--- Log opened Fri Oct 14 00:00:10 2011 | ||
--- Day changed Fri Oct 14 2011 | ||
@sonney2k | blackburn, the only other idea regarding reducing 'wrapper code size' is to remove Array | 00:00 |
---|---|---|
@sonney2k | I am afraid I don't know what else to remove... | 00:00 |
blackburn | will it help that much? | 00:00 |
@sonney2k | no | 00:00 |
@sonney2k | but any templated class would | 00:00 |
@sonney2k | I mean currently each tempalted class exists for ~8 data types | 00:00 |
@sonney2k | -> 8 wrappers | 00:01 |
blackburn | yes | 00:01 |
@sonney2k | grep template interfaces/modular/* | wc -l | 00:01 |
@sonney2k | 258 | 00:01 |
@sonney2k | grep template interfaces/modular/* | grep CArray | wc -l | 00:02 |
@sonney2k | 30 | 00:02 |
@sonney2k | better than nothing... | 00:02 |
blackburn | sonney2k: how can I handle CCache with steal_feature_matrix? | 00:03 |
blackburn | it corrupts stealed memory | 00:03 |
blackburn | when deleting | 00:03 |
@sonney2k | Builds of shogun have failed on several architectures because | 00:04 |
@sonney2k | SWIG's output proved too much for their compilers to handle; per | 00:04 |
@sonney2k | https://buildd.debian.org/status/package.php?p=shogun&suite=sid | 00:04 |
@sonney2k | the build blew out on ruby on s390 and on octave on i386, ia64, | 00:04 |
@sonney2k | powerpc, and sparc. Could you please take a look? | 00:04 |
blackburn | uh | 00:04 |
@sonney2k | debian bug report | 00:06 |
shogun-buildbot | build #226 of octave_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_modular/builds/226 blamelist: sonne@debian.org | 00:06 |
@sonney2k | ahh we are finally getting octave_modular bugs - very good :) | 00:08 |
blackburn | I was wondering why it is not working: | 00:10 |
blackburn | return SGMatrix<ST>(feature_matrix, num_features, num_vectors); | 00:10 |
blackburn | feature_matrix = NULL; | 00:10 |
blackburn | feature_matrix_num_vectors = 0; | 00:10 |
blackburn | feature_matrix_num_features = 0; | 00:10 |
blackburn | num_features = 0; | 00:10 |
blackburn | num_vectors = 0; | 00:10 |
blackburn | :D | 00:10 |
blackburn | not a cache bug | 00:11 |
blackburn | as expected | 00:11 |
@sonney2k | blackburn, clean the cache / preprocs | 00:12 |
blackburn | sonney2k: ? how? | 00:12 |
@sonney2k | look at Features.h for preprocs | 00:12 |
CIA-3 | shogun: Soeren Sonnenburg * r78c8694 / src/shogun/features/StringFeatures.h : fix documentation of clean_feature_vectors - http://git.io/hHd-yA | 00:13 |
blackburn | sonney2k: how to handle with cache? | 00:13 |
blackburn | there is no clean | 00:14 |
@sonney2k | feature_cache = NULL | 00:14 |
blackburn | ?! | 00:14 |
@sonney2k | also drop subset | 00:15 |
blackburn | sonney2k: not really needed - it is intended that features will be deleted after stealing | 00:15 |
@sonney2k | still | 00:15 |
@sonney2k | if someone then accesses this object later it should die sanely | 00:16 |
@sonney2k | so better remove more | 00:16 |
blackburn | k | 00:16 |
@sonney2k | java / ruby modular went through | 00:16 |
@sonney2k | python too | 00:16 |
@sonney2k | csharp had an internal error (gcc4.6) | 00:17 |
@sonney2k | blackburn, btw just one mad idea - shouldn't it be possible to somehow split the huge generated swig file? | 00:19 |
@sonney2k | and then later just join it for linking? | 00:20 |
blackburn | no idea | 00:20 |
blackburn | how can it help? | 00:20 |
blackburn | swig file = .cxx? | 00:20 |
@sonney2k | blackburn, well compiling files of size 1MB is better than compiling files of size 20MB | 00:21 |
blackburn | I guess compiling 20*1mb will took even a little more than 20mb | 00:22 |
@sonney2k | blackburn, but it can be done on multiple CPUs and won't require 3G of memory but 1/20th of that | 00:22 |
blackburn | hmm yes | 00:22 |
blackburn | I have no idea how to get it done | 00:22 |
blackburn | uh there are a lot of files to commit | 00:24 |
blackburn | I'm sorry but I have no power to split it into commits | 00:24 |
shogun-buildbot | build #283 of python_static is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/283 blamelist: sonne@debian.org | 00:25 |
CIA-3 | shogun: Sergey Lisitsyn * r04b6de9 / (110 files in 12 dirs): Introduced converters module, moved DR preprocessors to it, fixed examples, updated other classes according to changes - http://git.io/82jByg | 00:30 |
blackburn | sonney2k: a little intrusive change :D | 00:31 |
shogun-buildbot | build #284 of python_static is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/284 | 00:37 |
@sonney2k | blackburn, I've learned at work some extreme programming... | 00:40 |
@sonney2k | and one rule there is that someone else has to review the code before it gets accepted into the project | 00:41 |
@sonney2k | I think it would be a nice idea to apply this here too | 00:41 |
@sonney2k | then we could each work on our local github forks | 00:41 |
@sonney2k | and issue pull requests for each other :) | 00:41 |
@sonney2k | and you would know what I am doing and vice versa | 00:41 |
blackburn | sonney2k: for only two of us that would be a waste of time | 00:42 |
@sonney2k | I don't think so | 00:42 |
@sonney2k | it helps finding bugs early | 00:42 |
blackburn | sonney2k: but doing all the things through pull requests is a good thing | 00:42 |
blackburn | sonney2k: have you seen how it is done in scikit-learn? | 00:42 |
@sonney2k | what? | 00:42 |
blackburn | sonney2k: they do a lot of pull requests | 00:43 |
blackburn | I guess there is no direct access | 00:43 |
blackburn | sonney2k: but we both doing a lot of small fixes, etc | 00:43 |
@sonney2k | well there is but I think it is just better that code gets a review... | 00:43 |
blackburn | sure | 00:43 |
@sonney2k | I mean usually it is easy :) | 00:44 |
blackburn | sonney2k: so you are doing *extreme* programming at your job? | 00:44 |
@sonney2k | I am doing extreme extreme programming | 00:45 |
@sonney2k | or extreme^3 programming even ;-) | 00:45 |
blackburn | lucky one, I do only bugfixing on my job :D | 00:46 |
blackburn | cause netcracker oss is a piece of shit made by crazy students | 00:46 |
blackburn | (like me) | 00:46 |
@sonney2k | blackburn, what is this: examples/undocumented/python_modular/serialized_svm.bz2 | 00:46 |
blackburn | oh sh | 00:46 |
blackburn | junk | 00:46 |
@sonney2k | and examples/undocumented/python_modular/vw_cache.dat.cache | 00:47 |
@sonney2k | in your last commit | 00:47 |
CIA-3 | shogun: Sergey Lisitsyn * r864758f / (2 files): Removed junk from the last commit - http://git.io/CLvrTA | 00:47 |
@sonney2k | btw put the copyright to sth reasonable (to you for example) | 00:47 |
@sonney2k | blackburn, another crap file : src/interfaces/python_modular/Converter. | 00:48 |
@sonney2k | blackburn, and you forgot the documentation | 00:49 |
@sonney2k | blackburn, btw you can shorten statements like this: | 00:49 |
@sonney2k | SGMatrix<float64_t> new_feature_matrix = SGMatrix<float64_t>(m_target_dim,N); | 00:49 |
@sonney2k | SGMatrix<float64_t> new_feature_matrix(m_target_dim,N); | 00:49 |
blackburn | I know | 00:49 |
shogun-buildbot | build #210 of lua_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/lua_modular/builds/210 blamelist: blackburn91@gmail.com | 00:50 |
shogun-buildbot | build #218 of csharp_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/csharp_modular/builds/218 | 00:50 |
@sonney2k | blackburn, really lots of changes... but should be much nicer now | 00:51 |
@sonney2k | and one could even move the *obtain_from* functions from and turn them into converters | 00:51 |
blackburn | at least we can do embeddings in more clear way | 00:52 |
@sonney2k | anyway it is late and time to sleep! | 00:52 |
blackburn | I'm deadly tired :D | 00:52 |
blackburn | O(you) | 00:52 |
@sonney2k | and I am sick even | 00:52 |
blackburn | oh | 00:52 |
blackburn | I hope you will recover soon | 00:53 |
@sonney2k | but slept long hours at day time | 00:53 |
@sonney2k | th | 00:53 |
@sonney2k | x | 00:53 |
@sonney2k | anyway | 00:53 |
@sonney2k | sleep time! | 00:54 |
CIA-3 | shogun: Sergey Lisitsyn * r6e41476 / (3 files): Improved converters doc - http://git.io/pq4PgQ | 00:54 |
@sonney2k | blackburn: error in preprocessor_isomap_modular.lua | 00:55 |
@sonney2k | ================================================================================ | 00:55 |
@sonney2k | lua: preprocessor_isomap_modular.lua:13: attempt to call method 'apply_to_feature_matrix' (a nil value) | 00:55 |
@sonney2k | stack traceback: | 00:55 |
@sonney2k | preprocessor_isomap_modular.lua:13: in function 'preprocessor_isomap_modular' | 00:55 |
@sonney2k | preprocessor_isomap_modular.lua:20: in main chunk | 00:55 |
@sonney2k | you missed one example it seems | 00:55 |
blackburn | uh | 00:55 |
blackburn | ok | 00:55 |
blackburn | min | 00:55 |
* sonney2k hugs the buildbot | 00:55 | |
CIA-3 | shogun: Sergey Lisitsyn * r573e955 / src/interfaces/python_modular/Converter.i : Removed Converter.i junk from python_modular - http://git.io/ghESwg | 00:56 |
@sonney2k | blackburn, we need to get rid of ant in java | 00:57 |
blackburn | sonney2k: what do you propose? | 00:57 |
@sonney2k | it is really a useless requirement | 00:57 |
CIA-3 | shogun: Sergey Lisitsyn * r5cec685 / (2 files): Updated lua isomap example - http://git.io/7WEsTQ | 00:57 |
@sonney2k | just use jar manually to create the pkg | 00:57 |
@sonney2k | like I use gac for c# | 00:57 |
blackburn | I'm not a pro with doing such things | 00:58 |
@sonney2k | me neither | 00:58 |
@sonney2k | but ant is really not required just to create a .jar | 00:58 |
@sonney2k | anyway I am done for now | 00:59 |
@sonney2k | cu | 00:59 |
blackburn | see you | 01:01 |
-!- blackburn [~blackburn@188.168.2.213] has quit [Quit: Leaving.] | 01:04 | |
CIA-3 | shogun: Sergey Lisitsyn * r87344b7 / src/shogun/converter/LocallyLinearEmbedding.h : Fixed LLE doc - http://git.io/SYM1yw | 01:08 |
shogun-buildbot | build #292 of octave_static is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/292 blamelist: blackburn91@gmail.com | 01:27 |
shogun-buildbot | build #212 of python_modular is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/212 blamelist: blackburn91@gmail.com | 01:54 |
shogun-buildbot | build #212 of lua_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/lua_modular/builds/212 | 01:59 |
shogun-buildbot | build #213 of python_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/213 | 02:21 |
CIA-3 | shogun: Soeren Sonnenburg * r85b2e68 / (2 files in 2 dirs): fix documentation - http://git.io/_hbQIg | 09:39 |
CIA-3 | shogun: Soeren Sonnenburg * r8097efe / (2 files): fix octave examples - http://git.io/wI3xYw | 09:53 |
shogun-buildbot | build #211 of ruby_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/ruby_modular/builds/211 blamelist: sonne@debian.org | 10:04 |
shogun-buildbot | build #215 of lua_modular is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/lua_modular/builds/215 blamelist: sonne@debian.org | 10:21 |
shogun-buildbot | build #212 of ruby_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/ruby_modular/builds/212 | 10:24 |
CIA-3 | shogun: Soeren Sonnenburg * rbcb7967 / src/NEWS : update news - http://git.io/aLDGXg | 13:21 |
shogun-buildbot | build #216 of lua_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/lua_modular/builds/216 | 13:39 |
shogun-buildbot | build #213 of ruby_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org/buildbot/builders/ruby_modular/builds/213 blamelist: sonne@debian.org | 14:05 |
-!- alesis-novik [~alesis@dhcp-90-089.inf.ed.ac.uk] has joined #shogun | 14:05 | |
-!- blackburn [~blackburn@188.168.4.209] has joined #shogun | 20:15 | |
-!- alesis-novik [~alesis@dhcp-90-089.inf.ed.ac.uk] has quit [Quit: Leaving] | 23:52 | |
--- Log closed Sat Oct 15 00:00:51 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!