Open in new window / Try shogun cloud
--- Log opened Mon Aug 20 00:00:17 2012
shogun-buildbot_build #400 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/400  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>00:05
-!- hoijui [~hoijui@dslb-092-078-182-230.pools.arcor-ip.net] has quit [Ping timeout: 245 seconds]00:18
shogun-buildbot_build #292 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/292  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>00:29
-!- gsomix_ [~gsomix@178.45.41.90] has quit [Ping timeout: 252 seconds]00:30
shogun-buildbot_build #401 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/401  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>00:38
shogun-buildbot_build #316 of deb2 - static_interfaces is complete: Failure [failed test libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/316  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>00:40
blackburn2CIA-36: so you are back but will not tell us anything?00:45
blackburn2:D00:45
shogun-buildbot_build #293 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/293  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>00:54
shogun-buildbot_build #294 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/294  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>00:57
shogun-buildbot_build #402 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/402  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>01:04
shogun-buildbot_build #317 of deb2 - static_interfaces is complete: Failure [failed test libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/317  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>01:06
-!- blackburn2 [~blackburn@109.226.112.41] has quit [Quit: Leaving.]01:37
shogun-buildbot_build #403 of deb3 - modular_interfaces is complete: Failure [failed test libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/403  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>01:41
shogun-buildbot_build #295 of bsd1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/295  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>01:46
shogun-buildbot_build #404 of deb3 - modular_interfaces is complete: Failure [failed test libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/404  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>02:07
shogun-buildbot_build #405 of deb3 - modular_interfaces is complete: Failure [failed test libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/405  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>02:41
shogun-buildbot_build #318 of deb2 - static_interfaces is complete: Failure [failed test libshogun]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/318  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>02:43
shogun-buildbot_build #60 of nightly_none is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/6003:02
sr___can shogun perform some backward propagation neural network kind of stuff?03:03
shogun-buildbot_build #71 of nightly_default is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/7103:39
shogun-buildbot_build #56 of nightly_all is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_all/builds/5604:11
-!- puffin444 [180317bb@gateway/web/freenode/ip.24.3.23.187] has quit [Ping timeout: 245 seconds]04:14
-!- emrecelikten [~emre@trir-4d0d87fa.pool.mediaWays.net] has quit [Ping timeout: 244 seconds]04:26
-!- emrecelikten [~emre@trir-5d800761.pool.mediaWays.net] has joined #shogun04:41
-!- hoijui [~hoijui@dslb-092-078-182-230.pools.arcor-ip.net] has joined #shogun07:57
-!- hoijui [~hoijui@dslb-092-078-182-230.pools.arcor-ip.net] has quit [Quit: Leaving]08:28
@sonney2ksr___, no none of us here is an expert for NN09:13
@sonney2ks09:13
@sonney2kso we better not pretend that we are!09:14
shogun-buildbot_build #296 of bsd1 - libshogun is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/29609:41
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun10:05
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun10:08
shogun-buildbot_build #406 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/406  blamelist: Soeren Sonnenburg <sonne@debian.org>10:11
blackburnuh oh what a pain to get back to job noo10:14
shogun-buildbot_build #319 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/31910:17
-!- uricamic [~uricamic@2001:718:2:1634:dc2c:b66c:3452:bc42] has joined #shogun10:17
-!- thegruel [c1aa8a84@gateway/web/freenode/ip.193.170.138.132] has joined #shogun10:25
thegruelHi all10:25
blackburnhey10:25
thegruelThis time, I'm having trouble with CLibSVM's train() using python_modular. It sometimes returns "True", but has bias==NaN and empty SVs. Any tips?10:26
blackburnwhat is kernel you use?10:30
blackburnC?10:30
thegruelCustomKernel (precomputed linear, actually), but PosDef-ness should not be an issue, even after adding a ridge it failed.10:31
blackburnis kernel properly normalized?10:32
blackburnand did you try other solvers?10:32
thegruelYes, it's normalized. But that shouldn't matter, right? If it's the kernel's fault, the training shouldn't return "True", right?10:33
thegruelOther solvers instaed of LibSVM you mean? No, not yet. All the other SVMs version were super-slow with large kernel matrices.10:33
blackburnthegruel: you may try to check debug prints however I am unsure if it would help10:37
thegruelMSG_DEBUG says:10:41
thegruel[INFO] 96 trainlabels [INFO] Computing gradient for initial set of non-zero alphas [INFO]  optimization finished, #iter = 1 [INFO] nu = -nan [INFO] obj = -nan, rho = -nan [INFO] nSV = 0, nBSV = 0 [INFO] Total nSV = 010:41
thegruelOops... Sorry, there seems to be a NaN in the kernel matrix... How did that get there???10:43
blackburnhehe10:43
thegruelOkay, issue solved (or rather: false alarm) thanks!10:53
-!- thegruel [c1aa8a84@gateway/web/freenode/ip.193.170.138.132] has left #shogun []10:55
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has quit [Quit: Page closed]12:00
shogun-buildbot_build #407 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/407  blamelist: iglesias <fernando.iglesiasg@gmail.com>12:27
@sonney2khttps://www.youtube.com/watch?v=GSqrmuZQxz012:45
@sonney2kwiking, n4nd0 could you please fix those 2 warnings http://shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/320/steps/compile%20libshogun/logs/warnings%20%282%29 ?12:46
-!- blackburn [5bdfb203@gateway/web/freenode/ip.91.223.178.3] has joined #shogun12:48
shogun-buildbot_build #408 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/408  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>12:59
shogun-buildbot_build #293 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/293  blamelist: Soeren Sonnenburg <sonne@debian.org>13:05
shogun-buildbot_build #409 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/409  blamelist: Soeren Sonnenburg <sonne@debian.org>13:31
n4nd0sonney2k: sure13:50
blackburnoh I didn't know big displays are so cool14:00
blackburnwant 21" notebook :D14:00
-!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has joined #shogun14:13
yoohi all14:13
blackburnhey14:13
yoohow are you ?14:13
blackburnyoo: sorry I forgot to add multiclass things to xvaloutput14:13
blackburnI am setting up things for shogun dev at my job hehe14:14
yooblackburn: yea, since you didnt sent me the confirmation mail, I thought you didnt have time yet for this14:14
blackburnI can do that today I think14:15
yooblackburn: great14:15
yooblackburn: I tried last night to compile shogun with mosek support but I cant make shogun find it14:15
blackburnyeahh that is an issue14:16
yooblackburn: mosek is well installed14:16
yooblackburn:  ahah ok then :p14:16
n4nd0yoo: you can make a couple of modifications to your .config file in order to configure with MOSEK support14:17
n4nd0you'd do them after running ./configure and note that those will be removed every time you run ./configure again14:17
yooblackburn: when multiclass thgs will be added to xvaloutput, I will probably make up-to-date c++ examples and maybe create new ones14:18
yoothanks, i'll try it14:18
n4nd0yoo: let me know if you need to know what these modifications are14:19
yoon4nd0: do you have a log of your .config file ?14:20
n4nd0yoo: I already explained these changes by mail to other people, I can just forward that mail to you if you like14:20
yoon4nd0: yes, it will be kind of you!14:21
yoothereisnoknife@gmail.com14:21
n4nd0blackburn, sonney2k: I've just issued a PR with the warning fixed14:21
n4nd0yoo: hehe funny address :)14:21
blackburnI want to have a switch 'merge all PRs automatically'14:21
yooI am in the shogun mailing list by the way14:22
n4nd0yoo: sent14:23
yoon4nd0: thx14:23
yooclang++ compilation :p14:24
n4nd0yeah14:25
n4nd0try with the standard first14:25
n4nd0I had to use clang++ but it seemed that the problem was in my machine14:25
n4nd0sonney2k could compile the things with the compiler used in the makefile14:25
yoook14:26
n4nd0see you guys14:37
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]14:38
shogun-buildbot_build #410 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/410  blamelist: iglesias <fernando.iglesiasg@gmail.com>14:52
blackburnsonney2k: any idea what is going on with CIA-36 ?15:10
shogun-buildbot_build #411 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/411  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>15:24
shogun-buildbot_build #323 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/323  blamelist: Soeren Sonnenburg <sonne@debian.org>15:28
-!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has quit [Quit: Page closed]15:51
shogun-buildbot_build #412 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/41216:01
shogun-buildbot_build #324 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/32416:12
shogun-buildbot_build #294 of cyg1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/294  blamelist: iglesias <fernando.iglesiasg@gmail.com>, Sergey Lisitsyn <lisitsyn.s.o@gmail.com>16:46
-!- blackburn is now known as _________16:47
-!- _________ is now known as _____________16:48
shogun-buildbot_build #413 of deb3 - modular_interfaces is complete: Failure [failed test ruby_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/413  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>17:08
-!- blackburn [~blackburn@83.234.54.234] has joined #shogun17:46
shogun-buildbot_build #414 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/41417:51
-!- blackburn [~blackburn@83.234.54.234] has quit [Ping timeout: 265 seconds]17:54
-!- blackburn [~blackburn@188.168.2.3] has joined #shogun17:55
-!- wiking [~wiking@huwico/staff/wiking] has quit [Read error: Connection reset by peer]17:59
-!- wiking [~wiking@info2k1.hu] has joined #shogun18:06
wikingmmmm18:21
wikingthis died :)18:21
-!- wiking [~wiking@info2k1.hu] has quit [Changing host]18:21
-!- wiking [~wiking@huwico/staff/wiking] has joined #shogun18:21
uricamichi18:23
wikinghey !18:25
uricamiccan I submit some data to shogun-data? MINST database in svmlight format18:25
blackburnif it is not that big feel free :)18:25
uricamicI want to submit benchmark that needs this data18:25
blackburnwhat is size?18:26
uricamic3 files, trn file cca 108MB, validation and test about 10MB each18:26
blackburnoh that's big18:26
uricamicyep, the train set is quite big18:27
uricamicor I could just send the example and one would need to download the data from somewhere else18:27
uricamicit is 60k vectors, of dimension 78418:28
blackburnwhat if we compress it?18:31
uricamicthat could help18:31
uricamicand is it still possible to read it directly like any other svmlight formated file?18:32
blackburnthere should be some way using python (if you are going to use it)18:33
uricamicyep18:35
uricamicit is python example18:35
uricamicok, compressed it is 32MB total (3 files)18:44
blackburnuricamic: other way is to implement downloader18:45
uricamicblackburn: what do you mean by that?18:46
blackburnyou can post it on mldata.org and implement a script that downloads18:46
@sonney2kuricamic, why don't you compress the stuff w/ bz2?18:46
blackburnthe dataset on the first run18:46
uricamicI see, is there already such example?18:47
@sonney2kblackburn, what is wrong with CIA-3618:47
@sonney2k?18:47
blackburnsonney2k: no commit messages anymore18:47
uricamicsonney2k: compressed zip files are now 27.5 MB and 2x 9.1MB18:47
blackburnI miss they18:47
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun18:49
n4nd0hi all18:50
n4nd0what's the problem with our friend shogun-buildbot_?18:51
n4nd0retry exception slave lost!?18:51
-!- CIA-36 was kicked from #shogun by sonney2k [CIA-36]18:57
-!- CIA-52 [cia@198.71.88.9] has joined #shogun18:57
@sonney2kCIA-52, behave please!18:57
@sonney2kn4nd0, no idea - might be that the server the vm is running on is overloaded / got shut down...18:57
@sonney2kI wouldn't be surprised 32 C here...18:58
n4nd0sonney2k: aham I see18:58
n4nd0it was above 45 a couple of days in Spain while I was there18:58
n4nd0record!18:58
blackburnoh I fell asleep while writing commit message LOL19:01
n4nd0haha19:01
n4nd0that cannot be true!19:01
n4nd0blackburn: seriously?19:01
blackburnyeahh I had to wake up early today19:02
@sonney2kblackburn, wtf is _____________ ?19:02
n4nd0_____________: hey there, what's up?19:04
blackburnsonney2k: it is my computer at job19:05
n4nd0:)19:06
n4nd0blackburn: are you back to the job now?19:06
blackburnyeah :(19:06
@sonney2kblackburn, btw mailinglist has 233 subscribers nowadays...19:06
blackburnis it a big number?19:07
@sonney2kwell we started with 0 so it is inf times bigger :)19:08
@sonney2kblackburn, how big do you want it to be?19:08
@sonney2kis 1000 big?19:08
blackburnsonney2k: what was the number year before?19:08
@sonney2kblackburn, don't remember19:08
@sonney2kbut I think I told you back then19:08
blackburnyeah but I have not much better memory capacity so it is unknown19:09
blackburndenote it by x okay19:09
@sonney2kblackburn, ok I had a look at the shogun jmlr paper19:10
@sonney2kso end of 2009 it was 100019:10
@sonney2k10019:10
blackburn1000 or 100? :)19:10
@sonney2k100,00019:10
@sonney2k;-)19:10
@sonney2k10019:10
blackburnI see19:10
blackburnwell not the total count but quality matters19:11
blackburnsome great guys are subscribed19:11
-!- yoh_ [~yoh@washoe.dartmouth.edu] has joined #shogun19:11
@sonney2kI am actually not sure how big the ML community is19:11
yoh_sonney2k: Hi ;)19:11
@sonney2kmaybe 1000-10000 people or?19:11
n4nd0hey yoh_, are you yoo from this morning?19:12
blackburnno he isn't :)19:13
n4nd0who wanted to compile with MOSEK support19:13
n4nd0ups OK19:13
n4nd0yoh_: sorry19:13
yoh_nah -- I am from yesterday's evening ;)19:14
blackburnyoh_: it's been a while - when did you join us last time?19:15
yoh_sonney2k: so -- what are the chances/ways to get recent version to wheezy?19:15
yoh_blackburn: do not know really ;)19:15
blackburnif n4nd0 doesn't know you then it is greater than one year19:15
blackburnand actually I remember you were chatting with soeren in march, 2011 last time :D19:16
yoh_blackburn: for the curious ones: --- Log opened Thu Mar 24 14:04:20 201119:16
blackburnyeah19:16
yoh_blackburn: yeah19:16
yoh_;-)19:16
n4nd0blackburn: impressive memory! and you were saying you don't have much memory capacity ...19:16
yoh_I just whined to Soeren in email and he suggested to chat online19:16
yoh_so here I am19:16
blackburnyoh_: we are planning to release 2.0 September the first - our favourite day19:17
@sonney2kyoh_, ahh yaroslav19:17
@sonney2khey19:17
blackburnhahah19:17
yoh_blackburn: that is great... atm wheezy is scheduled to be released with 0.9 ! ;)19:17
blackburnsonney2k: werre you computing scores to estimate who could that be?19:18
blackburn0.9? oh nice19:18
@sonney2kyoh_, let me first explain the problems I have with packaging:19:18
@sonney2kbiggest problem of all is that compiling shogun's interfaces w/ gcc would require >3GB of memory19:18
yoh_sonney2k: I think I know the core -- building all those interfaces across platforms19:18
yoh_ah... and that ;)19:18
@sonney2kyoh_, and debian builders limit things to (I don't recall) but say 2GB19:19
yoh_I believe I ran into similar problems with some other package that there were simply no RAM avail on that architecture (s390?...)19:19
yoh_that is I why I suggest19:19
@sonney2kif we had a program that could split up a single 40MB .cpp file into multiple ones we could compile with gcc and all good19:20
yoh_to check with ftpmasters if they would simply allow removal of shogun from wheezy19:20
yoh_first and then where it builds -- it builds and gets into wheezy19:20
@sonney2kbut since this is not available we just use clang19:20
@sonney2kclang works just fine on some archs (requiring much less memory)19:20
@sonney2kbut obviously has bugs19:20
@sonney2kyoh_, I don't recall by heart but IIRC it crashes on powerpc and creates a weird linker error on s39019:21
@sonney2kboth bugs in clang but I simply don't have the time to prove that with isolated examples19:21
@sonney2kyoh_, then the other issue with python:19:21
@sonney2kI certainly never manually created anything under /usr/*/python*19:22
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 244 seconds]19:22
@sonney2khowever, I switched python packaging to hmmhh what was it dh_python2 or sth?19:22
@sonney2kso I guess there is some issue when the old package is still using the other debian python packaging system19:23
@sonney2kyoh_, so no idea about those errors19:23
@sonney2kthat's all19:23
@sonney2kyoh_, any suggestions?19:24
yoh_once again -- with all those problems on fancy architectures -- I would have talked to ftpmasters to see if they would agree to just remove shogun from wheezy altogether and allow it propagate from sid19:25
-!- uricamic [~uricamic@2001:718:2:1634:dc2c:b66c:3452:bc42] has quit [Quit: Leaving.]19:25
yoh_the same happened for ANTs and openmeeg19:25
yoh_but they were removed even before the freeze IIRC19:25
@sonney2ksure that would work19:26
yoh_as for python, I guess you meant the one I have filed: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=682166    just needs a look (deboostrap squeeze in a directory, install shogun) and go through upgrade to see WTF .  But probably better at once to check upgrade to 1.1.0 version in sid19:27
@sonney2kyoh_, yeah but I don't know what to do about it19:28
@sonney2kyoh_, that is dh_python2 (or the old system trying to remove stuff)19:29
yoh_I haven't looked into details of this one... but I have ran into it while upgrading from 1.0.0 not 0.9.3... and that one doesn't have those directories (checked on squeeze system)19:30
yoh_so might it be specific to upgrades from 1.0.0?19:30
yoh_ah damn -- I just saw it  -- you are not in wheezy already19:32
@sonney2kyoh_, it is weird: squeeze has19:36
@sonney2kKernel.py -> /usr/share/pyshared/shogun/Kernel.py19:36
@sonney2ketc in there19:36
yoh_right, so no /usr/lib/python2.6/dist-packages/shogun, which I guess came with 1.0.0 version right?19:38
@sonney2kmomplz kids crying19:39
yoh_as for http://release.debian.org/migration/testing.pl?package=shogun  - I am still not sure -- why does it wait for those architectures... or if it doesn't  -- why claims dependency on octave which you disabled entirely?19:39
yoh_and then "mono-2.0-devel " dependency -- was that automatically placed? may be just needs a rebuild?19:40
yoh_so if I were you I was bold and talked to ftp/release teams to figure out what is needed to be done19:41
@sonney2kyoh_, it would build on octave w/o trouble btw19:41
@sonney2kproblem is just that octave requires even more memory19:41
yoh_after testing upgrade path first for 0.9.3 -> 1.1.0-5 -- may be there is no problem19:41
@sonney2k(C++ templated code...)19:42
yoh_but now it is disabled (you commented out binary package right)?19:42
@sonney2kyoh_, I think so yes19:42
yoh_I doubt that you should worry about that ATM -- imho minimal plan should be to get current version as is into wheezy and then may be fix bugs up (e.g. if there is the upgrade problem ;) )19:43
@sonney2kyoh_, if I had time I would redo the packaging and split it up to compile just libshogun for all archs using gcc19:43
@sonney2kthen do the same for all interfaces separately...19:43
@sonney2ksure19:43
yoh_if you had a chance, not time ;)  those evil ftp/release- teams might simply not allow such radical changes19:43
yoh_;-)19:43
@sonney2kyoh_, alright ... let me debootstrap squeeze.19:45
yoh_and I would have checked with release team why britney (?) demands those builds for other architectures since "[2011-10-06] shogun REMOVED from testing (Britney)"19:57
@sonney2kyoh_, no problem installing shogun from squeeze -> sid20:06
shogun-buildbot_build #416 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/416  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>, Michal Uricar <uricar.michal@gmail.com>20:13
shogun-buildbot_build #417 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/417  blamelist: Michal Uricar <uricar.michal@gmail.com>20:32
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun20:35
n4nd0hi sonney2k20:39
n4nd0do you have a moment for a question about PLiFs?20:39
-!- zxtx [~zv@c-24-6-91-131.hsd1.ca.comcast.net] has joined #shogun20:48
shogun-buildbot_build #418 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/418  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>20:51
-!- blackburn [~blackburn@188.168.2.3] has quit [Ping timeout: 252 seconds]21:17
-!- blackburn [~blackburn@188.168.2.162] has joined #shogun21:18
shogun-buildbot_build #419 of deb3 - modular_interfaces is complete: Failure [failed compile java_modular]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/419  blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>21:23
-!- blackburn [~blackburn@188.168.2.162] has quit [Quit: Leaving.]21:34
yoh_sonney2k: [no problem installing shogun from squeeze -> sid] good -- leave a comment to that bug report and downgrade importance if it was high ;)21:49
yoh_so it is left to figure out why it didn't get into wheezy ;)21:50
-!- gsomix [~gsomix@109.169.157.174] has joined #shogun22:06
gsomixsonney2k, hey22:06
gsomixsonney2k, can I change some constructors and methods needed for zero-copy and write typemaps? foobar(SGMatrix) -> foobar(SGMatrix, bool copy), I mean22:08
gsomixthanks22:08
gsomixoh no, I cant :(22:11
n4nd0good night guys22:15
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving]22:15
@sonney2kyoh_, but actually there was someone noting how to reproduce the issue22:40
@sonney2kyoh_, just do apt-get install --reinstall shogun-python-modular on your machine and voila!22:40
@sonney2kgsomix, the problem is that for the typemaps to work you would then need either typemaps that operate on SGVector, bool22:42
yoh_sonney2k: ah22:42
@sonney2kor some flag inside SGVector22:42
yoh_sonney2k: indeed22:43
yoh_ah... you have all kinds of "stuff" in your shogun-python-modular.preinst22:44
yoh_and shogun-python-static.preinst -- I guess forgotten to be removed when switched from pycentral ;)22:45
@sonney2kyoh_, it can only happen on upgrade22:45
yoh_sonney2k: yeap -- all that on if [ "$1" = upgrade ]22:45
yoh_;)22:45
@sonney2kyoh_, ok there is a shogun-elwms-static.preinst22:46
yoh_sorry for misleading22:46
@sonney2ketc22:46
yoh_I guess it is rm -f /usr/lib/python$pyversion/$pydir/shogun/*22:46
yoh_in shogun-python-modular.preinst22:47
yoh_which is not conditioned on " dpkg --compare-versions "$2" lt 0.9.3-3 "22:47
yoh_so you always invoke it22:47
@sonney2kI have absolutely no idea if removing the *preinst* files would be OK...22:48
yoh_well -- theoretically all those manual cleanups should not be necessary at all, right?22:49
@sonney2kindeed prerm contains22:50
@sonney2kpyclean -p shogun-python-modular22:50
yoh_and you had them there for some previous versions... probably prior 1.0... so why not condition such and be done?  you probably would know more ;)22:50
yoh_pyclean (from manpage) just removes complied .py[co] -- that is different22:50
yoh_just look when/why you introduced those total cleanup loop -- may be you could simply condition it on some earlier version as well and be done ;)22:52
yoh_probably prior 1.xxx versions22:53
@sonney2kyoh_, I didn't no clue about this hack22:53
@sonney2kyoh_, OK in dh_python2 packages are done more robust: every symlink / dir is already inside the package visible via dpkg -L ...22:55
yoh_it was only your 2 commits which worked on that .preinst :-P22:55
@sonney2kand just /usr/share/pyshared contains the stuff22:55
yoh_sonney2k: yes, with dh_python2 everything is in the package (so needs binary rebuilds for new supported python versions)22:55
gsomixsonney2k, hm, there is another problem. typemaps doesn't make sense, when we need to overload free_feature_matrix with hack22:55
gsomix:(22:56
@sonney2kyoh_, that makes a *lot* of sense22:56
gsomixsonney2k, btw, I have implemented new inplace arithmetic mechanism. ?ow the behavior is the same as in numpy.23:00
gsomix*now23:00
@sonney2kgsomix, you mean you can do shogun dense featurematrix x : x+=1 ?23:02
@sonney2kthat worked already - so what did you change?23:02
gsomixsonney2k, yep23:04
gsomixworked? O_o23:04
gsomixwait... how?23:04
@sonney2kgsomix, err you did that a month back IIRC?!23:05
gsomixyep, but only for matrices with same type and shape and buffer23:05
@sonney2kyoh_, I think this preinst is still needed - at least I don't understand how the files in /usr/lib/python2.X/dist-packages/shogun/* will be removed?23:07
@sonney2kgsomix, ahh23:08
@sonney2kso now you can do these ops for scalars and other compatible shapes?23:08
gsomixyep23:08
gsomixlike in numpy, because I use numpy's number protocol (I need send letter to this guys)23:09
yoh_sonney2k: do you mean on new installations?  dh_python2 should take care about that ... it should add call to update-python-modules  -c in prerm23:11
gsomixat now I want only one thing - RealFeatures(my_cool_matrix, copy=False). :(23:11
@sonney2kyoh_, for new installations it will work23:12
yoh_so - for new packages there should be no manual anything in .pre/post scripts... at least I never add anything ;)  whatever problems you were resolving before should thus be conditioned on those old versions23:12
@sonney2kyoh_, since 1.0.0-1 it uses dh_python223:12
yoh_good -- so why not condition that manual rm on pre-1.0.0-1 of the version to be upgraded from?23:13
@sonney2kso maybe conditioning on that could work23:13
@sonney2kgsomix, I don't see that happening w/o adding new typemaps / constructors23:15
yoh_sonney2k: but once again -- figure out first is there anything else to fix to get current revision into wheezy23:15
-!- yoo [575b08cb@gateway/web/freenode/ip.87.91.8.203] has joined #shogun23:16
-!- yoo [575b08cb@gateway/web/freenode/ip.87.91.8.203] has quit [Client Quit]23:16
@sonney2kyoh_, I cannot fix clang though23:17
yoh_rright -- but that is only on some architectures, right?  ATM there is no shogun on ANY architecture in wheezy23:18
@sonney2kyoh_, there are of course no bug23:20
@sonney2ks23:21
yoh_of cause not -- are you writing a buggy software? ;)23:22
gsomixsonney2k, nite. PR will be tomorrow, and I will star work on typemap for modelsel.23:25
gsomixgood night guys23:25
yoh_gsomix: sleep tight!23:25
@sonney2kyoh_, alright - so before 0.9.3-3 shogun packages where using pycentral23:25
@sonney2kthat is what the first comparison is for23:26
@sonney2know I would jsut add23:26
@sonney2kif dpkg --compare-versions "$2" lt 0.10.0-1 ; then23:26
@sonney2karoudn the other rm -rf23:26
@sonney2krm -f that is23:26
@sonney2kyoh_, and do a new upload23:27
yoh_but why new upload? (or to experimental?)23:27
@sonney2kno to sid?23:28
yoh_I would have first figured out WTF will be happening with package -- will it go into wheezy at all at this point23:28
@sonney2kI mean the bug is there right?23:28
@sonney2kyoh_, no matter what - the current version is useless23:28
yoh_yeah... bug is there but you do not know what release team thinks in general about this situation23:28
yoh_yeah -- current one is buggy23:29
yoh_anyways -- up to you ;-)23:29
@sonney2kyoh_, well shogun 1.1 is pretty outdated anyways after gsoc.23:30
yoh_if I were you I would have just figured out "the plan" first ;)23:30
yoh_so you do not want any shogun in wheezy?23:30
@sonney2kso if it is not in - it doesn't hurt23:30
@sonney2kno I wanted 1.1 in that is why I tried to fix all bugs that got reported in time23:31
@sonney2kbut with these show stoppers I simply had no clue what to do23:31
@sonney2kand since the preinst scripts were not mine I could not even have imagined that we were messing with that stuff23:31
@sonney2kgsomix, thanks for your work - I would also suggest - you keep your frombuffer() functions for now23:34
@sonney2kand better work on model selection typemaps23:34
@sonney2kyoh_, btw do you still use shogun these days?23:41
@sonney2kyoh_, when you are awake I mean :D23:47
--- Log closed Tue Aug 21 00:00:17 2012