedora/besser82] has joined #shogun09:23
@iglesiasgsonne|work, you and I are immersed in an eternal battle for the longest streak in GitHub :P09:26
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun09:27
shogun-notifier-shogun: Fernando Iglesias :develop * 414c4c0 / data: https://github.com/shogun-toolbox/shogun/commit/414c4c025a03a311961b18dc219513f2c28f060309:27
shogun-notifier-shogun: Update data submodule commit09:27
shogun-notifier-shogun: Soeren Sonnenburg :develop * ebd3446 / data: https://github.com/shogun-toolbox/shogun/commit/ebd344684518d5e832a7d9be43652a44952352a409:27
shogun-notifier-shogun: Merge pull request #1675 from iglesias/develop09:27
shogun-notifier-shogun:09:27
shogun-notifier-shogun: Update data submodule commit09:27
sonne|workiglesiasg: same count :)09:28
sonne|workwhat a coincidence09:28
@iglesiasgsonne|work, now you are ahead again I think hehe09:28
shogun-notifier-shogun: lambday :develop * 9455535 / / (88 files): https://github.com/shogun-toolbox/shogun/commit/94555351b0e8089e750785b4269fd53a196c37ee09:29
shogun-notifier-shogun: renamed datatype complex64_t to complex128_t09:29
shogun-notifier-shogun: Soeren Sonnenburg :develop * fd0402b / / (88 files): https://github.com/shogun-toolbox/shogun/commit/fd0402bf41851af31d9fd638d1802c151141271f09:29
shogun-notifier-shogun: Merge pull request #1673 from lambday/develop09:29
shogun-notifier-shogun:09:29
shogun-notifier-shogun: renamed datatype complex64_t to complex128_t09:29
sonne|workgsomix: can you update the PR so I can finally merge this?09:29
shogun-notifier-shogun: Fernando Iglesias :develop * 86872b5 / doc/ipython-notebooks/metric/LMNN.ipynb: https://github.com/shogun-toolbox/shogun/commit/86872b519aa836ebad4dd5972cbd4f7fc7bef65109:30
shogun-notifier-shogun: Add bar chart to LMNN notebook for wine data section09:30
shogun-notifier-shogun: Soeren Sonnenburg :develop * 6c2a673 / doc/ipython-notebooks/metric/LMNN.ipynb: https://github.com/shogun-toolbox/shogun/commit/6c2a6733c60944e62567a0fbef5c83f32749017909:30
shogun-notifier-shogun: Merge pull request #1674 from iglesias/feature/lmnn09:30
shogun-notifier-shogun:09:30
shogun-notifier-shogun: Add bar chart to LMNN notebook for wine data section09:30
sonne|workiglesiasg: is the SOSVM SGD solver PR ready to be merged? I guess you take care of it...09:30
@iglesiasgsonne|work, no, it is not. Patrick wrote some comments yesterday that need to be addressed09:31
@iglesiasgsonne|work, and we need unit tests, hushell said he was on it09:31
sonne|workok09:31
wikingyo09:33
* wiking is back from a 2 days flu downtime09:34
gsomixsonne|work, yep. just a moment09:36
wikingsonne|work: png09:36
wikingping09:36
sonne|workwiking: a real phoenix you are09:36
wikinggsomix: ok i'm back now09:36
wikingsonne|work: fuck man flu took me badly09:36
wikingsonne|work: have u managed to change the script on 7nn for notebook generation?09:37
sonne|workwiking: lots of people here are out too09:37
sonne|workwiking: didn't do anything09:37
sonne|workwiking: we need the protoc support for gsomix09:37
wikingsonne|work: yeah i know09:37
wikingsonne|work: can u plz do the changes according to https://github.com/shogun-toolbox/shogun/issues/148209:39
sonne|workwiking: tonight09:40
wikingsonne|work: okie thnx09:41
wikingok so somebody who's protobuf person09:41
wikinghow does the .cc gets generated?09:42
sonne|workwiking: protoc --cpp *.proto09:42
sonne|workwiking: actually protoc filename.proto --cpp_out=OUT_DIR09:43
wikingsonne|work: ok and this has not been written yet right?09:43
sonne|workwiking: no - we have a commented PROTOBUF detection in there though09:43
wikingokok09:44
wikingsonne|work: fixing it now09:44
wikingbtw09:44
wikingdo we need the protobuf headers and library?09:45
wikingi have no idea about how protobuf works09:45
wikingthat's why i'm asking09:45
wikingbtw are we going to have .proto files anywhere else?09:46
wikingother than src/shogun?09:46
sonne|workwiking: well we could put them at some other place if you mean that... it is kind of the .i files - special09:47
wikingnono09:47
wikingi mean that are we going to have them elsewhere as well/09:47
wikingjust to know what to expect09:47
wikingi'm fine with it as it is now09:47
sonne|workok then just use the current location09:51
wikingare these the files that should be generated?09:54
wiking/Users/wiking/shogun/build/src/shogun/Chunks.pb.h/Users/wiking/shogun/build/src/shogun/Headers.pb.h/Users/wiking/shogun/build/src/shogun/ShogunVersion.pb.h09:54
wiking/Users/wiking/shogun/build/src/shogun/Chunks.pb.cc/Users/wiking/shogun/build/src/shogun/Headers.pb.cc/Users/wiking/shogun/build/src/shogun/ShogunVersion.pb.cc09:54
wikingor i guess they are in the wrong location...:S09:54
wikingwonder how to fix this with cmake...09:54
gsomixsonne|work, https://github.com/shogun-toolbox/shogun/pull/1670 updated!09:56
gsomixwiking, glad to see you, man09:56
shogun-notifier-shogun: Evgeniy Andreev :develop * 95deac2 / / (5 files): https://github.com/shogun-toolbox/shogun/commit/95deac24e6bf1a8990b99543ac6affd4b781823a09:56
shogun-notifier-shogun: add sparse matrix support for Protobuf file format09:56
shogun-notifier-shogun: Soeren Sonnenburg :develop * d22e14e / / (5 files): https://github.com/shogun-toolbox/shogun/commit/d22e14e561e81d78129e95ad7e6202875f84781509:56
shogun-notifier-shogun: Merge pull request #1670 from gsomix/feature/protobuf09:56
shogun-notifier-shogun:09:56
shogun-notifier-shogun: Add sparse matrix support for Protobuf file format09:56
gsomixOK. cu later guys - need to visit the clinic09:57
sonne|workgsomix: cu!10:01
shogun-buildbotbuild #1541 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1541  blamelist: Soeren Sonnenburg <sonne@debian.org>, Fernando Iglesias <fernando.iglesiasg@gmail.com>10:02
besser82G' Morning, Folks!  :)10:13
-!- gsomix [~gsomix@83.234.54.128] has quit [Ping timeout: 245 seconds]10:21
shogun-buildbotbuild #1813 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/1813  blamelist: Evgeniy Andreev <gsomix@gmail.com>, Soeren Sonnenburg <sonne@debian.org>10:21
-!- gsomix [~gsomix@83.234.54.128] has joined #shogun10:22
@iglesiasgbesser82, morning10:30
besser82iglesiasg: thanks, mornin'  :D10:30
besser82iglesiasg: How is it going?  I'm just compiling / test new checkout with updated data :)10:32
@iglesiasgbesser82, nice. Yeah , it is a data set used in the lmnn notebook10:33
@iglesiasga small one about wines.. yummy :D10:33
besser82iglesiasg: XD   wines are always tasty  ;)10:33
@iglesiasgbesser82, and Spanish ones even better10:33
besser82iglesiasg, you should definetivly taste some German ones, like those "Badische" Wines  ;)10:37
besser82iglesiasg, made in "Baden-W?rttemberg"10:37
shogun-buildbotbuild #148 of precise - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/148  blamelist: lambday <heavensdevil6909@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, Soeren Sonnenburg <sonne@debian.org>10:38
@iglesiasgbesser82, deffs I should. I don't recall having tried German wine actully :/10:38
besser82iglesiasg,  give them shot  :)  You can get pretty drunk from them, but feeling really good next day ;)10:44
besser82sonne|work:  Still no answer from TJ for SVMlight :(  He actually is a German, isn't he?10:45
-!- new_lido [~walid@193.227.20.2] has joined #shogun10:57
sonne|workbesser82: yes he is!10:57
besser82sonne|work:  :D As I supposed.  So he should be able to read my mail  ;)  Probably he's on vacation or so.10:58
sonne|workbesser82: well he is a big shot with little time10:58
besser82sonne|work: Or he just doesn't want to answer  :(10:58
shogun-buildbotbuild #149 of precise - libshogun is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/14910:59
sonne|workbesser82: we thought a couple of times writing 'SVM^bright'11:00
besser82sonne|work: ???11:00
sonne|worktotally GPL'd svm light clone11:00
besser82sonne|work: that would be cool, indeed  ;)11:00
sonne|workbut then libsvm etc are doing good enough11:00
besser82sonne|work: but isn't redistable  :(11:01
sonne|workyeah I know even worse I would say. not allowed to modify etc etc11:01
sonne|workI would very much like to get rid of it altogether11:02
besser82sonne|work: yes, but is there any suitable substitue?11:02
sonne|workbesser82: in principle there is but this version here is pimped (like the libsvm in shogun)11:03
sonne|workso they can be a bit faster (parallel, faster kernels, ...)11:03
besser82sonne|work: which one would be the substitute?  Perhaps I can take care of  ;)11:03
shogun-buildbotbuild #1542 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1542  blamelist: lambday <heavensdevil6909@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, Soeren Sonnenburg <sonne@debian.org>11:06
sonne|workbesser82: GPBTSVM is also a chunking based svm like svm light11:07
sonne|workbesser82: they claim to have a faster internal solver being able to process bigger chunks per iteration11:08
sonne|workhowever it doesn't do shrinking11:08
besser82sonne|work: so let me do some experiments with it  ;)  PErhaps we can replace SVMlight with this one, to be completely FOSS again ;)11:08
sonne|workand does not have the various hacks to make convergence more stable11:08
sonne|workbesser82: it is in shogun already11:09
besser82sonne|work: so another bundled thingy  ;)11:09
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun11:10
besser82sonne|work: I'll try to rewrite a binary / API compatible FOSS version of SVMlight, then  ;)11:12
lisitsynwhat?11:13
lisitsyn:D11:13
sonne|workbesser82: looks like you have too much time :)11:13
besser82sonne|work: Yes, sometimes ;)11:13
sonne|workbesser82: better do cmake fixes :)11:13
besser82sonne|work: I'm on them :)11:13
lisitsynbesser82: I believe it only makes sense to implement something that is *better* than svmlight11:14
besser82sonne|work: btw. How about that "`make install` downloads stuff" thing?11:14
besser82sonne|work: That would be a tough thing to write ;)11:15
besser82sonne|work: The better than SVMlight...11:15
lisitsynbesser82: well it is quite old so there must be some room for improvement11:18
lisitsynotherwise we are all stupid11:18
besser82lisitsyn: possibly there is, but I'm not too skilled in writing code ;)11:19
lisitsynbesser82: it is about math I think11:20
lisitsynbesser82: I am skilled in code but my brain is not capable to cope with that math haha11:20
lisitsynwell I didn't try but I am pessimistic about it11:20
besser82lisitsyn: that would be my domain  ;)  Let me try to understand the SVMlight sources ... ;)11:20
sonne|workhorrible11:21
sonne|workdon't even try to do that11:21
lisitsynyeah never try to read research code11:21
lisitsyn:D11:21
besser82lisitsyn, sonne|work:  :D  I can't code that well, but im BSc applied Maths  ;)11:21
sonne|worklike shogun you mean?11:21
lisitsynbesser82: same shit here, doesn't help :D11:21
besser82lisitsyn: Let's fetch some Asian then to do it  ;)11:22
lisitsynsonne|work: shogun is much more closer to industrial code now11:22
lisitsynbesser82: that's a good point11:22
sonne|worklisitsyn: yeah true the buildbot armada is actually as good as it gets IMHO11:22
lisitsynsonne|work: I don't have to tell you about research code I believe :D11:23
lisitsynyou must have seen terrible things11:23
lisitsynmuch more terrible than I have seen11:23
sonne|worksure11:23
sonne|workI wrote it myself back then11:23
lisitsynhaha good you joined engineering side ;)11:23
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has joined #shogun11:23
-!- mode/#shogun [+o HeikoS] by ChanServ11:23
sonne|workwell for quick tests I still do it this way11:23
lisitsynit is ok to write shitty code then improve11:24
besser82lisitsyn, sonne|work: like I propsed, let's get an asian to do it  ;)   ---> http://www.youtube.com/watch?v=p2EA2nA2ar011:24
lisitsynhahahah11:24
lisitsynI love family guy11:24
besser82lisitsyn: me, too XD11:24
wikingHeikoS: portugal? :)11:27
lisitsynwiking: yeah he is doing his surfing phd there11:29
shogun-buildbotbuild #43 of clang34 - static analysis is complete: Failure [failed analyse]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/43  blamelist: Soeren Sonnenburg <sonne@debian.org>, Fernando Iglesias <fernando.iglesiasg@gmail.com>11:33
shogun-buildbotbuild #27 of clang34 - undefined behaviour analysis is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20undefined%20behaviour%20analysis/builds/27  blamelist: Soeren Sonnenburg <sonne@debian.org>, Fernando Iglesias <fernando.iglesiasg@gmail.com>11:44
wikinglisitsyn: :D11:45
shogun-buildbotbuild #1543 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1543  blamelist: Evgeniy Andreev <gsomix@gmail.com>, Soeren Sonnenburg <sonne@debian.org>11:46
shogun-buildbotbuild #30 of clang34 - thread analysis is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/30  blamelist: Soeren Sonnenburg <sonne@debian.org>, Fernando Iglesias <fernando.iglesiasg@gmail.com>11:59
gsomixsonne|work, RE12:45
shogun-buildbotbuild #44 of clang34 - static analysis is complete: Failure [failed analyse]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/44  blamelist: lambday <heavensdevil6909@gmail.com>12:53
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]12:57
-!- gsomix [~gsomix@83.234.54.128] has quit [Remote host closed the connection]13:22
-!- gsomix [~gsomix@83.234.54.128] has joined #shogun13:25
-!- votjakovr [~votjakovr@host-46-241-3-209.bbcustomer.zsttk.net] has quit [Quit: Good bye!]13:33
shogun-buildbotbuild #45 of clang34 - static analysis is complete: Failure [failed analyse]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/45  blamelist: Evgeniy Andreev <gsomix@gmail.com>, Soeren Sonnenburg <sonne@debian.org>, Fernando Iglesias <fernando.iglesiasg@gmail.com>13:35
shogun-buildbotbuild #28 of clang34 - undefined behaviour analysis is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20undefined%20behaviour%20analysis/builds/28  blamelist: Evgeniy Andreev <gsomix@gmail.com>, lambday <heavensdevil6909@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, Soeren Sonnenburg13:44
shogun-buildbot<sonne@debian.org>13:44
shogun-buildbotbuild #31 of clang34 - thread analysis is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/31  blamelist: Evgeniy Andreev <gsomix@gmail.com>, lambday <heavensdevil6909@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, Soeren Sonnenburg <sonne@debian.org>13:58
-!- new_lido [~walid@193.227.20.2] has quit [Remote host closed the connection]14:39
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has joined #shogun14:54
travis-ci[travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/1176645914:54
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has left #shogun []14:54
@HeikoSwiking: hi!15:25
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun15:40
shogun-notifier-shogun: Roman Votyakov :develop * 8f62dd4 / / (18 files): https://github.com/shogun-toolbox/shogun/commit/8f62dd44fbb9f08813f9b55b5109e7ba107a4f5f15:40
shogun-notifier-shogun: refactor GPs framework15:40
shogun-notifier-shogun: Heiko Strathmann :develop * fd4203c / / (18 files): https://github.com/shogun-toolbox/shogun/commit/fd4203c3a9da1019d807930a2a07af9291467e7915:40
shogun-notifier-shogun: Merge pull request #1676 from votjakovr/feature/gp_refactoring15:40
shogun-notifier-shogun:15:40
shogun-notifier-shogun: refactor GPs framework15:40
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat]15:40
shogun-buildbotbuild #1814 of bsd1 - libshogun is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/181416:02
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has joined #shogun16:14
travis-ci[travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/1176651516:14
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has left #shogun []16:14
sonne|workgsomix: I've already passed you gsoc wise so now the only thing left would be the matlab code16:19
sonne|workgsomix: matlab reading/writing16:19
sonne|workgsomix: maybe you give it a try - might be easy with the opencv mat reader...16:19
shogun-buildbotbuild #1544 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1544  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>16:37
shogun-notifier-shogun: Heiko Strathmann :gp_notebook * c1afa6a / doc/ipython-notebooks/gaussian_process/gaussian_process_classification.ipynb: https://github.com/shogun-toolbox/shogun/commit/c1afa6aec30b9601879e538df5e127742d3e245516:41
shogun-notifier-shogun: -use posterior GP methods that Romand just implemented16:41
shogun-notifier-shogun: -some more updates16:41
gsomixsonne|work, yes, of course. I'll start playing with opencv reader soon.16:42
gsomixI also need refactor some old code and write examples.16:42
@HeikoShttp://nbviewer.ipython.org/668599016:42
@HeikoSsonne|work: since you asked how to do the "full" thing with GPs, this notebook contains a toy example for both classification and regression, including model-selection without any x-validation or grid-search16:43
-!- HeikoS [~heiko@218.4.166.178.rev.vodafone.pt] has quit [Quit: Leaving.]16:44
-!- sonne|osx [~sonne@89.204.138.94] has joined #shogun17:17
sonne|osxgsomix: btw one thing to clean up would be to return a sparse matrix even when reading CSV's17:18
wikingsonne|osx: dontforget https://github.com/shogun-toolbox/shogun/issues/1482#issuecomment-2473568017:18
sonne|osxgsomix: you could use the poor mans solution for that and just use the SGSparseMatrix(dense) thing for that17:19
sonne|osxwiking: tonight17:19
gsomixsonne|osx, yep, but not memory efficient.17:20
sonne|osxgsomix: and vice versa for libsvmfile you could return a dense matrix too17:20
shogun-buildbotbuild #1893 of deb3 - modular_interfaces is complete: Failure [failed test libshogun]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1893  blamelist: Roman Votyakov <votjakovr@gmail.com>17:20
sonne|osxgsomix: well better than nothing but feel free to do it better than that17:20
sonne|osxwiking: have you seen travis & java issues? I think we need some higher -Xmx for java to run17:21
-!- travis-ci [~travis-ci@ec2-54-211-105-97.compute-1.amazonaws.com] has joined #shogun17:32
travis-ci[travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/1176655817:32
-!- travis-ci [~travis-ci@ec2-54-211-105-97.compute-1.amazonaws.com] has left #shogun []17:32
shogun-buildbotbuild #1545 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1545  blamelist: Roman Votyakov <votjakovr@gmail.com>17:33
sonne|osxwiking: do you know why python modular takes 50mins on travis?17:37
sonne|osxall the rest is good in 8-30 mins17:38
sonne|osxgtg17:39
-!- sonne|osx [~sonne@89.204.138.94] has quit [Quit: sonne|osx]17:39
shogun-buildbotbuild #1892 of deb3 - modular_interfaces is complete: Failure [failed test ruby modular]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/1892  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>17:42
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.]17:46
-!- sonne|osx [~sonne@g225082005.adsl.alicedsl.de] has joined #shogun17:53
-!- zxtx [~zv@209-6-202-142.c3-0.smr-ubr2.sbo-smr.ma.cable.rcn.com] has joined #shogun18:25
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has joined #shogun19:13
travis-ci[travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/1176733719:13
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has left #shogun []19:13
-!- zxtx [~zv@209-6-202-142.c3-0.smr-ubr2.sbo-smr.ma.cable.rcn.com] has quit [Ping timeout: 240 seconds]19:16
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]19:41
@sonney2kwiking, nbconvert is run now19:56
wikingsonney2k: did u try to run it?20:03
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has joined #shogun20:03
-!- new_lido [~walid@41.218.177.108] has joined #shogun20:17
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has joined #shogun21:07
travis-ci[travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/1177833821:07
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has left #shogun []21:07
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun21:15
shogun-notifier-shogun: Soeren Sonnenburg :develop * 25f4d66 / applications/arts/__init__.py,applications/asp/__init__.py: https://github.com/shogun-toolbox/shogun/commit/25f4d663dce7a20810b0505ba40ac7927a53f5d721:15
shogun-notifier-shogun: add arts/asp __init__.py such that it can be imported21:15
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun21:22
-!- mode/#shogun [+o iglesiasg] by ChanServ21:22
@iglesiasghello people21:24
@iglesiasgwiking, sonney2k, ping ping21:24
gsomixiglesiasg, hi21:40
-!- KMcQuisten [d8338942@gateway/web/freenode/ip.216.51.137.66] has joined #shogun21:48
KMcQuistenHi there folks21:48
lisitsynKMcQuisten: hey21:48
lisitsynnice to see you here again ;)21:49
KMcQuistenSo, i've been trying to compile new versions of shogun using cmake, and no matter what I do, it just won't compile.  It will compile all the c++ objects, but then it just craps out whenever it tries to compile the python wrapper, and I can't figure out why.  Iv'e fiddled with every last parameter to point in things in what I think are the right directions, but it never works.  I need to update to fix some bugs, but it just isn't w21:50
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has joined #shogun21:52
travis-ci[travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/1178118721:52
-!- travis-ci [~travis-ci@ec2-184-72-85-210.compute-1.amazonaws.com] has left #shogun []21:52
@iglesiasgKMcQuisten, what kind of errors do you get?21:52
@iglesiasgKMcQuisten, what do you see related to python paths when you do ccmake ..?21:53
@iglesiasgmaybe you can check in there if the variables are well assigned for cmake21:53
lisitsynKMcQuisten: yeah please paste errors21:53
lisitsynsomewhere21:53
KMcQuistenI've been using a cmake gui, cmake28.  ccmake is not a command that my terminal recognizes21:54
besser82KMcQuisten: you're on RHEL6, right?  ;)21:54
KMcQuisteni'm aon a Centos6 workstation.  I went through hell for a while becasue it was not very clearly announced that python 2.6 would not be supported.  2.6 is the version my distro uses for its package manager, so I can't remove it.21:55
KMcQuistenso i've installed 2.7 in a local director and i've been pointing cmake to those directories21:55
KMcQuistenI don't have specific errors in front of me, because I rage-deleted my source folder21:56
KMcQuistenI have to clone the repo again21:56
besser82KMcQuisten: try ccmake28 or ccmake-2.821:56
KMcQuistenthe errors are all errors saying the the objects don't contain the propertiesthey are supposed to.21:57
KMcQuistenOne moment21:57
besser82KMcQuisten: and I guess your main problem is numpy or scipy.... not Python2.6  ;)21:59
KMcQuistenI freshly recomiled numpy and scipy to use python2.721:59
KMcQuistenI ran and tested numpy and scipy in the python 2.7 interpreter and they were playing nicely22:00
besser82OK22:00
besser82try to build them for Py2.6 and all should be fine, without python2.7  ;)22:01
KMcQuistenThey're already built and installed for 2.622:01
besser82which versions?22:01
besser82stock CentOS6?22:02
KMcQuistennumpy 1.6.2 and scipy0.12.022:02
KMcQuisteni built them from source.  I dind't use what was in the centos repo22:02
besser82should be recent enough...22:02
besser82mhhh....22:02
KMcQuistenyes, stock centos622:02
KMcQuistenif py2.6 is still supported, why to the installation instructions say it's no longer supported?22:03
besser82I dunno....22:03
besser82I'm curently working on RPMs for shogun supporting EL5 / EL6 and Fedora 18+22:04
besser82And didn't run to Python problems whether on el5 nor el622:04
besser82yet...22:04
besser82those rpms will be available during "official" channels btw.22:05
besser82EPEL repo for el5/el6...22:05
KMcQuistenThat would be lovely.  What's the timeframe on that?22:05
besser82est. 3 - 4 weeks  (depends on factor I can't influence)22:06
KMcQuistenwell that will be nice.22:07
KMcQuistenhopefully I will not have bashed my skull in from frustration by then ;)22:07
lisitsynKMcQuisten: so basically the only reason it stopped to work is cmake?22:11
KMcQuistenYeah, when you guys changed over to cmake, I couldn't get it to work anymore22:11
lisitsynsorry I still don't get what is the exact error22:12
KMcQuistenI can see the reasoning.  autotools are ancient and inflexible.  I'm compiling again so I can post the actual error22:12
besser82that would be nice ;)22:12
lisitsynok thanks22:13
lisitsynKMcQuisten: please bug us till we help you because it is us who should care about it22:13
lisitsynit is we* I mean22:13
lisitsyn:D22:13
lisitsynsonney2k: I tried to use multiboost22:14
lisitsyncrazy22:14
lisitsyncommand line oriented classifiers22:15
lisitsynI haven't thought it can be that bad22:16
shogun-buildbotbuild #1546 of cyg1 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1546  blamelist: Soeren Sonnenburg <sonne@debian.org>22:16
lisitsynthese guys don't give a shit how to code..22:18
@iglesiasglisitsyn, what is it you didn't like? :)22:22
lisitsyniglesiasg: well22:23
@iglesiasglisitsyn, all right. Did you like something :P?22:23
lisitsyniglesiasg: I want to use multiboost in C++22:23
lisitsynso how NORMAL PEOPLE do that22:23
lisitsynthey include shit and compile and link22:23
lisitsynand see how it goes22:23
@iglesiasgaaah it is just usable from the command line?22:23
@iglesiasgno API?22:24
lisitsyny22:24
lisitsyne22:24
lisitsyns22:24
lisitsynit takes Args22:24
lisitsynokay I thought22:24
lisitsynI'll put my arguments to args22:24
lisitsynand it will work22:24
lisitsynbut THERE IS NO METHOD TO SET ARGS22:24
lisitsynthey just parse them22:24
lisitsynhow can one think it is normal to do that?!22:24
@iglesiasgcrazy22:25
KMcQuistenBecasue most coders of scientific software are graduate students, not professional software developers?22:34
KMcQuistenYou guys are the glaring exception to this fact.  Even your graduate student coders write stuff that's readable and makes sense.22:34
shogun-buildbotbuild #1894 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/189422:34
KMcQuistenIt's quite impressive22:34
lisitsynKMcQuisten: these multiboost developers are not really students I think22:35
lisitsynoh sometimes I don't understand what researchers do because I code all the time to try something new22:36
lisitsynhow can one say come up with a new classifier w/o programming skills22:36
KMcQuistenI think their goal is just to get something that works without thinking about how it fits into pipelines and frameworks, and without thinking about the design of their code as somethign that's important in and of itself22:37
lisitsynit is usually about self-discipline22:37
lisitsynit doesn't help at all22:38
KMcQuistenExactly.  Shit code can be done quickly, good code takes time, effort, critical thinking, and patience22:38
lisitsynyou build bad designed tool22:38
lisitsynthen you use it22:38
lisitsynand spend time22:38
lisitsynbecause of its bad design22:38
KMcQuistenExactly.  It's short-term thinking causing long-term delay22:39
KMcQuistendoing it right puts the long term first, but then adaptation and change is straightforward when code's designed well22:39
lisitsynyeah you just save your time doing it right22:40
lisitsynthat's why I don't understand it22:40
@iglesiasgit is just about getting results and writing papers22:40
@iglesiasgthe faster you get results, the sooner you can a write a paper with sth22:41
lisitsyniglesiasg: haha! that's why I won't join that22:41
@iglesiasgjust a thought22:41
@iglesiasgprobably not all are like that22:41
KMcQuistenExactly.  they just want their answers and don't care adfter that22:41
lisitsynI have been convinced by all the stuff coming around that research is shit :D22:42
lisitsynthe more industry related it is the better it gets22:42
lisitsynonce you need something real the game changes22:42
KMcQuistenI'd say that's pretty accurate.  Companies want code that can be read and understood by someone who didn't write it22:43
lisitsynit seems to be quite frustrating to me22:43
KMcQuistenI agree totally.  Well designed code makes your life so much easier22:43
KMcQuistenI absolutely love the work you guys put into this, especially all the unit testing.22:45
KMcQuistenok, I have some errors for you:22:45
lisitsynoh finally22:45
lisitsyngodo22:45
lisitsyngood22:45
KMcQuisten[ 84%] Building CXX object src/interfaces/python_modular/CMakeFiles/_python_modular.dir/modshogunPYTHON_wrap.cxx.o22:46
KMcQuistenIn file included from /home/kmcquisten/shogun/build/src/interfaces/python_modular/modshogunPYTHON_wrap.cxx:5211: /home/kmcquisten/shogun/src/shogun/distance/DirectorDistance.h: In member function 'virtual float64_t* shogun::CDirectorDistance::get_distance_matrix_real(int32_t&, int32_t&, float64_t*)': /home/kmcquisten/shogun/src/shogun/distance/DirectorDistance.h:89: error: 'get_distance_matrix_real' is not a member of 's22:46
KMcQuistenIn file included from /home/kmcquisten/shogun/build/src/interfaces/python_modular/modshogunPYTHON_wrap.cxx:5211: /home/kmcquisten/shogun/src/shogun/distance/DirectorDistance.h: In member function 'virtual float32_t* shogun::CDirectorDistance::get_distance_matrix_shortreal(int32_t&, int32_t&, float32_t*)': /home/kmcquisten/shogun/src/shogun/distance/DirectorDistance.h:101: error: 'get_distance_matrix_shortreal' is not a mem22:46
lisitsynohh22:47
lisitsyndirectors..22:47
KMcQuistenit's not just directors22:48
KMcQuistenIn function 'PyObject* _wrap_CSVFile_set_order(PyObject*, PyObject*)': /home/kmcquisten/shogun/build/src/interfaces/python_modular/modshogunPYTHON_wrap.cxx:52943: error: 'csv_data_order' is not a member of 'shogun' /home/kmcquisten/shogun/build/src/interfaces/python_modular/modshogunPYTHON_wrap.cxx:52943: error: expected ';' before 'arg2' /home/kmcquisten/shogun/build/src/interfaces/python_modular/modshogunPYTH22:48
lisitsynha22:48
lisitsynKMcQuisten: do you have shogun installed already?22:48
KMcQuistenYes, i've left the old version alone so I could still get some work done while I was trying to figure this out22:49
lisitsynKMcQuisten: is it in /usr/local/include?22:49
KMcQuistennope, it's in usr/include22:49
lisitsynoh even that22:49
lisitsynKMcQuisten: could you move it out of the folder22:49
lisitsyntry to compile and get it back later22:49
KMcQuistenWhich folder?22:49
lisitsynI mean move shogun headers out of /usr/include22:50
lisitsynI have a suspect22:50
KMcQuistenI can try, but why would that matter for compilation?22:50
lisitsynlets see if he goes in custody22:50
lisitsynwell it uses old headers but new sources22:50
KMcQuistenBut I have it pointing ot hte new headers22:51
KMcQuistenok, I'll try it22:51
KMcQuisten:)22:51
lisitsynI have seen csv_data_order before22:51
lisitsynand it was the culprit22:51
KMcQuistenok, moved the old headers out of usr/include22:52
KMcQuistenUgh, it's recompiling everything.22:53
KMcQuistenthis might be a minute22:53
lisitsynKMcQuisten: I believe it should solve the problem22:57
KMcQuistenHere's hoping! :)22:57
-!- travis-ci [~travis-ci@ec2-54-211-105-97.compute-1.amazonaws.com] has joined #shogun23:01
travis-ci[travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/1179315723:01
-!- travis-ci [~travis-ci@ec2-54-211-105-97.compute-1.amazonaws.com] has left #shogun []23:01
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat]23:12
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun23:12
KMcQuistenSame load of error diarreah23:15
KMcQuistenIn function 'PyObject* _wrap_Math_min__SWIG_1(PyObject*, PyObject*)': /home/kmcquisten/shogun/build/src/interfaces/python_modular/modshogunPYTHON_wrap.cxx:323245: error: 'complex64_t' was not declared in this scope /home/kmcquisten/shogun/build/src/interfaces/python_modular/modshogunPYTHON_wrap.cxx:323245: error: expected ';' before 'arg1' /home/kmcquisten/shogun/build/src/interfaces/python_modular/modshogunPYTHON_23:16
KMcQuistenIn function 'PyObject* _wrap_InferenceMethod_get_posterior_approximation_covariance(PyObject*, PyObject*)': /home/kmcquisten/shogun/build/src/interfaces/python_modular/modshogunPYTHON_wrap.cxx:665624: error: 'class shogun::CInferenceMethod' has no member named 'get_posterior_approximation_covariance'23:19
shogun-buildbotbuild #46 of clang34 - static analysis is complete: Failure [failed analyse]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/46  blamelist: Soeren Sonnenburg <sonne@debian.org>23:20
lisitsynKMcQuisten: argh23:20
lisitsynKMcQuisten: different errors though23:21
KMcQuistenNo no, same errors, I just gave you a different flavor.  There's about 6 pages of them23:22
lisitsynah23:22
KMcQuistenit's not finding a header somewhere, but I can't figure out where or why23:22
lisitsynKMcQuisten: well there is really no such method like get_posterior_approximation_covariance23:22
KMcQuistenIs that deprecated or something?  It's sure looking for it23:23
lisitsynKMcQuisten: are you sure you don't have any other shogun headers in your include path?23:23
lisitsyn/usr/include /usr/local/include etc23:23
KMcQuistenit's possible23:24
KMcQuistenwhy would it be looking there during compilation, though?  I'm working from a fresh pull of the repo in my home directory.23:24
lisitsynKMcQuisten: that's an issue with cmake we have23:24
lisitsynKMcQuisten: once you have include paths it is a matter of their order I guess23:25
lisitsynwe didn't have such problems with make cause we did control the order explicitly23:25
KMcQuistenI see23:26
lisitsynI still think it is the reason23:26
KMcQuistenLet me check where things are23:26
lisitsynKMcQuisten: just look for that "get_posterior_approximation_covariance"23:26
lisitsynthere should be a file swig takes it from23:27
KMcQuistenRighto.  i'll try that23:27
shogun-buildbotbuild #29 of clang34 - undefined behaviour analysis is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20undefined%20behaviour%20analysis/builds/29  blamelist: Soeren Sonnenburg <sonne@debian.org>23:30
-!- KMcQuisten [d8338942@gateway/web/freenode/ip.216.51.137.66] has quit [Quit: Page closed]23:41
shogun-buildbotbuild #32 of clang34 - thread analysis is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/32  blamelist: Soeren Sonnenburg <sonne@debian.org>23:44
--- Log closed Thu Sep 26 00:00:15 2013