IRC logs of #shogun for Sunday, 2011-11-13

« Saturday, 2011-11-12 Index Monday, 2011-11-14 »
--- Log opened Sun Nov 13 00:00:56 2011
-!- blackburn [~blackburn@188.168.3.83] has quit [Quit: Leaving.]00:28
-!- blackburn [~blackburn@188.168.14.187] has joined #shogun10:59
-!- Netsplit *.net <-> *.split quits: shogun-buildbot, CIA-3714:24
-!- Netsplit *.net <-> *.split quits: @sonney2k, sonne|work, blackburn14:25
-!- Netsplit over, joins: blackburn, sonne|work, CIA-37, shogun-buildbot14:27
-!- Netsplit over, joins: @sonney2k14:28
-!- Netsplit *.net <-> *.split quits: sonne|work, blackburn14:44
-!- Netsplit *.net <-> *.split quits: shogun-buildbot, CIA-3714:44
-!- Netsplit over, joins: shogun-buildbot, CIA-3714:45
-!- sonne|work [~sonnenbu@194.78.35.195] has joined #shogun14:47
-!- blackburn [~blackburn@188.168.14.187] has joined #shogun14:47
@sonney2kblackburn, what do we need before we can release?21:03
blackburnsonney2k: oh I'm pretty unsure about it21:04
blackburnsonney2k: I can't recall anything we should21:04
@sonney2kI mean cygwin, osx work quite ok now21:04
@sonney2kI mean we could finish this .h -> .cpp transition21:04
@sonney2kbut that is still quite some work21:04
blackburnis it so necessary?21:04
blackburnerr that necessary21:04
@sonney2kwell it leads to much more readable header files and reduced compile times (only libshogun will be recompiled if the .cpp file is the only one that is changed)21:05
@sonney2kbesides that I don't know what else we wanted for 1.1.021:06
blackburnsonney2k: too bad we can't do 1.1.0rc1 :)21:06
blackburnwell we can but it is the same as 1.1.021:06
blackburncause we have 3.5 users totally21:07
@sonney2k3.5 ?21:07
blackburnjoke21:07
blackburnnot that much I mean21:07
@sonney2kblackburn, could you handle more bug reports / questions?21:07
blackburnsonney2k: if they was about dimension reduction :D21:08
blackburnwere*21:08
@sonney2k...the website was down recently for just a few hrs and a couple of people already complained...21:08
@sonney2kso yes we have quite some users21:08
@sonney2kheh21:08
blackburnwe need shogun users census21:09
@sonney2kthat's just some tiny fraction of shogun21:09
blackburnto make me believe we have users :)21:09
@sonney2kwe have enough traffic on the mailinglist...21:10
@sonney2kblackburn, ahh for the release we need to run tests again21:10
blackburnsonney2k: I would like to release on next weekend21:11
@sonney2kwhy weekend?21:11
blackburnsonney2k: cause I'm fully available only on weekends )21:11
@sonney2kyour b'day?21:11
blackburn:D21:11
@sonney2kand I am usually dead on weekends21:11
blackburnno, my b'day is in march21:11
blackburnthat's a trouble21:12
blackburnso what do you suggest?21:12
@sonney2kI would prefer to release ASAP21:12
@sonney2ki.e. we know that regression tests still work21:12
blackburnwell what about friday?21:12
blackburn:D21:12
@sonney2ktoday21:13
blackburnoh today21:13
blackburnokay21:13
@sonney2kor whenever we make it21:13
blackburnwhy not21:13
@sonney2kcould you update NEWS please?21:13
@sonney2kI don't know if today is possible - we will see what the tests say21:13
blackburnsonney2k: will you run tests?21:14
@sonney2kdoing already21:14
blackburnok21:14
blackburnwill update news in a min21:14
@sonney2kok crashes ahead21:16
blackburnsonney2k: what crashes?21:17
@sonney2khmmhh maybe clang related... recompiling21:20
@sonney2kblackburn, you have to generate test stuff21:32
@sonney2kconverter_diffusionmaps_modular.py setting 1/2               EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_diffusionmaps_modular0.txt'21:32
@sonney2kconverter_diffusionmaps_modular.py setting 2/2               EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_diffusionmaps_modular1.txt'21:32
@sonney2kconverter_hessianlocallylinearembedding_modular.py setting 1/2 EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_hessianlocallylinearembedding_modular0.txt'21:32
@sonney2kconverter_hessianlocallylinearembedding_modular.py setting 2/2 EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_hessianlocallylinearembedding_modular1.txt'21:32
@sonney2kconverter_isomap_modular.py setting 1/1                      EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_isomap_modular0.txt'21:32
@sonney2kconverter_kernellocallylinearembedding_modular.py setting 1/2 EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_kernellocallylinearembedding_modular0.txt'21:32
@sonney2kconverter_kernellocallylinearembedding_modular.py setting 2/2 EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_kernellocallylinearembedding_modular1.txt'21:32
@sonney2kconverter_kernellocaltangentspacealignment_modular.py setting 1/2 EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_kernellocaltangentspacealignment_modular0.txt'21:32
@sonney2kconverter_kernellocaltangentspacealignment_modular.py setting 2/2 EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_kernellocaltangentspacealignment_modular1.txt'21:32
@sonney2kconverter_laplacianeigenmaps_modular.py setting 1/2          EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_laplacianeigenmaps_modular0.txt'21:32
@sonney2kconverter_laplacianeigenmaps_modular.py setting 2/2          EXCEPTION [Errno 2] No such file or directory: '../../../testsuite/tests/converter_laplacianeigenmaps_modular1.txt'21:32
@sonney2kconverter_linearlocaltangentspacealig21:32
blackburnuff I don't really have much time for that right now :(21:33
blackburnsonney2k: any magic here?21:33
@sonney2kno just run the generator.py with these files21:34
blackburnsonney2k: btw I would like to not do that, it could vary very much but it is ok21:34
@sonney2kvary very much? how so21:34
@sonney2k?21:34
blackburnsonney2k: +-1 e.g.21:34
blackburnwe don't handle it as I remember21:35
@sonney2kwhy / how?21:35
blackburnsonney2k: eigenmethods21:35
@sonney2kI still dont' understand21:35
blackburnboth v and -v are eigenvectors21:35
@sonney2kand?21:35
@sonney2kisnt' what we do deterministic?21:35
blackburnno, it is not very deterministic21:36
blackburnI use at least two eigensolvers21:36
blackburnlapack one and arpack21:36
blackburnso results may vary21:36
blackburnit is perfectly ok for dimension reduction but tests will fail21:36
@sonney2kbut for lapack + arpack it is deterministic21:37
blackburne.g. if we have constructed feature matrix [1.0, 2.0; 4.0, 5.0]21:37
blackburnit is ok to get [-1.0, -2.0; 4.0, 5.0]21:37
blackburnand so on21:37
shogun-buildbotbuild #46 of nightly_default is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/4621:37
@sonney2kyeah but if you call it with arpack & lapack it will always be -1.0, -2.0, ...21:38
blackburnno, lapack can use random initial eigenvector21:38
blackburnare you suggesting to do three tests: with lapack, with arpack and with arpack+superlu?21:38
blackburnI would like to implement unit-tests working in different way21:39
blackburnthey could test result not by |A-B| but by target function or so21:39
blackburnsonney2k: can you remember any fixes I should mention?21:42
@sonney2kblackburn, can you seed lapack?21:42
@sonney2kgit log21:42
blackburnno, I don't think so21:42
@sonney2khmmhh21:42
@sonney2kbad21:42
blackburnsonney2k: I'm sure that typical ||A-B|| tests won't work here21:43
blackburnthey all are not that stable21:43
blackburnill-posed everywhere21:43
@sonney2kis there anything deterministic in your algos ?21:45
@sonney2kI mean we can check for in the test?21:45
blackburnsonney2k: yes, e.g. check distance matrices for MDS21:46
blackburnor check barycentric coordinates in LLE21:46
blackburnetc21:46
shogun-buildbotbuild #329 of octave_static is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/32921:46
blackburnsonney2k: I have no idea how to generalize it21:47
blackburnbut know how to test21:47
@sonney2kthen do that in the example and return a 'stable' result21:48
blackburnsonney2k: return error function you mean?21:48
shogun-buildbotbuild #330 of r_static is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/r_static/builds/330  blamelist: sonne@debian.org, blackburn91@gmail.com21:48
blackburnyes that would work smoothly21:48
shogun-buildbotbuild #346 of cmdline_static is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/346  blamelist: blackburn91@gmail.com21:49
blackburnsonney2k: but not now21:50
shogun-buildbotbuild #330 of octave_static is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/330  blamelist: sonne@debian.org, blackburn91@gmail.com21:51
shogun-buildbotbuild #322 of python_static is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/322  blamelist: sonne@debian.org, blackburn91@gmail.com21:51
CIA-37shogun: Sergey Lisitsyn master * r54fc133 / src/NEWS : Update NEWS - http://git.io/jtDo7A21:55
@sonney2klooks like things crash due to the custom kernel change...22:13
blackburnsonney2k: whoaa are you sure?23:02
--- Log closed Mon Nov 14 00:00:02 2011
« Saturday, 2011-11-12 Index Monday, 2011-11-14 »

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!