Open in new window / Try shogun cloud
--- Log opened Fri Aug 09 00:00:01 2013
--- Day changed Fri Aug 09 2013
@sonney2kHeikoS, we need to decide where to put the notebooks though00:00
@wikingsonney2k: google... i have no idea00:00
shogun-buildbotbuild #1689 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1689  blamelist: Soeren Sonnenburg <sonne@debian.org>00:00
@HeikoSsonney2k: have a dir on the website where they are pasted00:00
@HeikoSsonney2k: and then we autogenerate a html site with a list?00:00
lisitsynsonney2k: I once removed my source code being in a wrong directory00:00
@sonney2klisitsyn, yeah that is what I am afraid of00:02
lisitsynsonney2k: we can use something hybrid00:02
@sonney2kI do too many context switches so it is rather likely to happen00:02
lisitsynmakefile + CMake00:02
@sonney2kHeikoS, yeah and where do we put it on the website?00:02
lisitsynso calling make calls cmake00:02
lisitsynand make clean cleans cmake00:02
@wikingsonney2k: i think if u remove CMakeCache.txt00:03
@wikingthat's good enough00:03
-!- travis-ci [~travis-ci@ec2-54-211-42-245.compute-1.amazonaws.com] has joined #shogun00:03
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/1000438600:03
-!- travis-ci [~travis-ci@ec2-54-211-42-245.compute-1.amazonaws.com] has left #shogun []00:03
@HeikoSsonney2k: ah I see00:03
@HeikoSsonney documentation/tutorial00:03
@sonney2kHeikoS, or documentation/examples?00:03
@HeikoSsonney2k: yep00:04
@HeikoSsonney2k: I would like to activate file and line for WARNINGS and ERROR by default, do you agree?00:09
@HeikoSsonney2k: could you send me the scanned version of the signatures of the foundation?00:09
@sonney2kHeikoS, or the other thing (function?)00:11
@HeikoSsonney2k: what?00:11
@HeikoSsonney2k: see my PR00:11
@HeikoSI think its good for users00:11
@HeikoSto know from which file/class things come from00:11
shogun-notifier-shogun-data: Soeren Sonnenburg :master * 8024808 / toy/ (4 files): https://github.com/shogun-toolbox/shogun-data/commit/80248082fd7db96f7b65eb5581d049811c7440e200:12
shogun-notifier-shogun-data: byte data was stored as float - really save as byte data now00:12
shogun-notifier-shogun: Soeren Sonnenburg :develop * 873e32d / data: https://github.com/shogun-toolbox/shogun/commit/873e32d8e6b1d716e70f6ebd6275988b6479996600:12
shogun-notifier-shogun: require fixed byte data00:12
@sonney2kHeikoS, just do it00:12
shogun-notifier-shogun: Heiko Strathmann :develop * 103e071 / src/shogun/io/SGIO.cpp: https://github.com/shogun-toolbox/shogun/commit/103e071d3ca619e28fc45b48c04df3f0cf6630b700:12
shogun-notifier-shogun: warnings and worse always show file and line number00:12
shogun-notifier-shogun: Heiko Strathmann :develop * 01641bc / src/shogun/io/SGIO.cpp: https://github.com/shogun-toolbox/shogun/commit/01641bc25e64efa28296b6b5ab7b0c9ad5b3f49a00:12
shogun-notifier-shogun: error and warning always print file and line00:12
shogun-notifier-shogun: Heiko Strathmann :develop * 72e0ba8 / src/shogun/io/SGIO.cpp: https://github.com/shogun-toolbox/shogun/commit/72e0ba80db9409e30a7cb6ac452893e21216c0b800:12
shogun-notifier-shogun: Merge pull request #1382 from karlnapf/develop00:12
shogun-notifier-shogun:00:12
shogun-notifier-shogun: always line and file for errors and warnings00:12
shogun-buildbotbuild #1692 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1692  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>00:18
@sonney2kwiking, one more question - why do we need such new superlu?00:19
@iglesiasgok guys00:20
@iglesiasggood night!00:20
lisitsynwe don't need superlu at all :D00:20
@sonney2koops it is late00:20
lisitsynI don't use it anymore00:20
@wikingsonney2k: ?00:20
shogun-buildbotbuild #1693 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1693  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Soeren Sonnenburg <sonne@debian.org>00:20
@sonney2kwiking: -- Performing Test SUPERLU_MIN_VERSION_4_300:20
@sonney2k-- Performing Test SUPERLU_MIN_VERSION_4_3 - Failed00:20
@sonney2kso why that?00:20
-!- iglesiasg [~Fernando@s83-179-44-135.cust.tele2.se] has quit [Quit: Leaving]00:20
@sonney2kwiking, and I guess sth changed w/ unit tests? Generating base/clone_unittest.cc00:21
@sonney2kTraceback (most recent call last):00:21
@sonney2k  File "base/clone_unittest.cc.py", line 82, in <module>00:21
@sonney2k    output_file = sys.argv[2]00:21
@sonney2kIndexError: list index out of range00:21
@sonney2kmake[1]: *** [base/clone_unittest.cc] Error 100:21
@sonney2kwiking, I guess that is fixed once converting to cmake?00:21
@wikingdunno00:21
@wikingsonney2k: yes00:21
@wikingsonney2k: it's done00:22
@wikingsonney2k: i mean it's done with cmake00:22
@wikingsonney2k: cmake -DENABLE_TESTING=ON ../00:22
@wikingmake00:22
@wikingctest --output-on-failure00:22
@wikingand there u go00:22
@sonney2kwiking, well it still worked with 0e037a1968fa5a5b09259e3e8cf83c560c1aa0dc00:22
@wikingsonney2k: ?00:23
@sonney2kalright too late I have to sleep too00:23
@HeikoSsonney2k:  good night00:23
@wikingsonney2k: what did u wanted here actually: https://github.com/shogun-toolbox/shogun/blob/4daf5ee2f5d470bbd70cfaf5cd29d9c39472ad5a/examples/undocumented/python_modular/kernel_linear_byte_modular.py00:24
@wikingas u removed from tools.load import LoadMatrix00:24
@wikingbut not00:24
@wiking lm=LoadMatrix()00:24
@wikingohfuckme00:26
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has joined #shogun00:26
travis-ci[travis-ci] it's Viktor Gal'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/1000489500:26
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has left #shogun []00:26
@wikingversionheader change -> rebuild whole libshogun :(00:26
@wikingdoh this needs fixing00:27
@HeikoSwiking, lisitsyn, please have a look at the doc sprint application00:40
@HeikoSupdate your details etc00:41
@HeikoSthoralf: nice job with your patch! :)00:42
@wikingHeikoS: The difference between sample and -1.67990517588 is 0.25253461607449879, which exceeds 1e-1, where00:49
@wikingsample evaluates to -1.4273705598055013,00:49
@wikinginferencemethod....00:49
@wiking-1.67990517588 evaluates to -1.6799051758800001, and00:49
@wiking1e-1 evaluates to 0.10000000000000001.00:49
@HeikoSwiking: ah thats my test right?00:49
@wikingy00:51
shogun-buildbotbuild #1691 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1691  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>01:05
@HeikoSwiking:01:10
@HeikoSGenerating base/clone_unittest.cc Traceback (most recent call last):   File "base/clone_unittest.cc.py", line 82, in <module>     output_file = sys.argv[2] IndexError: list index out of range make[1]: Leaving directory `/home/buildbot/slaves/shogun-slave3/deb1_-_libshogun/build/tests/unit' make: Leaving directory `/home/buildbot/slaves/shogun-slave3/deb1_-_libshogun/build/src' make[1]: *** [base/clone_unittest.cc] Error 1 make: *** [unit-tests] Error 2 progra01:10
@wikingyeyey ye01:10
@wikingnot using the right command01:10
@wikingit should be ctest --output-on-failure instead of make unit-tests01:11
@wikingbut of course i can fix it if it's reeeallly necessary01:11
@wikingbut it's ancient code :P01:11
shogun-notifier-shogun: Heiko Strathmann :develop * b935813 / tests/unit/machine/gp/InferenceMethod_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/b9358130353c1e3285cb5157e1ce4a199a1448c801:14
shogun-notifier-shogun: lowered accuracy to avoid failing tests01:14
shogun-notifier-shogun: Heiko Strathmann :develop * e7834bd / tests/unit/machine/gp/InferenceMethod_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/e7834bd6514ec43247e2eec80e6f32a5cf3d754c01:14
shogun-notifier-shogun: lower once more to be sure :)01:14
shogun-notifier-shogun: Heiko Strathmann :develop * ed77532 / tests/unit/machine/gp/InferenceMethod_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/ed775328555fbc874451924cb7691e00f3ec724b01:14
shogun-notifier-shogun: Merge pull request #1383 from karlnapf/develop01:14
shogun-notifier-shogun:01:14
shogun-notifier-shogun: lower unit test accuracy in GP log-ml sampler01:14
-!- HeikoS [~heiko@90.204.166.211] has left #shogun []01:14
@wikingcool thnx01:15
shogun-buildbotbuild #1694 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1694  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>01:17
shogun-buildbotbuild #1696 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1696  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>01:20
-!- travis-ci [~travis-ci@ec2-54-224-126-10.compute-1.amazonaws.com] has joined #shogun01:30
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/1000634801:30
-!- travis-ci [~travis-ci@ec2-54-224-126-10.compute-1.amazonaws.com] has left #shogun []01:30
shogun-notifier-shogun: Viktor Gal :develop * a8013be / examples/undocumented/python_modular/kernel_linear_byte_modular.py: https://github.com/shogun-toolbox/shogun/commit/a8013bea1b520120fba90da8ded86b9991993c4c01:33
shogun-notifier-shogun: Switch kernel_linear_byte_modular.py example to use CSVFile reader01:33
shogun-buildbotbuild #1697 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1697  blamelist: Viktor Gal <viktor.gal@maeth.com>01:37
-!- hushell [~hushell@8-92.ptpg.oregonstate.edu] has joined #shogun01:46
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has joined #shogun02:00
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/1000636402:00
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has left #shogun []02:00
shogun-buildbotbuild #1695 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1695  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>02:05
-!- hushell [~hushell@8-92.ptpg.oregonstate.edu] has quit [Ping timeout: 264 seconds]02:24
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has joined #shogun02:44
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/1000802702:44
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has left #shogun []02:44
-!- travis-ci [~travis-ci@ec2-54-224-126-10.compute-1.amazonaws.com] has joined #shogun03:06
travis-ci[travis-ci] it's Viktor Gal'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/1000850203:06
-!- travis-ci [~travis-ci@ec2-54-224-126-10.compute-1.amazonaws.com] has left #shogun []03:06
shogun-buildbotbuild #483 of nightly_default is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/48303:24
shogun-buildbotbuild #418 of nightly_all is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_all/builds/41804:16
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun06:25
shogun-notifier-shogun: Viktor Gal :develop * 2086f20 / .travis.yml: https://github.com/shogun-toolbox/shogun/commit/2086f20a52db08488aaca8407af83698e4e61b1306:25
shogun-notifier-shogun: travis: add octave modular job to allowed to failed06:25
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has joined #shogun06:29
travis-ci[travis-ci] it's Viktor Gal'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/1000850206:29
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has left #shogun []06:29
shogun-buildbotbuild #1698 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1698  blamelist: Viktor Gal <viktor.gal@maeth.com>06:33
-!- gsomix [~gsomix@109.169.185.159] has joined #shogun06:48
gsomixgood morning06:51
-!- gsomix [~gsomix@109.169.185.159] has quit [Read error: Connection timed out]07:49
-!- gsomix [~gsomix@109.169.185.159] has joined #shogun07:50
shogun-notifier-shogun: Soeren Sonnenburg :develop * a68dbe4 / examples/undocumented/python_modular/kernel_linear_byte_modular.py: https://github.com/shogun-toolbox/shogun/commit/a68dbe4200136b5ff3d9463123b410c47da3e69e08:03
shogun-notifier-shogun: convert byte example to CSVFile08:03
@sonney2kwiking, how can one avoid a full libshogun rebuild? put the const's only in Version.cpp?08:04
shogun-buildbotbuild #1699 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1699  blamelist: Soeren Sonnenburg <sonne@debian.org>08:08
@sonney2kwiking, how can I specify I want -O0 ?08:19
shogun-notifier-shogun: Soeren Sonnenburg :develop * 3ec0181 / src/shogun/ (5 files): https://github.com/shogun-toolbox/shogun/commit/3ec01815727ad96c6bb7e2a385bf9172c304bd8508:21
shogun-notifier-shogun: move lib/versionstring.h include over to .cpp files to avoid full recompile08:21
shogun-notifier-shogun: when version changes (git pull...)08:21
@wiking:>08:28
@wikingjust woke up08:28
@wikingsonney2k: dunno08:29
@wikingi have to test08:29
@wikinghttps://github.com/shogun-toolbox/shogun/commit/3ec01815727ad96c6bb7e2a385bf9172c304bd8508:30
@wikingisnt solving it....08:31
@wikingimho the solution will be in the cmakefile...08:32
sonne|workwiking: why not?08:39
@wikingdunno08:40
sonne|workwiking: it is nowhere used in the headers any longer08:40
@wikingyes but it wasnt about that.... ever "P08:40
sonne|workwiking: so it will just link again08:40
sonne|workwiking: I mean the old Makefile.template is gone then right?08:40
@wikingversion header generation is a make target08:41
@wikingand libshogun depends on this target08:41
@wikinghence i think if the target is 'rebuilt' that triggers the dependent target to be rebuild again08:41
@wikingi have to figure it out08:41
sonne|workwiking: ohh so we have 2 reasons08:41
sonne|workwiking: sure it must08:42
@wikingbut we need to run the versionheader target08:42
sonne|workwiking: but it shouldn't rebuild08:42
@wikingeverytime08:42
sonne|workwiking: but just compile08:42
@wikingsince we cannot assure that things hasn't changed08:42
sonne|workerr compile lib/Version.cpp08:42
@wikingyeah i have to checkit08:42
@wikingbut first i have to wake up08:43
@wikingand get a coffee08:43
sonne|workheh08:43
@wikingbtw travis was green after a looooong time finally: build 173608:43
@wiking(i moved octave into allowed to fail mode)08:43
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun08:48
-!- mode/#shogun [+o iglesiasg] by ChanServ08:48
@wikingiglesiasg: morning... travis went green \o/09:08
@iglesiasgwiking, awesome!09:08
@iglesiasgI haven't seen what was that made fail the fisher example09:08
@iglesiasgwill have a quick look soon09:08
@iglesiasgwiking, btw, write in the google docs where do you live or would be coming from to the doc sprint09:09
@wikingiglesiasg: it's fixed...09:09
@wikingiglesiasg: yeah i have to read/edit that but first need to fix a cmake shit...09:09
@iglesiasgwiking, all right09:09
@iglesiasgwiking, just want to have a look, out of curiosity, even if it is fixed ;)09:09
sonne|workiglesiasg: it was a long double function missing09:10
@iglesiasgaham I see09:10
@wikingiglesiasg: it was the problem with not detection the libm functions...09:10
sonne|workiglesiasg: wow! the proposal heiko wrote is pretty good already!09:10
@iglesiasgyeah!09:10
-!- travis-ci [~travis-ci@ec2-54-211-42-245.compute-1.amazonaws.com] has joined #shogun09:11
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/1001565409:11
-!- travis-ci [~travis-ci@ec2-54-211-42-245.compute-1.amazonaws.com] has left #shogun []09:11
@wikingsonne|work: i've just tested now... seems your fix actually did the job... only09:12
@wiking[  1%] Building CXX object src/shogun/CMakeFiles/shogun.dir/base/class_list.cpp.o09:12
@wiking[  1%] Building CXX object src/shogun/CMakeFiles/shogun.dir/base/Version.cpp.o09:12
@wikingare rebuilt...09:12
@wikingand of course stuff are relinked09:13
@wikingbut that's all...09:13
@wikingcool09:13
sonne|workwiking: yeah exactly09:13
sonne|workwiking: why is class_list rebuilt?09:13
@wikingheh good question09:16
shogun-buildbotbuild #1700 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1700  blamelist: Soeren Sonnenburg <sonne@debian.org>09:16
@wikingsonne|work: btw on the buildbots: the right way to build would be as on travis09:16
@wikingfor configure: mkdir build && cd build && cmake -DENABLE_TESTING=ON -DBUNDLE_EIGEN=ON -DBUNDLE_JSON=ON ..09:16
@wikingcompile: make09:16
@wikingtest: ctest --output-on-failure09:17
@wikingsonne|work: u can test that on the fbsd bot09:17
@wikingsonne|work: it has cmake09:17
sonne|workwiking: I don't have access from work09:18
@wikingah :S09:18
@wikingthen i guess later tonight09:19
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has quit [Quit: Leaving.]09:27
thoralfHey.09:34
@iglesiasghello hello09:34
sonne|workwiking: what I was desperately missing is to totally disable gcc optimization flags - btw did you disable optimization flags when compiling the modular interfaces? Otherwise this really takes 3 times longer09:38
sonne|workthoralf: hey - is your PR ready to be merged now?09:38
thoralfsonne|work: I think it is - but Heiko wants me to change ASSERT to REQUIRE.09:38
thoralfsonne|work: Can't do that today, maybe monday.09:39
thoralfSorry.09:39
sonne|workyeah it is nice to have a reasonable error message09:39
sonne|workthoralf: np09:39
@iglesiasgwiking, getting a cmake problem here09:40
@iglesiasg-- Found CCache: /usr/bin/ccache09:40
@iglesiasg/usr/bin/ccache: invalid option -- 'd'09:40
@wikingiglesiasg: pastebin it09:41
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun09:41
@wikingsonne|work: what flags exactly do u mean?09:41
@iglesiasgwiking, http://pastebin.com/Np7sWdax09:41
sonne|workwiking: -O0 no tuning09:41
@wikingsonne|work: for swig generated cxx?09:42
sonne|workwiking: what we had with --disable-optimization . Otherwise compile time is too tough09:42
thoralfsonne|work: Preparing a PR from my local patches is quite a lot of work - so it's always a bit annoying doing shogun-housekeeping, too.09:42
sonne|workwiking: well one is for libshogun and interfaces09:42
thoralfsonne|work: I can continue sending PRs, but I've to take care that its not exceeding my capacities.09:42
@wikingsonne|work: currently the only optimization flag u have is -mtune=native...09:42
sonne|workthoralf: I cannot follow sorry09:43
sonne|workwiking: well use -march=native and -O9 or 309:43
sonne|workwiking: but not for compiling modular interfaces!09:43
@wikingsonne|work: no i have notspecified -O9 or 3 anywhere09:44
sonne|workwiking: ohh you have to otherwise optimization is off09:44
@wikingsonne|work: yes09:44
@wikingwell start editing the cmake file :)09:45
@wikingjust make sure you do it for the right target09:45
@wikingi.e. Debug or Release09:45
thoralfsonne|work: Ehrm.  I've been moving code around - most of the assertions have been there already.  A bit annoying to do the migration from ASSERT to REQUIRE.09:45
@wikingiglesiasg: wtf...09:45
sonne|workthoralf: ahh ok well then we can merge it :D09:45
@wikingiglesiasg: what ht efuck is invalid option -- 'd'09:45
sonne|workthoralf: but please do the require too it is like 10 mins work I guess09:46
sonne|workthoralf: I mean do it later but do it09:46
@iglesiasgwiking, is it passed to ccache?09:46
@wikingwell it seems but why and what i dont get it09:46
shogun-notifier-shogun: Thoralf Klein :develop * ea623c7 / src/shogun/lib/SGSparseVector.cpp: https://github.com/shogun-toolbox/shogun/commit/ea623c72aa56571c637d378e665f4203fb5afab109:46
shogun-notifier-shogun: Bounds checking for dense_dot on sparse vectors (if vec is smaller than largest feature_index).09:46
shogun-notifier-shogun: Thoralf Klein :develop * eead651 / src/shogun/features/SparseFeatures.cpp: https://github.com/shogun-toolbox/shogun/commit/eead651c088acd9d180105f59649146a9efbb9c609:46
shogun-notifier-shogun: Removed code duplication in CSparseFeatures<ST>::dense_dot: forwarding call to SGSparseVector<ST>::dense_dot09:46
shogun-notifier-shogun: Thoralf Klein :develop * 768a187 / src/shogun/lib/ (3 files): https://github.com/shogun-toolbox/shogun/commit/768a18721d7ef2201bb46e6362a24573d51ec33609:46
shogun-notifier-shogun: Moved sort_features() from SGSparseMatrix to SGSparseVector.  Fixed ASSERT to allow duplicate feature ids.09:46
shogun-notifier-shogun: Thoralf Klein :develop * da5891c / src/shogun/lib/SGSparseVector.cpp,tests/unit/lib/SGSparseVector_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/da5891ce17410d0284c57a7b84e489ba3e43b3a109:46
shogun-notifier-shogun: SGSparseVector::sort_features() compresses duplicate features; wrote basic unit tests for sort_features().09:46
shogun-notifier-shogun: Thoralf Klein :develop * b629081 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/b629081c749789a4b96374e4c9df4110a3fcff1009:46
shogun-notifier-shogun: Moved get_feature(index) from SGSparseFeatures to SGSparseVector.  Added unit-tests to fix behaviour on duplicate entries.09:46
shogun-notifier-shogun: Thoralf Klein :develop * f128965 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/f12896518117741b4066cfc95e83aaa97a1e7fcc09:46
shogun-notifier-shogun: Moved get_dense() from SGSparseFeatures to SGSparseVector.  Added unit-tests to fix behaviour.09:46
shogun-notifier-shogun: Thoralf Klein :develop * 9b4d950 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/9b4d95003f482f05a4754c1c729bbab6ec9e3f9009:46
shogun-notifier-shogun: Added get_num_dimension() and parameter-less get_dense() to SGSparseVector.  Added unit tests.09:46
shogun-notifier-shogun: Thoralf Klein :develop * b115168 / src/shogun/features/SparseFeatures.cpp,src/shogun/lib/SGSparseVector.cpp: https://github.com/shogun-toolbox/shogun/commit/b115168c673b6bb74ae97f6b4bc7fc02279ea7bd09:46
shogun-notifier-shogun: Fixed formatting issues.09:46
shogun-notifier-shogun: Soeren Sonnenburg :develop * eab098f / / (5 files): https://github.com/shogun-toolbox/shogun/commit/eab098fe15e337a38be73ee5eda19450c50ecb8609:46
shogun-notifier-shogun: Merge pull request #1380 from tklein23/sparse_fixes09:46
@wikingcan u add a line to your ./CMakeLiss.txt: to line 23: MESSAGE(${CMAKE_CXX_COMPILER_ARG1}) MESSAGE(${CMAKE_CXX_COMPILER})09:48
@wikingand rerun cmake09:48
@iglesiasgwiking, sure09:48
thoralfsonne|work: Okay, so it's only REQUIRE(condition, "format string", *arguments)?09:49
sonne|workthoralf: yes09:49
thoralfsonne|work: for instance: REQUIRE(num>=0 && num<get_num_vectors(), "get_feature(num=%d,index=%d): num exceeds [0;%d]\n", num, index, get_num_vectors());09:51
thoralfsonne|work: Too verbose?09:51
sonne|workthoralf: no perfect09:51
@iglesiasgwiking, I am sorry, I think I am doing something wrong. Parse error.  Expected a newline, got identifier with text "MESSAGE".09:51
@iglesiasgall right09:52
@wikingiglesiasg: indeed u r doing it wrong09:52
@wikinggit diff09:52
@iglesiasgit had to be in a couple of lines09:52
@wikingand show me wtf09:52
@iglesiasgMESSAGE09:52
@iglesiasgMESSAGE09:52
@iglesiasg/usr/bin/c++09:52
@iglesiasg/usr/bin/ccache09:52
@iglesiasgthat this what we get with those messages09:52
@wikingmmm and where does then that fucking d comes fomr? :)09:53
shogun-buildbotbuild #1701 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1701  blamelist: Thoralf Klein <thoralf.klein@zib.de>09:53
@wikingdo the same for CMAKE_C_COMPILER and CMAKE_C_COMPILER09:53
@wikingCMAKE_C_COMPILER_ARG109:53
@wikingand CMAKE_C_FLAGS CMAKE_CXX_FLAGS09:53
@wikingif doesn't help then -DENABLE_CCACHE=OFF09:53
@wiking;)09:53
@wikingi have no idea09:54
@wikingwhere do u get that d09:54
shogun-buildbotbuild #1703 of deb1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1703  blamelist: Soeren Sonnenburg <sonne@debian.org>, Thoralf Klein <thoralf.klein@zib.de>09:57
@iglesiasgwiking, flags are empty in both cases09:59
@wikinglol10:00
@wikingi'm sorry man10:00
@wikinghtis is weird10:00
@iglesiasgI can just deactivate ccache10:00
@wikingbut it would be great to figure out why10:00
@wikingsonne|work: ok so DEBUG = -O2 RELASE = -O9?10:01
@wikingany other flags u would like?10:01
@wiking-Wall -Wno-unused-parameter -Wformat -Wformat-security -Wparentheses -Wshadow -Wno-unknown-pragmas -Wno-deprecated10:01
@wiking?10:02
@wiking-O9 -fexpensive-optimizations\ -frerun-cse-after-loop -fcse-follow-jumps -finline-functions -fschedule-insns2\ -fthread-jumps -fforce-addr -fstrength-reduce -funroll-loops -mfpmath=sse10:02
@wikingfor release ^?10:02
@wikingand i guess u want '-g' always10:02
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has joined #shogun10:02
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/1001734210:02
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has left #shogun []10:02
@wikinghahah this merge was killing it all10:03
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun10:20
@wiking lisitsyn i think i've found the solution for the dyn generated src list10:22
lisitsynwiking: !10:22
lisitsynhow?10:22
@wikinglisitsyn: almost like the case of versionstring.h10:23
@wikinglisitsyn: i just don't know how to pass back a variable10:23
@wikinglisitsyn: if we do openmp, then swig should have those flags as well? :)10:23
lisitsynwiking: do you think so?10:26
lisitsynI don't know10:26
lisitsynit won't hurt anyway though10:26
@wikingi have no idea10:27
@wikinglisitsyn: so if we compile libshogun with c++1110:33
@wikingdo we have to have the SWIG compiled with c++11 as well?10:33
lisitsynwiking: good question10:33
lisitsynI hope no but they changed ABI or so10:34
lisitsynso it could get us some problems10:34
@wikingi mean now i'm doing it10:34
@wikingbut hten sometimes i cannot have c++1110:34
lisitsynwiking: where are you located now?10:34
@wikingas swig is not fully c++11 ready10:34
lisitsyncroatia?10:34
@wikinglisitsyn: yep10:34
lisitsynzagreb?10:34
@wikingoh shit today i have to read/write10:34
@wikinglisitsyn: yes10:34
sonne|workwiking: debug = -O0 and yes all the warning flags please10:35
lisitsynwiking: just to fill in this docsprint application10:35
sonne|workwiking: yes and always -g10:35
@wikingsonne|work: debug -O0 not -O2?10:35
@wikingsonne|work: at least in past it was -O010:35
lisitsyndebug is -O0 for sure10:35
@wikingi mean -O210:35
@wikingCOMP_DEFAULT_OPTS="$COMP_WARNINGS -O2"10:35
@wikingthis was in ./configure10:35
lisitsynreally/10:35
lisitsyn?10:35
lisitsynehmhemhem10:35
@wikingyes10:36
lisitsynwhy :D10:36
@wikingdunno but ok10:36
@wikingi mean if no -O is set10:36
@wikingthen it's -O0 by default10:36
lisitsynyes10:36
@wikingas it is now...10:36
lisitsynO2 is already optimizing out functions10:36
@wikingthen i just have to fix the RELEASE flags...10:36
lisitsynso I'd prefer O010:36
sonne|workwiking: no -O3 by default?10:36
@wikingsonne|work: the old ./configure was like this10:37
@wikingCOMP_DEFAULT_OPTS="$COMP_WARNINGS -O2"10:37
@wikingCOMP_NO_OPTS="$COMP_WARNINGS -O0"10:37
@wikingcmake is now by default building Debug mode10:37
@wikingthe only flag was now still missing is -g10:37
@wikingbut i'm fixing that now10:37
sonne|workwiking: we had 3 modes10:39
sonne|workwiking: 1) no optimization at all -> -O0 for all10:39
sonne|workwiking: 2) tune to the max -> -O9 -march=native for libshogun and -O0 for modular interfaces10:40
shogun-buildbotbuild #1702 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1702  blamelist: Thoralf Klein <thoralf.klein@zib.de>10:40
sonne|workwiking: 3) portable optimizations (as in shipped by debian) -> -O2 for libshogun -O0 for modular10:40
sonne|workwiking: please make 2) the default since that is what *users* compiling shogun want10:41
@wikingsonne|work: so10:41
@wikingsonne|work: distribution build type?10:42
sonne|workwhat?10:42
@wikingwell we can have10:42
@wikingbuild types: debug, release, distribution10:42
@wikingdebug = no optimization -O010:42
@wikingrelease = -O9 -march=native10:42
sonne|workyes that is what we usually use for developers10:43
@wikingdistribution = -O210:43
sonne|workdebug I mean10:43
sonne|workyes perfect10:43
@wikingand the default build type is debug10:43
@wikingok setting it up10:43
@wikingsonne|work: distribution: -march=native ?10:43
@wikingaaah you mean we should build release10:43
@wikingok10:43
@wikingsonne|work: -g always i guess10:45
sonne|workwiking: yes?10:46
@wikingsonne|work: -march=native for 3) (distribution) ?10:46
@wikingoerror: invalid value '9' in '-O9'10:48
@wiking:DDD10:48
@wikingoh yeah10:48
@wiking       -O0 -O1 -O2 -Os -Oz -O3 -O410:49
@wikingfor clang ;)10:49
lisitsynwiking: just O310:49
lisitsynI have no idea what is O910:49
@wikingyeah10:49
lisitsynI guess that's something like quantum computing optimization10:50
@wikingit'll b -IO310:50
lisitsynin they year 215310:50
@wikingO310:50
lisitsynwiking: O4 is dangerous like a mine field10:50
lisitsynit kabooms the kaboom10:50
@wikingok testing O310:51
@wikingbut i guess on travis we want to build in Debug mode :P10:54
sonne|workwiking: well what we ship to users is what we should test :)10:57
@wikingsonne|work: ok10:57
@wikingsomebody fix the last merge error11:04
sonne|workthoralf: ^ any clues for https://travis-ci.org/shogun-toolbox/shogun/jobs/10017343 ?11:13
@wikingsonne|work: these were not used for swig cxx: -Wall -Wno-unused-parameter -Wformat -Wformat-security -Wparentheses -Wshadow -Wno-unknown-pragmas -Wno-deprecated11:19
@wikingsonne|work: do we want them?11:19
-!- van51 [~van51@ppp-94-66-52-171.home.otenet.gr] has joined #shogun11:20
-!- hoijui [~hoijui@adsl-84-227-37-53.adslplus.ch] has joined #shogun11:20
sonne|workwiking: maybe not sure. put them in and we check later11:35
sonne|workvan51: morning!11:35
-!- HeikoS [~heiko@nat-163-6.internal.eduroam.ucl.ac.uk] has joined #shogun11:39
-!- mode/#shogun [+o HeikoS] by ChanServ11:39
van51sonne|work: morning!11:48
van51sonne|work: I was grabbing a bite :)11:49
van51sonne|work: I think that PR is good to merge11:49
van51sonne|work: I ran the lang detection example last night and it did quite well11:50
van51sonne|work: now I would like to make a dataset of small senteces to compare results  w/ and w/o quadratic11:50
van51sonne|work: and then turn it into a demo11:50
-!- hoijui [~hoijui@adsl-84-227-37-53.adslplus.ch] has quit [Quit: Leaving]11:51
sonne|workvan51: quite well == ?11:54
van51sonne|work: it had like 99.99% accuracy on large documents and I just typed a few small sentences that it didn't miss11:55
van51sonne|work: and the training time took ~2200 secs on 30k docs11:56
sonne|workvan51: that was for large documents right?11:59
van51sonne|work: yea11:59
thoralfsonne|work: I'll check.12:00
van51sonne|work: well they 're not enormous either :P just an average wiki article12:00
sonne|workvan51: alright then create smaller snippets and try again12:00
thoralfsonne|work: Sorry for the delay; meeting.12:00
@HeikoSsonne|work, lisitsyn, wiking, iglesias please more tuning of the proposal! Want to submit soon12:00
sonne|workvan51: well you could take say just say 5-50 words12:00
lisitsynHeikoS: yeah going through it12:01
@HeikoSsonne|work, lisitsyn, wiking, iglesias I have the feeling that if we send it too late, they might decide before weekend12:02
@HeikoSleaving us out12:02
@HeikoSso lets send it ASAP12:02
@HeikoSbest thing is if they have in in their mailbox friday morning12:02
@HeikoSwhich is soon12:02
@HeikoSsonne|work, lisitsyn, wiking, iglesias what do you think if we push some individuals to apply for SHOGUN?12:03
@HeikoSlike thoralf, or kevin?12:03
sonne|workHeikoS: you mean thoralf and who else?12:03
thoralfsonne|work: The break happened because I switched from SGVector<T> dense(0) to dense().12:03
@HeikoSsonne|work: pickle blogs a lot about shogun and does cool stuff12:03
sonne|workthoralf: errm ok then use dense(0) (wtf?!?)12:03
@HeikoSdont know12:04
sonne|workHeikoS: well or even lambday12:04
@HeikoSI think if there are some individuals applying our chances get batter12:04
@HeikoSyeah12:04
@HeikoSsonne|work: but that might be a bit difficult12:04
@HeikoSsonne|work: do you think its a good idea?12:04
sonne|workbutter batter better yes12:04
@HeikoSthoralf: interested is applying to the doc sprint?12:05
thoralfHeikoS: What's this?12:05
@HeikoSsonne|work: Ill write to the mailing list maybe?12:05
@wikingHeikoS: gimme access for vigsterkr@gm12:05
thoralfHeikoS: Sorry, didn't follow your thread.12:05
@HeikoSthoralf: getting together for three days in the US and write a book on shogzun12:05
thoralfHeikoS: I'm in. ;)12:05
@HeikoSthoralf: so we apply as an organization, but individuals may also apply12:06
@HeikoSand we think that this might help12:06
@HeikoSthoralf: sent you an email12:06
sonne|workHeikoS: well they would need to apply by end of today right?12:06
@HeikoSwiking: shared12:06
@HeikoSsonne|work: the thing is: its friday, I guess they will decide today12:07
@HeikoSso better earlier (for our big one=)12:07
@HeikoSI will submit in 3hrs12:07
@HeikoSso please everyone tune tune tune :)12:08
@HeikoSgot to do some work  now see you later12:08
sonne|workHeikoS: no they won't decide today12:08
sonne|workHeikoS: I guess not before 1 week12:08
sonne|workHeikoS: but still deadlines are deadlines12:08
@HeikoSok12:09
@HeikoSsonne|work: Ill write an email to the mailing list to push individual users to apply12:10
@HeikoSsonne|work, thoralf: see mail on shogun-list12:19
-!- van51 [~van51@ppp-94-66-52-171.home.otenet.gr] has quit [Quit: Leaving.]12:21
@iglesiasgwiking, hello hello12:24
@iglesiasgwiking, I am getting this when doing ccmake12:24
@iglesiasgYour compiler cannot expand a variadic template parameter into a fixed-length argument list.12:24
@wikingiglesiasg: this should have been fixed12:24
@iglesiasgIt doesn't look like an error though12:24
@wikingiglesiasg: ah ok no it's not an error12:24
@iglesiasgwiking, should I worry about it?12:25
@wikingiglesiasg:12:25
@wiking    # to avoid problems with variadic template handling in gcc 4.6.312:25
@wiking    # with -std=c++0x mode in gmock-matchers.h12:25
@wiking    # error: unimplemented: cannot expand 'Tail ...' into a fixed-length argument list12:25
@wiking    # TODO: check if there's a version12:25
@wikingso your compiler doesn't support c++11 but c++0x12:25
@wikingbut c++0x will cause some troubles with gmock12:25
@wiking(see the variadic template parameter thingy)12:26
@wikinghence c++0x is not enabled12:26
@wikingnothing to worry about12:26
@iglesiasgok12:26
@wikingthoralf: you did the sparsevector stuff?12:28
thoralfwiking: Yes, why?12:28
@wikingif yes then in SGSparseVector.cpp:23012:28
@wikingwhy: SGVector<T> dense;12:28
@wikingi mean SGVector<T> dense();12:28
@wikingwhy is the () there?12:28
thoralfwiking: Does it hurt?12:28
@wikinguyes12:29
@wikingit kills the compiler ;)12:29
thoralfI just replaced it by (0)12:29
thoralfwiking: That's all?12:29
thoralfPuh.12:29
thoralfThanks.12:29
@wikingwhy not SGVector<T> dense;12:29
@wiking?12:29
@wikingit'll call SGVector<T> () ctor12:29
thoralfwiking: Didn't know there's a difference between dense() and dense.12:29
@wikingok i'll finish this and then commit the patch12:30
@wikingso dont send a PR for this12:30
thoralfwiking: Ehrm.  I just prepared a PR containing this and a few other fixes. ;)12:30
@wikingthoralf: ok then send it12:31
@wikingand after that i'll send my stuff12:31
thoralfPR is out.12:33
thoralfHeikoS: It contains your annoying REQUIRE-thing. ;)12:34
@HeikoSthoralf: thanks!12:35
@HeikoSthoralf: I know annyoing but very helpful for users, as they otherwise only see assertion failed12:35
@HeikoSthoralf: could you change one more thing? We have an 80 characters per line style12:35
thoralfwiking: Can you explain the difference between "SGVector<T> dense()" and "SGVector<T> dense"?12:35
thoralfHeikoS: Ey!12:36
@HeikoSthoralf: ? :)12:36
thoralfHeikoS: I was just moving code, so the assertions were not mine.  I fixed it on your behalf and now it's too long?12:36
thoralfNo! ;)12:36
thoralfstriking!12:36
@HeikoSthoralf: Well then just keep it in mind next time12:37
@HeikoSits easy, just press enter ;)12:37
thoralfHeikoS: I will. ;)12:37
@HeikoSbut thanks for the patch12:37
@HeikoSvery good stuff!12:37
@HeikoSthoralf: dont blame me just trying to keep code clean12:37
thoralfHeikoS: Yeah, it's okay.12:38
thoralfHeikoS: I wish there would be a kind of indent-style to apply before committing.12:38
@HeikoSthoralf: thats why we need the doc camp :) shogun developer guidelines :)12:38
thoralfHeikoS: Or code sniffer.12:38
thoralfYah.12:38
thoralfYeah.12:38
@HeikoSthoralf: I have that set up for eclipse btw12:38
@HeikoSdoesnt work with  macros though12:38
@HeikoSbut for other code things,12:38
thoralfHeikoS: Which style did you choose?12:38
@HeikoSjust press ctrl-shift-f12:39
@HeikoSthere are things in some readme that I obey12:39
@HeikoSsuch as the 80 char limit12:39
@HeikoSspaces, newlines etc12:39
thoralfHeikoS: I'm switching between eclipse and command-line, escpecially for remote editing.12:39
@HeikoSthoralf: yeah for remote emacs is cool, or vim12:40
@HeikoSbut my fingers go on strike when I use them :)12:40
thoralfHeikoS: My laptop has only 2 cores, my workstation 8.  So it's better to compile and check remote. :)12:40
@HeikoSthoralf: did you get the email on shogun-list about the doc sprint?12:40
thoralfHeikoS: Yes, I did.12:40
@HeikoSthoralf: ha! indeed :) btw use cmake, compiling is done with ccache automagically12:40
thoralfHeikoS: How to use this?12:41
@HeikoSwiking sent an email12:41
@HeikoSshogun with cmake is the subject12:41
@HeikoSits easy12:41
@HeikoScreate dir12:41
@HeikoSrun cmake with your options12:41
@HeikoStyp emake12:41
@wikingdidin't get anything12:41
@HeikoSwiking: no mail on the shogun-list?12:41
@wikingah ok ... not talking to me12:41
@HeikoSwiking: ah :)12:42
thoralfsites google tells me "You need permission"12:42
@iglesiasgyes, it says that for me too12:42
@iglesiasgbtw, why is it we lost the green in travis already? :(12:43
thoralfHeikoS: Ah, it's new.  Didn't read that one yet.12:43
@wikingiglesiasg: hehhe wrong ctor by thoralf12:43
@wikingthoralf: where's the PR?12:43
thoralfhttps://github.com/shogun-toolbox/shogun/pull/138412:44
@iglesiasgit is not the new PR though12:44
@iglesiasgsomething was already bad before12:44
@wikingiglesiasg: diagonal shit12:44
@wikingin the unit test12:44
thoralfwiking: Damn.  I broke my unit tests.12:44
@wikingor at least it was that before12:45
thoralfIt's always the same: "One quick PR and I'm done"12:45
thoralfNever thinking that again. (Or kill me)12:45
@wikingthoralf:12:46
@wiking[  FAILED  ] HashedDocDotFeaturesTest.computed_features_test12:46
@wiking[  FAILED  ] HashedDocDotFeaturesTest.dense_dot_test12:46
@wiking[  FAILED  ] SparseFeaturesTest.subset_get_full_feature_vector_identity12:46
@wiking[  FAILED  ] SparseFeaturesTest.subset_get_full_feature_vector_permutation12:46
@wiking[  FAILED  ] SparseFeaturesTest.subset_get_full_feature_vector_smaller12:46
@wikingPLEASE COMPILE AND TEST YOUR PRs *BEFORE* sending even the PR12:46
@wikingit's just waste of compile time ...12:46
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]12:46
@wikingbecuase now we got that shit merged12:47
@wikingit's going to break all other ppl's PR broken12:48
@wikingand it's again where we were yesterday12:48
thoralfwiking: Do you think I don't know that?12:48
@wikingwhere the travis output was not reliable12:48
thoralfSeriously, I'm sorry.12:48
@wikingthoralf: making sure that u know12:48
@wiking;)12:48
@wikingit took me like weeks to get that travis into a working shape with cmake12:48
@wikingso that finally we can go on green12:48
@iglesiasgwe should also stop merging stuff that doesn't pass travis12:49
@wikingok let's fix the unit test12:49
@iglesiasgI mean it is not only responsible the person who does the PR12:50
@wikingiglesiasg: yes12:50
thoralfFITCInferenceMethod.get_cholesky12:50
thoralf*** Error in `./shogun-unit-test': free(): invalid pointer: 0x0000000001fb52c0 ***12:50
@wikingiglesiasg: but it was a catch 22 i think in this PR case12:50
thoralfUnit tests are broken.12:50
@wikingiglesiasg: because the PR was sent before cmake marge12:50
@wikingiglesiasg: so the output of travis was unreliable12:50
@iglesiasgwiking, aah ok I see12:50
@wikingand even with the rebuild of travis build12:50
@iglesiasgI thought it was because of the new pull requests that got merged recently12:50
@wikingwould fail as cmake wasn't even in yet12:50
@iglesiasgI mean I looked at this12:51
@iglesiasghttps://travis-ci.org/shogun-toolbox/shogun/builds12:51
@wikingiglesiasg: yeah but that PR got in before cmake merge12:51
@wikingso there's nothing we could have done about it12:51
@iglesiasgaaham all right!12:51
@wikingthoralf: afaik not... :))12:52
@wikingthoralf: make clean the unit tests12:52
@wikingusually some shit got stuck in with the old unit test way12:52
@wikingand that caused horrible segfaults12:52
thoralfFile "base/clone_unittest.cc.py", line 82, in <module>: output_file = sys.argv[2]12:56
thoralfIndexError: list index out of range12:56
thoralfwhen running "make" in tests/unit/12:56
@iglesiasgwiking, I found some trouble compiling with test enabled. I guess it is because I should give the path to gmock and/or gtest somewhere?12:58
thoralfproblems when running "base/clone_unittest.cc.py base/clone_unittest.cc.jinja2" - any idea?12:59
thoralfIndexError: list index out of range13:00
@iglesiasgthoralf, not really actually13:02
@iglesiasgdo you know if that happens generating the unit tests for clone? afterwards actually running them?13:03
thoralfiglesiasg: No, I think generating is failing.13:04
thoralfiglesiasg: Since the py script is failing and the clone tests donn't exist yet.13:05
@iglesiasgthoralf, that looks funny indeed then. No idea why would that depend on the machine where it runs13:06
@wikingiglesiasg: no13:15
@iglesiasgwiking, why then?13:16
@wikingthoralf: what r u using? i guess not cmake13:16
@wikingiglesiasg: what was the error13:16
@iglesiasgwiking, let me show you13:16
thoralfwiking: make13:16
@iglesiasgwiking, http://pastebin.com/wcqmNYtn13:17
thoralfwiking: Your last commit on clone_unittest introduced new parameters, but you didn't change the Makefile?13:17
@wikingthoralf: yes bcau makefile is deprecated13:17
@wikingiglesiasg: lolface13:17
@wikingiglesiasg: let's do this this way13:17
thoralfwiking: How to run the unit-tests then?13:18
@iglesiasgwiking, is cmake downloading gtest or something?13:18
@iglesiasgwiking, I had it in /usr/src and now I see onw in /usr/include13:19
@wikingiglesiasg: yes and it doesn't care abou tit13:19
@HeikoSwiking, lisitsyn we really should have a readme on this :)13:19
lisitsynHeikoS: I'll modify it rather soon13:19
@wikingHeikoS: i told to lisitsyn13:19
@iglesiasgthis is not even 24h old guys, it is normal that these problems arise :)13:20
@wikingiglesiasg: ^ pull this version.13:20
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun13:20
shogun-notifier-shogun: Viktor Gal :develop * 356d031 / / (6 files): https://github.com/shogun-toolbox/shogun/commit/356d03174f91784d73b26dd91c4fc2ad9084590213:20
shogun-notifier-shogun: Add CTestConfig and add Release, Debug and Distribution CFLAGS13:20
shogun-notifier-shogun: set CMAKE_BUILD_TYPE to Release by default13:20
@wikingi mean this13:21
@iglesiasgwiking, all right13:21
@wikingand then13:21
@wikingrm -rf build13:21
@iglesiasgwiking, rm -rf build13:21
@wikingyep13:21
@wikingmkdir build13:21
@iglesiasghehe read your mind :P13:21
@wikingcd build13:21
@wikingcmake -DBUILD_DASHBOARD_REPORTS=ON <and whatever other flags u r using for cmake> ..13:22
@wikingand then NOT 'make' but13:22
@wikingctest -D Experimental -j2 (or as many jobs u r usually using for building)13:22
@wikingand then wait13:22
@wikinguntil it compiles and runs the whole thing13:22
@wikingi'll get the output myself ;)13:23
@iglesiasgok13:23
@wikingthoralf: you can quickfix this by adding to tests/unit/Makefile line 5713:23
@wikinginstead of @$(PYTHON) $@.py $@.jinja213:23
@wiking@$(PYTHON) $@.py $@.jinja2 ../../src/shogun/base/class_list.cpp13:24
@wikingso append there ../../src/shogun/base/class_list.cpp13:24
@wikingand u should be fine13:24
@iglesiasgwiking, it is on progress already13:24
thoralfwiking: Thanks.13:24
@wikingiglesiasg: yeah it'll take a whiel as it compiles and stuff13:24
@wikingiglesiasg: but it's writing you the ... stuff right?13:24
@iglesiasgwiking, all right13:24
@iglesiasgwiking, yes13:24
@wikingcool13:24
@wikingso then just wait13:25
@iglesiasglot of lines with lot of points and then Size numberK13:25
@wikingi should get an email once it uploaded the reports13:25
@wikingyes yes that's it13:25
@iglesiasgthis is neat13:25
thoralfwiking: There's still another parameter missing.13:25
@wikingthoralf: ?13:26
thoralfclass_list_file = sys.argv[3]13:26
thoralfYou told me output_file = sys.argv[2]13:26
@wikingah fuck13:26
@wikingforgot that one13:26
shogun-buildbotbuild #1704 of deb1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1704  blamelist: Viktor Gal <viktor.gal@maeth.com>13:26
@wiking@$(PYTHON) $@.py $@.jinja2 $@ ../../src/shogun/base/class_list.cpp13:27
@wikingliek that13:27
@wikingman eigen 3.1.2 is like shadowing everything in shogun13:29
thoralfwiking: jinja2.exceptions.TemplateNotFound: base/clone_unittest.cc.jinja213:29
thoralfwiking: But: 874 Jul 24 17:28 base/clone_unittest.cc.jinja213:29
thoralfThe file exists.13:29
@wikingthoralf: where are u running the make for the unit tests?13:30
thoralftests/unit13:30
@wikingmmm13:30
@wikinginteresting13:31
@wikingthoralf: ahahahha i know13:31
@wikingthe script13:31
@wikingbase/clone_unittest.cc.py:6013:31
@wiking    templateLoader = jinja2.FileSystemLoader(searchpath="/")13:31
@wikingshould be     templateLoader = jinja2.FileSystemLoader(searchpath="/")13:31
@wikingi mean ./ instead of /13:32
@wikingso relative path instead of full path13:32
thoralfwiking: Man, you should test your stuff before committing! :D13:32
@wikingthoralf: ...13:32
@wikingthoralf: use cmake13:32
@wikingnone of this would be a problem13:33
@wikingneither the segfault of the unit test13:33
@wikingiglesiasg: still running?13:33
@iglesiasgwiking, yep13:33
@iglesiasgwiking, there is a line with some !!!! instead of ...., worrying?13:34
@wikingmmm yeah i guess so :))) but lets see what's going to be the output13:34
@wikingafaik !! stands for error13:34
-!- gsomix [~gsomix@109.169.185.159] has quit [Ping timeout: 260 seconds]13:34
@wikingand * is warning or something like this13:35
@wikingit's written there13:35
thoralfwiking: I was all integrated into my eclipse and working... since your mail is <24h away, I hadn't time to read and adapt.13:35
thoralfI know, not your fault. ;)13:35
@wikingthoralf: cmake -G Eclipse13:36
thoralfBut you at least could have said that the old make will be deprecated on the next git pull.13:36
@wikingand have fun13:36
-!- gsomix [~gsomix@109.169.185.159] has joined #shogun13:36
@wikingthoralf:13:36
@wikingEclipse CDT4 - Ninja        = Generates Eclipse CDT 4.0 project files. Eclipse CDT4 - Unix Makefiles13:36
thoralfwiking: Cool.13:36
@wikingyou can even choose if u want Ninja or Unix makefiles13:36
thoralfwiking: Really impressed.13:36
thoralfwiking: CMake Error: Could not create named generator Eclipse13:39
thoralfwiking: Do I need some cmake extensions?13:39
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has joined #shogun13:39
travis-ci[travis-ci] it's Viktor Gal'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/1002285213:39
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has left #shogun []13:39
@wikingone of the good sideeffects of cmake13:39
@wikingyou can generate even fucking visual studio solution file13:40
@wikinglet along xcode proj file13:40
@wikingthoralf: cmake --help13:40
@wikingthoralf: and you'll see what generators u have13:40
@wikingbut yeah i guess it should have been13:40
@wikingcmake -G Eclipse CDT4 - Unix Makefiles13:41
@wikingor cmake -G "Eclipse CDT4 - Unix Makefiles"13:41
@wikingsoemthing like this13:41
@wikingbut just check what your cmake supports and act accordingly13:41
thoralfwiking: Yeah, worked.13:41
@wikingthoralf: i cannot guarantee anything with that13:41
@wikinghaven't tested it13:42
@wikingshould work though...13:42
@wiking:P13:42
@wikingif not let me know and we can fixit13:42
@wikingiglesiasg: still in building phase?13:42
thoralfBtw., my PR is updated.  Compiles and even unit-tests run locally. ;)13:42
thoralfwiking: FYI: make -G replaces the Makefile which still in the project root14:00
@wikingthoralf: because you have not read the email14:02
@wikingthoralf: in shogun's root14:02
@wikingmkdir build14:02
@wikingcd build14:02
@wikingand there do the 'cmake -G <whatever> ..'14:02
@wikingiglesiasg: ? :)14:05
@wikingiglesiasg: got it :D14:05
@HeikoSwiking: I currently cannot compile14:05
@HeikoSIn file included from /home/heiko/Desktop/shogun/shogun/src/shogun/lib/SGSparseVector.cpp:2:0:14:05
@HeikoS/home/heiko/Desktop/shogun/shogun/src/shogun/../shogun/lib/SGVector.h:38:3: error:   initializing argument 1 of 'shogun::SGVector<T>::SGVector(index_t, bool) [with T = std::complex<double>; index_t = int]' [-fpermissive]14:05
@wikingHeikoS: yeah i know thoralf is working on it14:05
@HeikoSok14:05
@wikingHeikoS: i didn't merge that one...14:06
@wiking;)14:06
@wikingthe quickfix for that error is this14:06
@HeikoSyeah merging always goes to quick, we should always ask travis before14:06
@HeikoSits green right?14:06
@HeikoSapart from this14:06
thoralfMy PR is green.14:06
@wikingno14:06
thoralfKindof green.14:06
@wikingthoralf: with the new?14:06
@wikingthoralf: gimme link14:06
thoralfhttps://github.com/shogun-toolbox/shogun/pull/138414:06
thoralfOSX is still running.14:07
@wikingthoralf: ok just a sec14:07
@wikingrestarting gcc job14:07
@wikinglet's wait the gcc job to restart/finish14:08
@wikingoh this is getting even better14:11
@wikingrestarting again but this seems to be a major error14:12
@wikingwe cannot merge it14:12
@wikingiglesiasg: the problem with your thing is bascially that current shogun develop is utterly broken14:13
@wikingwe should really just drop that merge14:13
@wikingi mean reverse14:13
@wikingbecause the quickfix as well seems to be killing it14:13
@wikingk14:15
@wikingwe cannot compile currently shogun14:15
@wikingThe log length has exceeded the limit of 4 Megabytes (this usually means that test suite is raising the same exception over and over).14:16
@wikingthere's just too much shit with Eigen14:16
@wikingand it's shadowing everything14:16
shogun-notifier-shogun: Viktor Gal :develop * 4150fbf / CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/4150fbf0f5595900be67dab3fb1fbd71bba570ce14:20
shogun-notifier-shogun: Remove -Wshadow flag see issue #138514:20
lisitsynwiking: -isystem!14:21
sonne|workHeikoS: I am fine with it!14:21
lisitsynwe need to use it somehow14:21
sonne|worklisitsyn: yeah don't drop shadow14:21
@wikinglisitsyn: ok it'll be fixed14:21
@wikingor fix it14:21
lisitsynwiking: include_directories(SYSTEM "${EIGEN3_INCLUDE_DIR}")14:21
lisitsynuse this14:21
lisitsynjust add SYSTEM14:22
lisitsynbefore14:22
@wikinglisitsyn: edit + commit + push? :)14:22
lisitsynand you'll get now warnings14:22
@wikingi mean it's not only me who's allowed to edit cmakefiles14:22
sonne|workHeikoS: we can nicely explain all the user issues like installation etc and extend as much as we want into ML algorithms and development so it is good enough for me now14:22
lisitsynwiking: yeah sure but you can get mad if I modify something you work on ;)14:22
@wikinglisitsyn: now it's already merged14:22
@wikingso it's free for grag14:22
@wikinggrab14:22
sonne|workwiking: well I am clueless about cmake anyway14:23
lisitsynwiking: oops you use list herfe14:24
@wikinglisitsyn: btw when u see how we includes14:24
lisitsynbad bad14:24
@wikingthis was the only way14:24
lisitsynyeah14:24
lisitsynwell14:24
@wikingremember the other conversation we had earlier...14:24
lisitsynwiking: but all includes are from libraries righT?14:24
@wikinglike a week ago when i had to refactor the shit14:25
@wikinglisitsyn: git grep "LIST(APPEND INCLUDE"14:25
@wikingbut yeah i guess so14:25
@wikingcannot remember anymore14:25
lisitsynwiking: where do you put the INCLUDE list to includes?14:26
shogun-buildbotbuild #1705 of deb1 - libshogun is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1705  blamelist: Viktor Gal <viktor.gal@maeth.com>14:26
@wikinglisitsyn: couples of places14:30
@wikingone is src/shogun/CMakeLists.txt14:30
@wikingthe other is cmake/CommonModularInterface.cmake14:30
@wikinglisitsyn: imo the best way to do this is to maintain 2 list14:31
@wikingone system list and one normal include list14:31
@wikingand now it's only eigen....14:31
@wikingbut who knows in the future14:31
sonne|workHeikoS: ping ping?14:37
lisitsynwiking: we can include all the libraries as system actually14:37
@wikingwhy?14:38
@wikingi mean why we would do that?14:38
sonne|workwiking: I am fine with the SYSTEM include way14:38
lisitsynwiking: we don't care about other code warnings14:38
shogun-notifier-shogun: Thoralf Klein :develop * ce4ee05 / src/shogun/features/SparseFeatures.cpp,src/shogun/lib/SGSparseVector.cpp: https://github.com/shogun-toolbox/shogun/commit/ce4ee057ce3980725275ced9416c889f5cdf8b9814:38
shogun-notifier-shogun: * replaced ASSERT by REQUIRE in several places14:38
shogun-notifier-shogun: * removed redundant assertions on get_num_vectors(): get_sparse_feature_vector(num) already checks it14:38
shogun-notifier-shogun: * replaced SGVector<T> dense() by dense to fix compiler errors.14:38
shogun-notifier-shogun: Viktor Gal :develop * e2ad36b / src/shogun/features/SparseFeatures.cpp,src/shogun/lib/SGSparseVector.cpp: https://github.com/shogun-toolbox/shogun/commit/e2ad36bc0879f29a7c68213bcbef0d90276c2aaa14:38
shogun-notifier-shogun: Merge pull request #1384 from tklein23/sparse_fixes14:38
shogun-notifier-shogun:14:38
shogun-notifier-shogun: replaced ASSERT by REQUIRE; fixed SGVector<T> dense() compiler error14:38
lisitsynwhether it is eigen or shmeigen14:38
lisitsyn;)14:39
lisitsynI think we should not see other code warnings I mean14:39
lisitsynbut if we include shogun code in INCLUDES14:39
lisitsynwiking: do we?14:39
@wikinglisitsyn: we do14:39
lisitsynahh14:39
@wikingor wait14:39
@wikinginclude_directories(${INCLUDES} . ..)14:40
@wikingthis could be split in two easily14:40
@wikingINCLUDE_DIRECTORIES(${INCLUDES} ${CMAKE_SOURCE_DIR}/src)14:40
@wikingsame here14:40
@wikingand same here in unit tests14:40
@wikinginclude_directories(${INCLUDES} ${CMAKE_SOURCE_DIR}/src  ${source_dir}/include ${source_dir}/gtest/include)14:40
@wikinglisitsyn: shogun code is not among ${INCLUDES}14:41
-!- travis-ci [~travis-ci@ec2-54-211-42-245.compute-1.amazonaws.com] has joined #shogun14:41
travis-ci[travis-ci] it's Viktor Gal'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/1002438814:41
-!- travis-ci [~travis-ci@ec2-54-211-42-245.compute-1.amazonaws.com] has left #shogun []14:41
@wikingso if u want u can include the ${INCLUDES} with SYSTEM14:41
@wikingand the others in normal way14:41
@wiking${INCLUDES} are indeed just 3rd party libs14:41
@wikingiglesiasg: i have a good news... i was wrong14:43
@wikingiglesiasg: the PR is always applied on top of HEAD in travis14:43
lisitsynwiking: I don't mind to include 3rd party as system then14:44
@wikingiglesiasg: or not :D14:44
lisitsynthat looks okay, isn't it?14:44
@wikingiglesiasg: i dont understand exactly :P14:44
shogun-buildbotbuild #1707 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1707  blamelist: Viktor Gal <vigsterkr@gmail.com>14:53
thoralfwiking: I followed the instructions from your mail but cmake still overwrites local Makefiles.14:54
thoralfwiking: in root: mkdir build; cd build; cmake .. -> build is empty, but files are overwritten14:55
thoralfwiking: in src: does not work14:55
@wikingthoralf: woah14:56
@wikingthoralf: never seen this before14:56
@wikingthoralf: r u sure u have the HEAD of 'develop' branch?14:56
sonne|workthoralf: weird it did work here14:57
thoralfwiking: Yes.14:57
@wikingthoralf: mmm there's something wrong...14:57
@wikingas this way it has to do out of source compilation14:57
sonne|workthoralf: so you are in shogun/ dir ?14:57
sonne|workthoralf: and then you do mkdir build ; cd build ; cmake .. ?14:57
thoralfsonne|work: ./src/shogun?  No, just tried ./ and ./src14:57
@wikingthoralf: copypaste the output of this here: head -n 1 CTestConfig.cmake14:57
sonne|worknot shogun/src/shogun but just shogun/14:58
sonne|workwiking: I guess I cannot say it enough - great work... lets hope that we can remove configure & friends in about a week14:58
@wikingsonne|work: heheh yeah i was just thinking about that part14:59
@wikingsonne|work: but static is still missing14:59
@wikingsonne|work: btw what's the status of RModular?14:59
@wikingas i really haven't done anything about the rmodular interface14:59
@wikingas it was broken14:59
thoralfSET (CTEST_PROJECT_NAME "shogun")14:59
sonne|workwiking: but static just requires the same settings like the modular interface but just compiling that single file15:00
@wikingthoralf: hehe ok then there15:00
sonne|workwiking: so it is really easy15:00
@wikingmkdir build && cd build && cmake ..15:00
@wikingsonne|work: what do u mean single file?15:00
@wikingsonne|work: ah u mean what's in python_static15:00
@wiking?15:00
@wikingPythonInterface.cpp ?15:01
thoralfwiking: Ran successfully, but build directory is still empty.15:01
sonne|workwiking: interfaces/octave_static/OctaveInterface.cpp etc15:01
sonne|workyes15:01
@wikingmmm15:01
thoralf-- Build files have been written to: xxx/shogun-github15:01
@wikingthoralf: gimme a shell to your machine :))))15:01
@wikingthoralf: i really want to see this15:01
@wikingsonne|work: and then i link that .o with libshogun?15:02
@wikinglisitsyn: what have u decided ? :)15:03
thoralfwiking: I can create a new user, if it's what you wand.15:03
thoralfwant15:03
sonne|workwiking: yes to a -shared .so and that's it15:03
@wikingthoralf: yeah... adduser wiking15:03
@wikingsonne|work: ok got it.. will do soon15:03
sonne|workwiking: ohh exception: cmdline is a real executable !!15:03
@wikingsonne|work: hehe15:03
sonne|workbut for the rest it is exactly like that15:03
thoralfwiking: My head is e2ad36b.., which is latest.15:04
@wikingindeed15:04
@wikingthoralf: just testing w815:04
@wikingthoralf: ls /home/wiking/shogun/build15:05
@wikingthoralf: here's the history of my console15:05
@wikinggit clone https://github.com/shogun-toolbox/shogun.git15:05
@wikingcd shogun/15:05
@wikingmkdir build15:05
@wikingcd build15:05
@wikingcmake ..15:05
@wikingok sorry rm -rf -ed shogun15:05
@wikingdo u need it still ?15:06
@wikingi can reproduce :D15:06
thoralfwiking: All these files are in my project root.15:06
thoralfwiking: Delete.15:06
@wikingu can delete my user now15:06
@wikingthoralf: woah man15:06
@wikingthoralf: u tried with eclipse generator?15:06
thoralfwiking: Maybe it's I ran it once in the wrong directory?15:06
thoralfYes.15:06
@wikingw815:06
@wikingi'll try that as well15:06
@wikingjust a sec15:06
@wikinghaven done that before15:06
@wikingmmm15:08
@wiking-- Could not determine Eclipse version, assuming at least 3.6 (Helios). Adjust CMAKE_ECLIPSE_VERSION if this is wrong.15:08
@wikingand15:08
@wikingCMake Warning in CMakeLists.txt: The build directory is a subdirectory of the source directory.15:08
@wikingThis is not supported well by Eclipse.  It is strongly recommended to use a build directory which is a sibling of the source directory.15:08
@wikingthese were the warnings15:08
@wikingbut still15:08
@wikingthere are tons of files in build15:08
@wikingincluding the .project file15:08
@wikingthoralf: yeah maybe you ran once in a bad way cmake15:08
@wikingtry rm -rf build15:09
@wikingmkdir build && cd build15:09
@wikingand then run again the cmake with the parameters u want15:09
thoralfwiking: Okay, it works with a clean checkout.15:12
thoralfwiking: Now I'm trying to remove the build artifacts from the root directory and see it fixes15:13
@wikingthoralf: there's a git command for this15:13
@wikingdontremeber it again... lisitsyn wrote it yesterday15:13
thoralfYou don't mean git stash?15:13
@wikingthoralf: nono15:13
@wikingthoralf: it's a reset that removes all non repo related files as well15:14
lisitsynwiking: git clean -dfx15:14
@wikingthoralf: ^15:14
@wikinglisitsyn: heheh i saved it now :)15:14
@wikingthnx15:14
thoralfwiking: Works now, thanks.15:15
@wikingnw15:16
@wikingthoralf: eclipse?15:16
thoralfwiking: Importing "build" as a new eclipse project and hitting build works.15:20
sonne|workthoralf: so how does this work?15:21
sonne|workthoralf: you create a build dir and then how do you get the eclipse project file?15:22
thoralfsonne|work: build$ cmake -G "Eclipse CDT4 - Unix Makefiles" ..15:22
thoralfBut actually it's not what I want.15:22
thoralfSince the build files are not under version control.15:22
@wikingok travis is back to green15:23
sonne|workthoralf: ? I thought they are just symlinks?15:23
@wikingthoralf: ?15:23
@wikingthoralf: what files are not under version control?15:24
@wikingthoralf: http://www.vtk.org/Wiki/Eclipse_CDT4_Generator15:24
@wikingthoralf: http://www.vtk.org/Wiki/Eclipse_CDT4_Generator#Version_Control_Integration_in_Eclipse15:25
@wikingi guess something like this u r looking for15:25
@wikingthoralf: or http://www.cmakebuilder.com/15:25
@wikingthoralf: please let me know your experience15:25
@wikingand if i can ask15:25
@wikingwrite couple of lines if u get a something u like15:26
thoralfwiking, sonne|work: If I create a prject from build/, then I don't see the source files in the browser.  But I was wrong as-in there is no copy of the source files in the build directory.15:26
@wikingso we can add that to the readme15:26
thoralfI just can't navigate.15:26
@wikingthoralf: yes there's no copy15:26
thoralfBut I can build.15:26
@wikingthoralf: take a look at those tutorials i've just sent u15:26
@wikingthey fix the problem u want15:26
@wikingcmake -G"Eclipse CDT4 - Unix Makefiles" -DECLIPSE_CDT4_GENERATE_SOURCE_PROJECT=TRUE ../certi_src15:27
@wikingThis will create your normal project in the build tree and additionally an extra project in the source tree, we call it the "source-project". In Eclipse you can then import this source-project the same way as you import the normal project. This way you'll have two (or more) projects, one for browsing the sources and doing version control, the other for building your project.15:27
sonne|workthoralf: if you get this to work I will use it too :D15:27
@wikingsonne|work: it can be15:27
@wikingsonne|work: tons of ppl use cmake projects with eclipse15:27
@wikingu just need to know the right flags for cmake15:27
@wikingto generate what u r looking for15:27
thoralfOh.  Inside "targets"/"[lib] shogun"/"source" I get an alphabetical list of all source files, but no directories. ;)15:28
@wikingsonne|work: i mean when i have to do someting in m$ VS15:28
@wikingsonne|work: i use cmake to generate the solution file ;P15:28
@wikinghe following generators are available on this platform:15:29
@wiking Visual Studio 6             = Generates Visual Studio 6 project files.15:29
@wiking Visual Studio 7             = Generates Visual Studio .NET 2002 project15:29
@wiking                               files.15:29
@wiking Visual Studio 10            = Generates Visual Studio 10 project files.15:29
@wiking Visual Studio 11            = Generates Visual Studio 11 project files.15:29
@wiking Visual Studio 7 .NET 2003   = Generates Visual Studio .NET 2003 project15:29
@wiking                               files.15:29
@wiking Visual Studio 8 2005        = Generates Visual Studio 8 2005 project fil15:29
@wiking Visual Studio 9 2008        = Generates Visual Studio 9 2008 project fil15:29
@wiking Borland Makefiles           = Generates Borland makefiles.15:29
@wiking NMake Makefiles             = Generates NMake makefiles.15:29
@wiking NMake Makefiles JOM         = Generates JOM makefiles.15:29
@wiking Watcom WMake                = Generates Watcom WMake makefiles.15:29
@wiking MSYS Makefiles              = Generates MSYS makefiles.15:29
@wiking MinGW Makefiles             = Generates a make file for use with15:29
@wiking                               mingw32-make.15:29
@wikingthese are the generators on windows ;)15:29
@wikingso theoretically we can haz vs 2010 solution for shogun15:29
@wikingonly thing is that the compiler flags are not handled well15:29
@wikingthoralf: google: cmake eclipse generator15:30
@wikingthat's your friend to have all the funky stuff u want15:30
sonne|workwiking: nice indeed15:30
@wikingsonne|work: the good sideeffect of using cmake for VS solutions that u dont have to care/fidle around with importing for example VS 2006 solution to vs 201115:31
@wiking:P15:31
shogun-buildbotbuild #1706 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1706  blamelist: Thoralf Klein <thoralf.klein@zib.de>15:31
@wikingsonne|work: i have a feeling that i need to fix the .deb package generator now ;)15:32
@wikingsonne|work: have u seen that twitter line i've just copied like a week ago from gstreamer core developers?15:33
sonne|workwiking: no15:33
@wikingsonne|work: edward (one gstreamer core dev) said that having various binary packages shipped (like osx, deb, rpm etc) brought more developers to the project15:33
sonne|workyeah sure15:34
sonne|workwiking: more users == more developers15:34
sonne|workI would says sth like for 10000 users there are 100 who do a bug report and 1 who contributes15:35
@wikingsonne|work: #?guadec? ?#?gstreamer? "Doing binary releases for win, macosx, ios, android brought more developers"15:35
@wikingin our case ios + android is not something we r aiming15:35
sonne|workwiking: well do it! start with osx right?15:35
@wikingbut a win release yes15:35
@wikingsonne|work: done15:35
sonne|workwiking: we have a .dmg now?15:35
@wikingsonne|work: that was my first thing i did with cmake+cpack15:35
@wikingsonne|work: yes15:35
sonne|workwiking: but libshogun only right?15:36
@wikingsonne|work: gimme space to upload :)15:36
sonne|worknot python or so?15:36
@wikingsonne|work: no15:36
@wikingsonne|work: libshoung+allmodular15:36
sonne|workahh nice15:36
@wikingsonne|work: and u can choose from the installer what u wanna install15:36
sonne|worknice15:36
@wikinglike only libshogun + some modular interface15:36
sonne|workhow can one do this automatic (nightly)15:36
@wikingor header files as well...15:36
@wikingetc.15:36
@wikingsonne|work: make package15:36
sonne|workwell if I am on linux...15:36
@wikingsonne|work: then u get deb or rpm + tar.gz + tar.bz215:37
@wikingdepends what package generator is available15:37
@wikingsonne|work: for osx package u need osx15:37
@wikingsonne|work: as it's using the native package generator15:37
@wikingsonne|work: either we get gunnar to put that osx bot online soon15:37
@wikingsonne|work: or i can trick travis osx builder15:38
@wikingto geneerate the package and upload it to somewhere15:38
@wiking;)15:38
@iglesiasgwiking, so were you wrong or not finally? :)15:41
@wikingiglesiasg: wrong :(15:41
@iglesiasgwiking, any idea how to fix it?15:42
@wikingiglesiasg: your problem was that u tried to compile a broken commit ;)15:42
@wikingiglesiasg: git pull the HEAD of develop and compile now15:42
@wikingiglesiasg: should work as travis just went green15:42
@iglesiasgwiking, eehm ok15:42
@iglesiasgwiking, but I was compiling the HEAD before ;)15:43
@wikingiglesiasg: yes but at that moment it was broken15:43
@wiking;)15:43
@iglesiasgI see15:43
@iglesiasgthen it makes sense!15:43
@wikingindeed15:43
@iglesiasgremoving -Wshadow commit fixes it?15:44
@wikingsonne|work: the problem with the travis trick is that: a) which commit will be the nightly release b) the osx job on travis already exceeds a bit the soft time limit of travis15:44
@wikingiglesiasg: no... that's another fix15:44
@wikingiglesiasg: thoralf sent a PR that fixed his previous PR15:45
@iglesiasgwiking, so from the last time I commited15:45
@iglesiasgthe only new things are thoralf's commit and -Wshadow15:45
@iglesiasgwhat was failing in mine was compiling agains gtest15:45
@iglesiasgbecause of these undefined references15:45
@wikingiglesiasg: according to cdash that's not true15:45
@wikingiglesiasg: these were your problems http://maeth.com/cdash/viewBuildError.php?buildid=315:46
@wikingit has nothing to do with gtest15:46
@iglesiasgwiking, I think they were not15:46
@iglesiasglet me double check please15:46
@wikingiglesiasg: well your last build was this: http://maeth.com/cdash/viewBuildError.php?buildid=315:46
@wikingthat's for sure15:46
@wiking;P15:46
@iglesiasgwiking, http://pastebin.com/wcqmNYtn15:46
@wikingyes15:47
@wikingand after that15:47
@wikingyou did ctest -D Experimental15:47
sonne|workwiking: with the buildbot I am currently generating the nightly stuff - so I was hoping that we could do the releases on that15:47
@wikingand that clearly showed that thers's something else wrong....15:47
@iglesiasgok15:47
sonne|workwiking: so only when everything compiles just fine15:47
@wikingiglesiasg: and i dont know how u ended up having that error15:47
@wikingiglesiasg: tryagain15:47
@wikingsonne|work: yep.... get gunnar to put that osx machine online15:48
@wikingor someBODY DONATE US AN OSX BUILDBOT15:48
@wiking^^^^^^^^^^^^^^^15:48
@iglesiasgwhat about travis' OSX machines?15:49
@wikingiglesiasg: @wiking> sonne|work: the problem with the travis trick is that: a) which commit will be the nightly release b) the osx job on travis already exceeds a bit the soft time limit of travis15:52
@iglesiasgwiking, I see, thank you and sorry for making you repeat15:52
@wikingnw it's a c-p ;)15:53
@wikingor i can install as i said in a VB an osx on my server15:53
@wikingbut that implies some licensing problems :(15:54
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has joined #shogun15:54
travis-ci[travis-ci] it's Viktor Gal'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/1002484515:54
-!- travis-ci [~travis-ci@ec2-54-234-6-198.compute-1.amazonaws.com] has left #shogun []15:54
@wikingyey back to green15:54
@wiking"... you are granted a limited, non-transferable, non-exclusive license: [...] (iii) to install, use and run up to two (2) additional copies or instances of the Apple Software within virtual operating system environments on each Mac Computer you own or control that is already running the Apple Software."15:58
@wikingi do 2 copies of osx16:00
@wikingso some legal help here plz16:02
@wiking" to install, use and run up to two (2) additional copies or instances of the Apple16:02
@wikingSoftware within virtual operating system environments on each Mac Computer you own16:02
@wikingor control that is already running the Apple Software, for purposes of: (a) software16:02
@wikingdevelopment; (b) testing during software development; (c) using OS X Server; or (d)16:02
@wikingpersonal, non-commercial use."16:02
sonne|workIANAL16:03
-!- sonne|work [~sonnenbu@91-64-72-127-dynip.superkabel.de] has quit [Quit: Leaving.]16:03
lisitsynwiking: ^ look he feel nervous about not being a lawyer ;)16:04
@wiking:D16:05
@wikinglisitsyn: so what have u decided?16:06
-!- thoralf [~thoralf@enki.zib.de] has quit [Quit: Konversation terminated!]16:08
lisitsynwiking: about?!16:08
@wikinglisitsyn: -isystem vs -I16:09
lisitsynahh16:10
@wikinglisitsyn: and about whether i can haz oSX in VM16:10
@wiking:D16:10
@wikinglegally16:10
@wikinglisitsyn: although the problem will be that that VM will need about 3 gigs of ram16:10
@wikingand i really need that 3 gigs of ram on my server16:10
@wikingso16:10
@wikingwho has a spare server with spare 3 gigs of ram in it16:11
@iglesiasgwiking, it built smooth. Running ctest atm16:22
@wikinggreat!16:23
@wikingi think there was something there halfway broken16:23
@wikingthat's how u ended up having that problem16:23
@iglesiasgthis runs examples as well as unit tests, right?16:24
@wikingiglesiasg: everything it can16:24
@wikingu can restrict it by the -E and -R flags of ctest16:25
@wikingsee ctest --help output16:25
@wikingso u can do: i want only unit test ctest -R unit16:25
@iglesiasgthanks!16:25
@wikingyou want only python tests: ctest -R python16:25
@wikingor you want everything tested apart from python: ctest -E python16:25
@wikingof course it's good to have --output-on-failure as well16:26
@wikingjust to see what has really went wrong16:26
@iglesiasgthere were three things failing here16:26
@iglesiasgI guess the same ones you mentioned?16:26
@iglesiasgThe following tests FAILED:16:27
@iglesiasg 97 - integration-python_modular-tester (Failed)16:27
@iglesiasg332 - python_modular-evaluation_cross_validation_multiclass_storage (Failed)16:27
@iglesiasg350 - python_modular-features_string_char_compressed_modular (Failed)16:27
@wikingi mean u can 'make test' as well but that doesn't show u the failed shit16:27
@wikingiglesiasg: oh16:27
@wikingiglesiasg: no they should be good16:27
@wikingi mean on travis its good16:27
@wikingwith the head16:27
@wikingbut then again integration-python_modular-tester fails big time on my machine as well...16:27
@iglesiasgok16:27
@wikingbut not that other two python16:28
-!- HeikoS [~heiko@nat-163-6.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]16:28
@wikingiglesiasg: that integration-python_modular-tester only works on travis :P16:28
@iglesiasghehe16:28
@iglesiasgsomeone tuned it for travis16:28
@wikingand i guess sonney2k's machine16:28
@iglesiasgI didn't see in the output unit tests however16:28
@iglesiasgmaybe I missed them though16:28
@wikingiglesiasg: integration-python_modular-tester16:28
@wikingno16:28
@wiking--output-on-failure16:29
@wikingotherwise it won't print it16:29
@wikingjust that it failed16:29
-!- HeikoS [~heiko@nat-163-6.internal.eduroam.ucl.ac.uk] has joined #shogun16:29
-!- mode/#shogun [+o HeikoS] by ChanServ16:29
@wikingiglesiasg: if unit test fails then the whole gtest ouput is printed out...16:29
@iglesiasggot it!16:29
@iglesiasgre-running with --output-on-failure to see how it goes16:30
@wikingheheh i'll have to finish up something now in a windows project.. i'll try cmaking shogun :DDD16:32
@wikingis there like python for windows? :)16:33
@iglesiasgyes, it is16:36
@iglesiasgabout the other two tests failing in my machine16:36
@iglesiasgI think they probably happen because I didn't compile with gplk and lzo16:36
@iglesiasgbut that should be guarded somehow16:36
@wikinghahah this is a historical moment: http://pastebin.com/41qdJPBp16:38
@wiking:>16:38
@wikingbut basically nothing has been found16:39
lisitsynwiking: shogun on windows?16:39
@wikinglisitsyn: indeed16:39
@wikinglisitsyn: gotta finish some shitty work in m$ env so i thought to give it a go16:40
@iglesiasgdid the unit tests get much faster now?16:50
@wikingiglesiasg: dunno16:50
@wikingiglesiasg: but be aware that u might be running Release now :(16:50
@iglesiasgwiking, that means with optimizations, right?16:51
@iglesiasgthe old --disable-optimizations thingy16:51
@iglesiasgwithout s16:51
@wikingyes16:52
@wikingiglesiasg: -DCMAKE_BUILD_TYPE16:52
@wikingfor cmake16:52
@wikingi mean16:52
@wiking-DCMAKE_BUILD_TYPE=Debug16:52
@iglesiasgit was in Release, yes16:53
@wikingiglesiasg: sonney2k said that that should be by default16:53
@iglesiasgwiking, we have Release and Debug, right?16:54
@wikingand Distribution16:55
@iglesiasgI don't know yet how to feel about this https://soundcloud.com/recordsyourcomments/48-116:58
@HeikoSwiking, iglesiasg, lisitsyn, sonney2k,  I will submit the doc camp application now16:59
@HeikoSdid you all have a look?16:59
@wikingHeikoS: yeps16:59
@iglesiasgyeah!16:59
lisitsynHeikoS: I did17:00
lisitsynno serious changes though17:00
@HeikoSok17:00
@HeikoSthen Ill submit17:00
@HeikoSlets hope for the best :)17:00
@HeikoSPretty good text17:00
@HeikoSbetter than last year17:00
@wikingHeikoS: thnx heaps for putting this together!17:05
@HeikoSwelcome :)17:06
@HeikoSok submitted17:06
@wiking\o/17:07
@iglesiasglet's keep our fingers crosses17:20
@iglesiasgor crossed17:20
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Read error: Connection reset by peer]17:23
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]17:38
@iglesiasgwiking,  I didn't lie Kyle in the ML, did I?17:52
@iglesiasgI mean, now with cmake, configure won't work fine17:52
@iglesiasgmaybe I screwed though17:52
-!- HeikoS [~heiko@nat-163-6.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]17:54
-!- van51 [~van51@ppp-94-66-52-171.home.otenet.gr] has joined #shogun18:01
@wiking?18:10
@wikingwhaaat? :)))18:10
@wikingsorry i wasn't paying attetion :)18:10
@wikingaah mailing list18:10
@wikinglet me check18:10
@wikingok sent18:16
@wikingiglesiasg: hehehe18:18
@wiking1 minute difference18:18
@iglesiasgwiking, you are the fast guy! :P18:18
@wikingyeah more to the point18:19
@wikingbut it's great you went into a deeper discussion ;)18:19
@wikinglet's keep the users happy18:19
@wikingthat little we have18:19
@iglesiasghaha yeah18:19
@iglesiasgI think he must be one of the very few users (the only?) who uses Shogun from cutting edge branch18:20
@wiking:D18:20
@wikingyeah i mean maaaan18:21
@wikingmaster branch must be like18:21
@wikingsooooo outdated :)18:21
@wikingalthough not that much18:21
@wikingit was branched out like in april or something18:21
@wikingbut a lot of code got in the meanwhile18:21
@iglesiasgit depends on what he is using Shogun for, no doubt18:21
@iglesiasghave to dash off for some minutes now, see you later18:21
@wikinglaterz18:22
@wikingshould get python 3.3 testing working on trav18:22
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has joined #shogun18:24
@wikingctest -D ExperimentalMemCheck18:28
@wikingthis is naaasty ;)18:28
@wikingbut only with -DBUILD_DASHBOARD_REPORTS=ON18:28
@wikingmaybe we should just enable the whole CTest as is18:31
-!- van51 [~van51@ppp-94-66-52-171.home.otenet.gr] has quit [Quit: Leaving.]19:23
@wikinghttp://maeth.com/cdash/viewDynamicAnalysisFile.php?id=119:35
@wikinghavefun19:37
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun19:37
@wikingyou pickle2719:37
@wikingi mean YO119:37
@wiking!19:37
pickle27wiking: hey19:38
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has left #shogun []20:24
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has joined #shogun20:24
lisitsynwiking: do you know any way to add a 'synonym' for some class in other namespace?20:42
lisitsynlike create an alias of class A::C in namespace B20:42
lisitsyntypedef should be a way!20:46
@wiking:)20:55
@wikingindeed20:55
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat]21:55
-!- gsomix [~gsomix@109.169.185.159] has quit [Quit: Leaving]22:02
-!- thoralf [~thoralf@p5B32571C.dip0.t-ipconnect.de] has joined #shogun22:16
thoralfHello.22:16
@sonney2kthoralf, hey hey23:55
thoralfHey sonney2k23:55
@sonney2kwiking, lisitsyn - did we apply for doc camp now?23:55
thoralfJust writing my application.23:55
lisitsynsonney2k: well Heiko submitted sth23:56
thoralfHeiko told that individuals should apply too, right?23:56
lisitsynthoralf: ehm!23:56
@sonney2kthoralf, sure do it!23:57
lisitsynhow?23:57
lisitsynI must be missing something23:57
--- Log closed Sat Aug 10 00:00:08 2013