Open in new window / Try shogun cloud
--- Log opened Fri Jun 28 00:00:08 2013
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving]00:04
lisitsyniglesiasg: what? :)00:36
@iglesiasglisitsyn: football!00:36
@iglesiasgconfederations cup00:36
lisitsyniglesiasg: ahh00:36
lisitsyniglesiasg: you are in convenient conditions with sweden for hockey and spain for football :D00:37
@iglesiasgI am a smart guy :P00:37
@iglesiasghttp://shogun-toolbox.org/page/Events/workshop2013_program00:38
@iglesiasgI updated the workshop program btw ^00:38
lisitsyngut iglesiasg00:39
@iglesiasgtack lisitsyn00:39
lisitsyniglesiasg: ack00:39
lisitsyn:D00:39
@iglesiasghehe00:39
lisitsynshall we speak strange short words?00:39
lisitsyn:D00:40
@iglesiasgn00:40
lisitsyniglesiasg: I added some things to my formatting stuff00:42
@iglesiasglisitsyn: nice, let me see00:42
lisitsyniglesiasg: now supports hex, oct and raw pointers00:42
lisitsyniglesiasg: https://github.com/lisitsyn/formatting/blob/master/test/wrappers.cc00:42
lisitsyniglesiasg: now if you really want pointer one may use formatting::raw(pointer)00:43
lisitsynif one*00:43
lisitsyn:)00:43
@iglesiasgreinterpret_cast wow00:44
lisitsyniglesiasg: the only way to cast int to pointer00:44
lisitsynthat's just for test00:44
lisitsynin usual case one hasn't to do that :)00:45
@iglesiasgbut one thing00:45
@iglesiasgabout hexadecimal and oct00:45
lisitsyniglesiasg: yes?00:45
@iglesiasgmmm so how would it actually work with variables00:46
@iglesiasgI mean what type of variables would you give to oct() or hex()00:46
@iglesiasgan integer in decimal?00:46
lisitsyniglesiasg: well actually any it just sets a flag00:46
lisitsynI didn't put any restriction to type yet00:46
@iglesiasgso say I do00:47
lisitsyn<< std::hex << value;00:47
@iglesiasgoct(8)00:47
@iglesiasgthat prints 1000:47
lisitsynyes00:47
@iglesiasgok :)00:47
lisitsyniglesiasg: I never used octal though00:47
lisitsyndid you?00:47
lisitsynabsolutely no idea what is it for :D00:47
@iglesiasgto change permissions haha00:48
lisitsyniglesiasg: I am now thinking about precision00:49
lisitsyniglesiasg: how do you like precision[3](3.1415)?00:49
@iglesiasglisitsyn: I like it00:50
lisitsynlet me put it this way then00:50
lisitsyniglesiasg: can you imagine better name?00:51
lisitsynprecised[3]00:51
lisitsynprecision[3]00:51
@iglesiasglet me think00:52
@iglesiasgno, not precised00:52
@iglesiasgprecision much better00:52
@iglesiasgthe only but is may look a bit verbose00:53
@iglesiasgother than that, jag gillar det00:53
lisitsyniglesiasg: what is jag gillar det? :)00:53
@iglesiasgI like it :)00:54
lisitsynahh00:54
lisitsyniglesiasg: I actually have a feeling such things are rather rarely used00:55
@iglesiasgI normally use the precision when I print floats00:56
@iglesiasgthe precision and the padding00:56
@iglesiasgI like to have the numbers well aligned :)00:56
lisitsyniglesiasg: argh now I think it should be here too00:57
@iglesiasgooops sorry00:58
lisitsyniglesiasg: this is less trivial as adds combinations00:59
@iglesiasgor more fun!01:01
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun01:08
shogun-notifier-shogun: iglesias :develop * f366d03 / tests/unit/mathematics/Math_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/f366d03906fe7ad24961157fa442130855ab081e01:08
shogun-notifier-shogun: linspace unit test01:08
shogun-notifier-shogun: Fernando Iglesias :develop * 27c9a1c / tests/unit/mathematics/Math_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/27c9a1cd0cb53183b7d86c2b6529c16ce133688101:08
shogun-notifier-shogun: Merge pull request #1193 from iglesias/develop01:08
shogun-notifier-shogun:01:08
shogun-notifier-shogun: linspace unit test01:08
-!- nube [~rho@49.244.117.250] has joined #shogun01:08
@iglesiasggood night!01:09
lisitsyngood night01:09
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]01:10
shogun-buildbotbuild #1140 of deb2 - static_interfaces is complete: Failure [failed test octave_static]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1140  blamelist: iglesias <fernando.iglesiasg@gmail.com>01:18
shogun-buildbotbuild #1153 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1153  blamelist: iglesias <fernando.iglesiasg@gmail.com>01:21
shogun-buildbotbuild #985 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/985  blamelist: iglesias <fernando.iglesiasg@gmail.com>01:24
shogun-buildbotbuild #1154 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1154  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>01:33
-!- travis-ci [~travis-ci@ec2-23-20-149-45.compute-1.amazonaws.com] has joined #shogun01:35
travis-ci[travis-ci] it's Fernando Iglesias'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/852298101:35
-!- travis-ci [~travis-ci@ec2-23-20-149-45.compute-1.amazonaws.com] has left #shogun []01:35
shogun-buildbotbuild #986 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/986  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>01:37
shogun-buildbotbuild #1141 of deb2 - static_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/114101:40
-!- lisitsyn [~lisitsyn@37.61.181.61] has left #shogun []01:42
shogun-buildbotbuild #1275 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/127501:53
-!- FSCV [~FSCV@50.7.50.60] has quit [Quit: Leaving]02:16
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has left #shogun ["PART #gsoc-gr :QUIT :Leaving."]02:17
shogun-buildbotbuild #1276 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1276  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>02:23
-!- foulwall [~user@2001:da8:215:6100:217b:2a2b:84cc:e3ee] has joined #shogun03:16
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]04:08
-!- nube [~rho@49.244.117.250] has quit [Quit: Leaving.]04:50
-!- foulwall [~user@2001:da8:215:6100:217b:2a2b:84cc:e3ee] has quit [Ping timeout: 245 seconds]05:41
-!- hushell [~hushell@67.160.139.116] has joined #shogun05:42
-!- nube [~rho@116.90.239.3] has joined #shogun06:11
-!- foulwall [~user@2001:da8:215:6100:2cc6:b0c6:86bb:3894] has joined #shogun06:37
-!- zxtx [~zv@cpe-66-68-190-37.austin.res.rr.com] has joined #shogun06:44
-!- foulwall [~user@2001:da8:215:6100:2cc6:b0c6:86bb:3894] has quit [Remote host closed the connection]07:59
-!- foulwall [~user@2001:da8:215:6100:2cc6:b0c6:86bb:3894] has joined #shogun08:22
-!- sonne|work [~sonnenbu@sams-office-nat.tomtomgroup.com] has quit [Quit: Leaving.]08:34
-!- sonne|work [~sonnenbu@sams-office-nat.tomtomgroup.com] has joined #shogun08:34
sonne|workfoulwall: hey there!08:35
foulwallhey sonne|work08:35
foulwalljust sent a pr08:36
sonne|workfoulwall: could you install this on the nn.7nn.de machine I gave you?08:37
foulwallok08:37
foulwallwait me a minute08:37
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun08:41
shogun-notifier-shogun-demo: foulwall :master * 3c408c5 / / (18 files): https://github.com/shogun-toolbox/shogun-demo/commit/3c408c5dae5bd6f893317458624dc5f9257dc58608:41
shogun-notifier-shogun-demo: 1. stretch the mouse input js out. 2. make the toy_data generator08:41
shogun-notifier-shogun-demo: modular 3. redesigned the left side panel, make it modular. 4. make the08:41
shogun-notifier-shogun-demo: content aligned center. 5. make an option for whether the coordinate08:41
shogun-notifier-shogun-demo: system has a background heatmap. 6. add the README08:41
shogun-notifier-shogun-demo: Soeren Sonnenburg :master * 5f28d8d / / (18 files): https://github.com/shogun-toolbox/shogun-demo/commit/5f28d8ddb65d097e0812f77d21a6c03ee81f370a08:41
shogun-notifier-shogun-demo: Merge pull request #8 from foulwall/refractor08:41
shogun-notifier-shogun-demo:08:41
shogun-notifier-shogun-demo: new toy_data generator, style, readme. and did some code refractor08:41
sonne|workfoulwall: just do git update there08:43
sonne|workfoulwall: then do08:43
sonne|workscreen -d -r08:43
sonne|workkill fapws3 running there08:44
sonne|work(ctrl+c)08:44
sonne|workstart it again (cursor up to get last cmd)08:44
sonne|workthen do ctrl+a d08:44
sonne|workto detach from the screen session08:44
foulwallsonne|work: weird08:46
foulwallsonne|work: when I run clustering demo, got segmentation fault08:46
sonne|workwhat is the error?08:48
foulwallsonne|work: wait me a moment, and I'll have a check08:50
-!- foulwall [~user@2001:da8:215:6100:2cc6:b0c6:86bb:3894] has quit [Ping timeout: 264 seconds]09:00
-!- nube [~rho@116.90.239.3] has quit [Ping timeout: 276 seconds]09:45
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun09:47
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun09:55
hushellDoes anybody knows the SWIG csharp interfaces?10:09
hushellI got an error when generating this function:10:10
hushellSWIGEXPORT void SWIGSTDCALL CSharp_StructuredModel_init_primal_opt(void * jarg1, double jarg2, void * jarg3, int size, double* jarg4, int rows, int   cols, double* jarg5, void * jarg6, int size, double* jarg7, int size, double* jarg8, void * jarg9)10:10
hushellwhich gives me multiple int size10:10
sonne|workhushell: yeah it is an annoying issue with our swig typemaps occurring whenever you have e.g. two SGVectors as argument10:15
sonne|workhushell: the workaround solution is to not expose the function in csharp by putting it in modular/*ignores*.i10:15
-!- nube [~rho@116.90.239.3] has joined #shogun10:23
-!- hushell [~hushell@67.160.139.116] has quit [Ping timeout: 268 seconds]10:24
-!- hushell [~hushell@c-67-160-139-116.hsd1.or.comcast.net] has joined #shogun10:27
hushellsonne|work: Thanks! This is where the problem comes from. I changed the name of init_opt() to init_primal_opt() but didn't update in modshogun_ignores.i10:29
-!- nube [~rho@116.90.239.3] has quit [Ping timeout: 248 seconds]10:29
hushellAre there such tricks for other modulars?10:29
sonne|workno10:30
van51sonne|work: hello10:34
van51sonne|work: I got a question while making the changes10:34
sonne|workvan51: shoot!10:34
van51sonne|work: while the SGVector is limited by the dim size, the dynamic array is limited by the number of tokens10:35
van51sonne|work: and isn't more probable to have a larger number of tokens than dim size?10:35
van51sonne|work: because the dim size is already supposed to be << than the original feature space dimension10:36
sonne|workvan51: usually dim is huge10:36
sonne|workas in you want 4 grams10:36
van51sonne|work: I get that originally dim would be huge, I just thought that the reduced dimension would be reasonable10:36
sonne|workvan51: IIRC the yahoo spam detector uses 32 bit for hash size10:37
sonne|workbut you are right10:38
sonne|workit depends on what you want to process10:38
sonne|workit would make sense to have 2 implementations - one for small dim10:38
sonne|workand one for large dim >> document size10:38
-!- nube [~rho@116.90.239.3] has joined #shogun10:40
van51sonne|work: maybe if I worked with a map, instead of a dynamic array?10:40
van51sonney2k: although I'm guessing it would be rather slower10:40
sonne|workit doesn't really matter *that* much for a converter btw10:41
sonne|workthe heavy part is the training later10:42
sonne|worksince you currently can assume everything fits in memory the large dim solution would just be fine10:42
van51sonne|work: ok then10:44
-!- lambday [67157f37@gateway/web/freenode/ip.103.21.127.55] has joined #shogun10:57
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun11:07
-!- mode/#shogun [+o iglesiasg] by ChanServ11:07
-!- nube [~rho@116.90.239.3] has quit [Read error: Connection reset by peer]11:16
-!- nube [~rho@116.90.239.3] has joined #shogun11:17
hushelliglesiasg: for my PR, every interface passed except python...11:19
@iglesiasghushell: yeah! I am checking at the moment11:19
@iglesiasggreat11:19
hushelliglesiasg: sorry for updating so late11:19
@iglesiasgpython's fail is still due to SerializableJsonFile11:19
@iglesiasgno problem :)11:19
hushellokay, I go to sleep now :)11:19
@iglesiasggood night then11:20
hushellCU guys11:20
-!- hushell [~hushell@c-67-160-139-116.hsd1.or.comcast.net] has quit [Quit: WeeChat 0.3.7]11:20
sonne|workwiking: could you please fix the install of json-c on travis?11:22
sonne|workwe still have this annoying bug11:23
@iglesiasgyeah please, that would be great to have the build again green11:23
sonne|workiglesiasg: I hope you merge his PR then?11:24
sonne|worknow I mean11:24
@iglesiasgsonne|work: yes, I am just having a last look atm11:24
sonne|workkthx11:24
@iglesiasgI like making the StructuredModel cleaner11:25
shogun-notifier-shogun: hushell :develop * 3142600 / / (33 files): https://github.com/shogun-toolbox/shogun/commit/3142600b5951013bee76e42f4ec7e63b2a5349a411:28
shogun-notifier-shogun: move risk() from StructuredModel to DualLibQPSOSVM11:28
shogun-notifier-shogun: Fernando Iglesias :develop * 2135513 / / (33 files): https://github.com/shogun-toolbox/shogun/commit/2135513c8e74c6a40a830036ef469debf27a2ab511:28
shogun-notifier-shogun: Merge pull request #1185 from hushell/develop11:28
shogun-notifier-shogun:11:28
shogun-notifier-shogun: move risk() from StructuredModel to StructuredOutputMachine11:28
sonne|workvan51: btw what is with your PR https://github.com/shogun-toolbox/shogun/pull/1186 ?11:33
sonne|workvan51: btw it would make a lot of sense to actually send a PR for the tokenizer & its unit test11:34
sonne|workvan51: that one is already good to merge11:34
sonne|workvan51: then also gsomix could start to use this11:36
van51sonne|work: ok11:36
van51sonne|work: although I expect to be done with the converter as well any time now11:36
sonne|workI am just saying small logical units please to avoid big bites11:36
sonne|workbut do as you want this time11:37
van51sonne|work: just for this time since you've already reviewed most of it11:37
van51sonne|work: I think it would also take me more time to find how to do it in git :P11:38
sonne|workvan51: np just finish it then11:38
shogun-buildbotbuild #1155 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1155  blamelist: hushell <hushell@hushell-U510.(none)>11:42
shogun-buildbotbuild #987 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/987  blamelist: hushell <hushell@hushell-U510.(none)>11:45
-!- nube [~rho@116.90.239.3] has quit [Quit: Leaving.]11:51
-!- travis-ci [~travis-ci@ec2-54-226-241-56.compute-1.amazonaws.com] has joined #shogun11:52
travis-ci[travis-ci] it's Fernando Iglesias'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/853423211:52
-!- travis-ci [~travis-ci@ec2-54-226-241-56.compute-1.amazonaws.com] has left #shogun []11:52
shogun-buildbotbuild #1156 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1156  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>11:53
-!- nube [~rho@116.90.239.3] has joined #shogun11:56
shogun-buildbotbuild #988 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/988  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>11:58
van51sonne|work: I updated the PR12:00
van51sonne|work: but I don't understand what's wrong with travis..12:00
-!- nube [~rho@116.90.239.3] has quit [Quit: Leaving.]12:01
@iglesiasgvan51: it died!12:02
sonne|workvan51: did you change data?12:03
sonne|workor who did and committed the new data version?12:03
@iglesiasgall the builds are gray right?12:03
sonne|workwell someone did commit a new data version12:03
sonne|workthat is not yet pushed12:03
-!- HeikoS [~heiko@nat-179-220.internal.eduroam.ucl.ac.uk] has joined #shogun12:03
-!- mode/#shogun [+o HeikoS] by ChanServ12:03
sonne|workto shogun-data12:03
sonne|workHeikoS: hey there12:03
@HeikoSsonne|work: hi!12:03
sonne|workI will be off for lunch soon but could you please open the form?12:03
van51sonne|work: no I don't think I changed data12:04
@HeikoSsonne|work: could you send the link again?12:04
* sonne|work sent it 5 times already12:04
* sonne|work looking12:04
@HeikoSmy irc does not store private chats12:04
sonne|workhttps://docs.google.com/a/google.com/forms/d/1zDCB8aozZOQJjqDm5rLsOso2xAbIn7OBcYGdLhL7OTs/viewform?pli=112:05
sonne|workHeikoS: no via email too12:05
sonne|workanyway12:05
@HeikoSsonne|work: will fill it in12:05
van51sonne|work: made it with a new commit and seems to work now12:05
sonne|workHeikoS: I think you did it last year too12:05
sonne|workHeikoS: do you know your Vendor ID Number12:05
sonne|workall the rest is totally clear12:06
sonne|workans simple12:06
@HeikoSvendor id, no idea12:06
sonne|workHeikoS: did we get one last year?12:06
@HeikoSExact Name Used for Your Mentoring Organization's Supplier Enrollment  ?12:06
sonne|workI think you registerd as supplier12:06
@HeikoSyeah12:06
@HeikoSlet me search12:06
sonne|workwell from http://www.google-melange.com/gsoc/accepted_orgs/google/gsoc201312:07
sonne|workShogun Machine Learning Toolbox12:07
@HeikoSno thats a different one12:07
sonne|workHeikoS: but have a look at Exact Name Used for Your Mentoring Organization's Supplier Enrollment12:07
sonne|workoh yes12:07
sonne|workmisread12:07
sonne|workwhat did you use last time?12:08
@HeikoSjust found a mail with that12:08
* sonne|work too12:08
@HeikoSis this the supplier enrollment form?12:08
sonne|workno12:08
sonne|workyou are a supplier already12:08
@HeikoSsonne|work: I changed my bank, no idea whether thats a problem12:08
sonne|workyes it is12:08
sonne|workHeikoS: so you have to do the supplier enrollment form again12:09
@HeikoSyep, searching12:09
sonne|workHeikoS: *sigh*12:09
@HeikoSthe supplier form doesn not contain bank data12:10
@HeikoSjust found it12:10
sonne|workSupplier resources can be found on our Supplier portal, including a12:10
sonne|worktool that lets you check your invoice status!12:10
sonne|workhttps://services.google.com/supplier-enrollment/12:11
sonne|workHeikoS: you can go there and click12:11
sonne|workI have previously enrolled as a Google supplier and am updating that information12:11
sonne|work12:11
sonne|workHeikoS: I guess then you can also immediately fill out the other form12:11
sonne|workvan51: weird then12:11
@HeikoSthe other?12:12
@HeikoSyou mean the PO one?12:12
sonne|workthis one Request for Google Summer of Code 2013 Mentoring Payments12:12
sonne|workhttps://docs.google.com/a/google.com/forms/d/1zDCB8aozZOQJjqDm5rLsOso2xAbIn7OBcYGdLhL7OTs/viewform?pli=112:12
sonne|workHeikoS: so do the supplier enrollment update with your new bank account first and then the mentoring payments one12:13
@HeikoSsonne|work: ah man, I forgot everything :(12:14
@HeikoSVAT?12:14
@HeikoSdid I give this?12:14
sonne|worklook you have the old email12:14
sonne|workwhere it is all written12:14
van51sonne|work: I g2g now12:15
van51got a dentist appointment :(12:16
van51sonne|work: hope everything is ok now12:16
sonne|workvan51: thanks!12:16
van51sonne|work: I'll look for your comments in a couple hours when I'll be back12:16
@HeikoSsonne|work: what are the time contraints on the PO?12:16
sonne|workASASP12:17
sonne|workit is like 5 minutes work12:17
@HeikoSsonne|work: so the thing is I recently quit my bank account and made another one, I sent the paperworks last week and it should be ready soon, but I dont have one yet12:17
@HeikoScould give one of my family (which I currently use) but thats not my name and might cause problems later12:17
@HeikoSI think the new account should be ready within 1-2 weeks12:18
sonne|workI see12:18
@HeikoSso thats why the question12:18
@HeikoSwhat are the time contraints12:18
sonne|workno in this case we should wait.12:18
sonne|workwell we cannot start to book flights before we have a PO12:18
sonne|workotherwise I just want to have all the paperwork off the table12:19
@HeikoSsonne|work: so in principal we could  do it over another account since we can donate it to the ev this year (which removes it from tax) but thats risky if things dont work out12:19
@HeikoSso lets just wait a bit, should be soon with the bank12:19
@HeikoSI already applied for the account ages ago, just needed to send my identification from which I did last week12:20
@HeikoSI will call them on Monday and ask12:20
sonne|workHeikoS: then please do that all yourself12:20
sonne|workI forget this stuff12:20
@HeikoSsonne|work: sure, btw I will probably stay in the US for a bit longer, about 1 week or so12:21
@HeikoSwill search for a flat/room on airbnb that should be cheaper than a hotel12:21
@HeikoSand maybe rent a car12:21
shogun-buildbotbuild #1277 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/127712:27
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has quit [Quit: Leaving.]12:39
shogun-buildbotbuild #1278 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1278  blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com>12:43
shogun-notifier-shogun: iglesias :develop * b83fa92 / .gitignore: https://github.com/shogun-toolbox/shogun/commit/b83fa92d34d5fd6e0689e57e1c37e635ce7f83c413:20
shogun-notifier-shogun: Update gitignore with files created by unit tests13:20
shogun-notifier-shogun: json files and combined_kernel.weights13:20
shogun-notifier-shogun: iglesias :develop * 9464e9e / tests/unit/io/Serialization_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/9464e9e5d7d52a9e4f2b912074459fb5dbaac68813:20
shogun-notifier-shogun: Fix typo in file extension: from hfd5 to hdf513:20
shogun-notifier-shogun: Fernando Iglesias :develop * e5d5b16 / .gitignore,tests/unit/io/Serialization_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/e5d5b1674cb2cd9e50fd4443ff3ab714019783df13:20
shogun-notifier-shogun: Merge pull request #1194 from iglesias/develop13:20
shogun-notifier-shogun:13:20
shogun-notifier-shogun: Update gitignore and typo fix13:20
@iglesiasgsonne|work: any special reason not to use commas after macros like SG_REF, SG_UNREF, SG_PRINT, etc?13:22
thoralfFound another double-free in shogun...13:23
thoralfCOnlineLibLinear *svm = new COnlineLibLinear(C);13:23
thoralfsvm->train();13:23
thoralf{ SGVector<float32_t> w = svm->get_w(); }13:23
thoralf{ SGVector<float32_t> w = svm->get_w(); }13:23
shogun-buildbotbuild #989 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/989  blamelist: iglesias <fernando.iglesiasg@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>13:23
thoralfThe brackets are only used to close the scope and force destruction of w.13:24
@iglesiasgthoralf: a double call to SG_UNREF?13:24
thoralfiglesiasg: No.  w gets initialized by SGVector(array, len)13:24
thoralfiglesiasg: And upon destruction, it frees memory behind svm.w13:25
thoralfget_w() is just: return SGVector<float32_t>(w, w_dim);13:25
thoralfWhich does not reference counting on *w*.  It simply passes a reference, which is then freed by SGVector... the state is already corrupted after destruction of the first w above.13:27
@iglesiasgthoralf: can you tell me which get_w you mean?13:28
@iglesiasgthe one in LinearMachine?13:28
thoralfCOnlineLinearMachine::SGVector<float32_t> get_w()13:28
@iglesiasgthoralf: why doesn't it reference count on w?13:29
@iglesiasgshogun/machine/StructuredOutputMachine.cpp13:29
@iglesiasgshit, sorry13:30
@iglesiasgI meant13:30
@iglesiasgSGVector(T* v, index_t len, bool ref_counting=true)13:30
shogun-buildbotbuild #1157 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1157  blamelist: iglesias <fernando.iglesiasg@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>13:31
@iglesiasgsonne|work: it seems that SerializableJsonFile import error happens in the buildbot as well as in travis13:32
thoralfiglesiasg: I don't understand.  *w* is a array of floats - SG_REF can only be called on SGReferencedData objects, right?13:33
-!- foulwall [~user@2001:da8:215:6110:dd27:e03b:e5a:cb89] has joined #shogun13:34
foulwallhey sonne|work, sorry for the campus network crash...13:35
@iglesiasgthoralf: no. SGReferencedData are objects for which automatic memory management is provided13:38
@iglesiasgso there is no need to call SG_REF and SG_UNREF with them13:38
@iglesiasgSG_REF and SG_UNREF are used for all SGObjects13:39
thoralfiglesiasg: Okay, but then there is no way to track w's references.13:39
@iglesiasgthoralf: I am not sure I understand, can you ellaborate?13:41
@iglesiasgI mean, if a SGVector is created via SGVector(T*,index_t)13:41
@iglesiasgthen I believe that there is no new memory allocated13:41
thoralfCorrect.  SGVector(w,len)13:41
@iglesiasgapart from the attributes of the SGVector (the pointer, the length, etc)13:42
thoralfbut w is still beeing used by onlinelinearmachine13:42
thoralfSo if SGVector is destructed, w is freed.13:42
@iglesiasgwhy is it a problem?13:42
@iglesiasgno13:42
thoralfBut it is!13:42
thoralfI see it in the debugger. ;)13:42
shogun-buildbotbuild #1158 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1158  blamelist: iglesias <fernando.iglesiasg@gmail.com>13:42
@iglesiasgw is freed if there is no other reference pointing13:43
@iglesiasgaham! then there should be a bug13:43
thoralfI think there are two solutions:13:43
@iglesiasgok I think maybe I get it now13:43
thoralf1. copy w into new memory13:44
thoralf2. onlinelinearmachine maintains an own sgvector to pass around.13:44
@iglesiasgyes13:44
shogun-buildbotbuild #990 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/990  blamelist: iglesias <fernando.iglesiasg@gmail.com>13:44
thoralflike linearmachine does13:44
@iglesiasgI think the solution is the second, that the onlinelinearmachine uses an sgvector13:44
@iglesiasgor13:44
@iglesiasgthat the get method creates the vector with the third argument to false13:45
thoralfOkay, I just tried copying w - and it seems to work without leaking.13:46
@iglesiasghow without copying w?13:47
thoralfiglesiasg: Then I need to add a private member to store a sgvector instance.13:47
thoralfiglesiasg: No, I would need to refactor the class: replacing (float * w, int w_dim) by (sgvector w) to make sure there's no confusion which one to keep up-to-date.13:55
thoralfiglesiasg: (float * w, int w_dim) gets resized while learning.13:55
@iglesiasgthoralf: I agree with you, using SGVector is better13:55
thoralfiglesiasg: The code is nearly untested - it would be suicide to start refactoring. ;)13:57
@iglesiasgyes, it may be a good idea to add a couple of unit tests first and then refactor13:57
thoralfiglesiasg: If OnlineLinearMachine turns out to be useful, I see what I can do.14:04
shogun-buildbotbuild #1279 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/127914:05
-!- travis-ci [~travis-ci@ec2-23-22-49-203.compute-1.amazonaws.com] has joined #shogun14:05
travis-ci[travis-ci] it's Fernando Iglesias'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/853650014:05
-!- travis-ci [~travis-ci@ec2-23-22-49-203.compute-1.amazonaws.com] has left #shogun []14:05
@iglesiasgthoralf: could you please write a github issue anyway? In case we don't forget there is something weird there14:07
thoralfiglesiasg: On the way.14:07
@iglesiasgthank you!14:08
shogun-buildbotbuild #1280 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1280  blamelist: iglesias <fernando.iglesiasg@gmail.com>14:32
sonne|workfoulwall: so it works now?14:40
-!- nube [~rho@49.244.121.237] has joined #shogun14:46
sonne|workfoulwall: ping?14:56
sonne|workthoralf: yeah I think we have no option - we need to switch to using SGVector completely15:01
thoralfiglesiasg: I sent a pull request with my hotfix (copying memory) - I won't mind if you reject, but at least it's linked to the ticket.15:01
sonne|workthoralf: I started with this but it is *a lot* of work15:01
thoralfsonne|work: I know.  That's why I'm avoiding this... :(15:02
-!- lisitsyn [~lisitsin@mxs.kg.ru] has left #shogun []15:10
-!- lambday [67157f37@gateway/web/freenode/ip.103.21.127.55] has quit []15:37
-!- iglesiasg [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]15:38
-!- foulwall [~user@2001:da8:215:6110:dd27:e03b:e5a:cb89] has quit [Ping timeout: 264 seconds]15:48
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun15:48
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]16:20
sonne|workvan51: teeth polished?16:20
van51sonne|work: more like tooth removed!16:21
sonne|workvan51: so you have a fancy smile now?16:21
van51i;m even prettier :O16:21
van51sonne|work: should I make the dynamic array size a public field?16:48
sonne|workvan51: why?16:56
sonne|worklets just waste this 1MB16:57
sonne|workactually 416:57
van51sonne|work: so I just make it 1024*1024?16:59
sonne|workyeah17:00
-!- sonne|work [~sonnenbu@sams-office-nat.tomtomgroup.com] has quit [Quit: Leaving.]17:02
@sonney2kvan51, done?17:30
van51sonney2k: yea I pushed17:30
@sonney2kvan51, let me check17:31
@sonney2kohh funkloch17:32
@sonney2kvan51, fine with it17:35
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun17:35
shogun-notifier-shogun: van51 :develop * c868efb / / (11 files): https://github.com/shogun-toolbox/shogun/commit/c868efbc9dc180433fce5b8bf04f77d56fc59c2a17:35
shogun-notifier-shogun: Basic classes17:35
shogun-notifier-shogun: Soeren Sonnenburg :develop * 1a53852 / / (11 files): https://github.com/shogun-toolbox/shogun/commit/1a53852fdda84022f89dd1629e186e07cfedbbb317:35
shogun-notifier-shogun: Merge pull request #1186 from van51/feature/hashing17:35
shogun-notifier-shogun:17:35
shogun-notifier-shogun: Hashing N-gram / String Converter and tokenizer17:35
@sonney2kvan51, ohh I just see n-bits - could you make 20 the default17:36
@sonney2knot 5017:36
@sonney2kthat would be a bit too much17:36
van51sonney2k: oops it was 50?17:36
@sonney2kvan51, now could you do that with dot-features17:36
van51sonney2k: I already have a class ready17:37
@sonney2kvan51, so you have string features underneath and do the dense_dot and add_to_dense_vec functions17:37
@sonney2kin the very same way you did that before17:37
van51sonney2k: I wanted to add some covariance and average methods that I saw that exist17:37
van51sonney2k: and I was looking for a collection to test it17:37
@sonney2konce that is done we try some simple example distinguishing say german from english text which you can then extend to say 30-50 languages17:38
@sonney2kvan51, you can create a simple test for add_to_dense_vec17:38
shogun-buildbotbuild #991 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/991  blamelist: van51 <vangelis_51@hotmail.com>17:38
@sonney2kvan51, just use get_computed_features17:38
@sonney2kand compare against what you get with your converter17:39
@sonney2kvan51, similar with dense_dot17:39
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun17:39
@sonney2kwith example this should take you next week (but no more) I would estimate17:39
@sonney2kalright17:39
@sonney2kgtg17:39
@sonney2kcu17:39
shogun-buildbotbuild #1159 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1159  blamelist: van51 <vangelis_51@hotmail.com>17:46
shogun-buildbotbuild #992 of cyg1 - libshogun is complete: Failure [failed configure]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/992  blamelist: Soeren Sonnenburg <sonne@debian.org>17:50
shogun-buildbotbuild #1160 of bsd1 - libshogun is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1160  blamelist: Soeren Sonnenburg <sonne@debian.org>17:58
-!- thoralf [~thoralf@enki.zib.de] has quit [Quit: Konversation terminated!]18:02
-!- travis-ci [~travis-ci@ec2-54-226-241-56.compute-1.amazonaws.com] has joined #shogun18:10
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/854459518:10
-!- travis-ci [~travis-ci@ec2-54-226-241-56.compute-1.amazonaws.com] has left #shogun []18:10
shogun-buildbotbuild #1281 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/128118:21
-!- lisitsyn [~lisitsyn@5-28-21-24.clients.tlt.100megabit.ru] has joined #shogun18:41
shogun-buildbotbuild #1282 of deb3 - modular_interfaces is complete: Failure [failed test python_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1282  blamelist: Soeren Sonnenburg <sonne@debian.org>18:48
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has left #shogun ["QUIT :Leaving."]19:38
-!- HeikoS [~heiko@nat-179-220.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]20:04
-!- nube [~rho@49.244.121.237] has quit [Quit: Leaving.]20:13
-!- papan [704f246a@gateway/web/freenode/ip.112.79.36.106] has joined #shogun20:18
papani'm new to shogun............ how to use matlab binding20:19
papan?????20:19
@sonney2klisitsyn, did you receive my email?20:26
lisitsynsonney2k: just started spamming people20:26
lisitsynsonney2k: took a while to put names here20:27
lisitsyn:D20:27
lisitsynall out20:28
-!- HeikoS [~heiko@nat-179-220.internal.eduroam.ucl.ac.uk] has joined #shogun20:29
-!- mode/#shogun [+o HeikoS] by ChanServ20:29
lisitsynsonney2k: HeikoS: wiking: did you receive a mail?20:30
-!- papan [704f246a@gateway/web/freenode/ip.112.79.36.106] has quit [Ping timeout: 250 seconds]20:32
-!- papan [7563149b@gateway/web/freenode/ip.117.99.20.155] has joined #shogun20:33
@sonney2klisitsyn, just arrived it seems20:33
@sonney2klisitsyn, you bad russian spammer!20:33
lisitsynsonney2k: haha yes vodka spamming20:34
papanis there any compiled copy available.... of shogun matlab20:34
lisitsynpapan: hey20:34
lisitsynpapan: by compiled you mean binaries, right?20:34
papanyes......20:34
papanyes......... compiled...20:34
lisitsynunfortunately no20:34
papanyes binaries20:34
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]20:35
papansee i've windows..... 'n i need matlab interface....20:35
papanwhat can i do?????20:35
papan@lisitsyn20:36
lisitsynpapan: you'd need cygwin to compile there20:36
papanbut cygwin itself is large.... anything else....20:36
papanthere????????????20:37
lisitsynpapan: no, I regret but that's a requirement to compile shogun under windows20:37
@sonney2kpapan, try linux or osx20:38
@sonney2kmuch easier to get shogun to work under these systems20:38
-!- HeikoS [~heiko@nat-179-220.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]20:39
papanone more thing.... is sg.oct file is all what i need?20:39
lisitsynsg.oct is used for octave I think20:40
papansg.mexglx is for matlab?????????????????????20:41
@sonney2kshogun-buildbot, force build shogun_modular20:45
shogun-buildbotno such builder 'shogun_modular'20:45
@sonney2kshogun-buildbot, force build 'deb3 - modular_interfaces'20:45
shogun-buildbotbuild forced [ETA 36m41s]20:45
shogun-buildbotI'll give a shout when the build finishes20:45
@sonney2kpapan, yes20:45
papan@sonney2k ........ then sg.mexglx file is all what i need?????????20:46
@sonney2k?20:52
@sonney2kunder linux?20:52
-!- papan [7563149b@gateway/web/freenode/ip.117.99.20.155] has quit [Ping timeout: 250 seconds]20:52
lisitsynsonney2k: due to timezones my flight takes 35 minutes :D20:54
@sonney2klisitsyn, close to teleportation!20:59
lisitsynsonney2k: 2.50 actually20:59
lisitsynor 2.3520:59
lisitsynsomewhere around20:59
-!- HeikoS [~heiko@nat-179-220.internal.eduroam.ucl.ac.uk] has joined #shogun21:00
-!- mode/#shogun [+o HeikoS] by ChanServ21:00
@sonney2kHeikoS, thanks for the mail draft!21:00
-!- HeikoS [~heiko@nat-179-220.internal.eduroam.ucl.ac.uk] has quit [Client Quit]21:01
@sonney2kwiking, hey there please fix json-c installs on the bsd bot21:02
@sonney2kwiking, and on github21:03
shogun-buildbotbuild #1283 of deb3 - modular_interfaces is complete: Failure [failed compile lua_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/128321:03
@sonney2klisitsyn, btw it is freezing cold in germany now :)21:06
@sonney2kshogun-buildbot, force build 'deb3 - modular_interfaces'21:06
shogun-buildbotbuild forced [ETA 36m41s]21:06
shogun-buildbotI'll give a shout when the build finishes21:06
lisitsynsonney2k: gut!21:06
@sonney2klisitsyn, we had like 13 C max recently21:06
@sonney2kthe week before that 35 C21:06
@sonney2kcrazy wheather21:06
lisitsynsonney2k: going to tell my gf to take some clothes for that then!21:06
@sonney2knext week summer is supposed to come back21:09
@sonney2k25 C or so21:09
@sonney2khmmhh luuuua21:10
shogun-buildbotbuild #1284 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/128421:52
-!- van51 [~van51@athedsl-408350.home.otenet.gr] has joined #shogun23:49
--- Log closed Sat Jun 29 00:00:09 2013