--- Log opened Fri Aug 12 00:00:07 2011 | ||
-!- in3xes [~in3xes@182.156.166.217] has quit [Ping timeout: 264 seconds] | 00:15 | |
serialhex | sonney2k: u around?? | 00:26 |
---|---|---|
-!- f-x [~user@117.192.211.222] has left #shogun ["ERC Version 5.3 (IRC client for Emacs)"] | 00:38 | |
CIA-87 | shogun: Baozeng Ding master * rf97132e / src/interfaces/ruby_modular/swig_typemaps.i : fix ruby stringlist bug by removing SGFREE(str) - http://bit.ly/pRYhgF | 06:34 |
CIA-87 | shogun: Baozeng Ding master * rfbcb1ec / (2 files in 2 dirs): fix ruby stringlist bug, add result=arr statement - http://bit.ly/qC0i1e | 06:34 |
CIA-87 | shogun: Soeren Sonnenburg master * r7594a76 / (2 files in 2 dirs): | 06:34 |
CIA-87 | shogun: Merge pull request #289 from sploving/master | 06:34 |
CIA-87 | shogun: ruby stringlist can work now - http://bit.ly/oLI3mt | 06:34 |
CIA-87 | shogun: Soeren Sonnenburg master * rdbfda65 / (10 files in 3 dirs): | 06:53 |
CIA-87 | shogun: Merge pull request #288 from frx/streaming_vw | 06:53 |
CIA-87 | shogun: Added StreamingVwFeatures. - http://bit.ly/q5yJV3 | 06:53 |
-!- in3xes [~in3xes@49.201.194.134] has joined #shogun | 08:52 | |
-!- blackburn [~blackburn@188.122.238.50] has joined #shogun | 11:16 | |
@sonney2k | blackburn, that is from the new serialization stuff - how did you get it? | 12:25 |
@sonney2k | blackburn, btw did the other interfaces compile? | 12:26 |
blackburn | created SerializableAsciiFile and tried to save_serializable | 12:26 |
@sonney2k | blackburn, if so it would be a good idea to commit the new typemaps | 12:26 |
blackburn | yes, all ok | 12:26 |
@sonney2k | ok then commit | 12:26 |
@sonney2k | and then we continue | 12:26 |
blackburn | I'll commit it with java readObject | 12:27 |
blackburn | soon | 12:27 |
-!- f-x [~user@117.192.212.86] has joined #shogun | 13:18 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 13:37 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 13:41 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 13:47 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 13:47 | |
blackburn | shogun-buildbot: status | 14:28 |
shogun-buildbot | cmdline_static: idle | 14:28 |
shogun-buildbot | csharp_modular: idle | 14:28 |
shogun-buildbot | java_modular: idle | 14:28 |
shogun-buildbot | libshogun: idle | 14:28 |
shogun-buildbot | lua_modular: idle | 14:28 |
shogun-buildbot | matlab_static: idle | 14:28 |
shogun-buildbot | octave_modular: idle | 14:28 |
shogun-buildbot | octave_static: idle | 14:28 |
shogun-buildbot | python_modular: idle | 14:28 |
shogun-buildbot | python_static: idle | 14:28 |
shogun-buildbot | r_modular: idle | 14:28 |
shogun-buildbot | r_static: idle | 14:28 |
shogun-buildbot | ruby_modular: idle | 14:28 |
blackburn | wow | 14:28 |
blackburn | stop stop | 14:28 |
blackburn | :D | 14:28 |
CIA-87 | shogun: Sergey Lisitsyn master * rcfab782 / (5 files): Added write/readObject for java serialization, added ignores and removed unnecessary modular classes - http://bit.ly/n42Fih | 14:34 |
CIA-87 | shogun: Sergey Lisitsyn master * r9778993 / src/interfaces/modular/Library.i : Fixed wrong deletion of SGNDArray<uint16_t> - http://bit.ly/q2iahU | 14:49 |
-!- f-x [~user@117.192.212.86] has quit [Ping timeout: 260 seconds] | 15:02 | |
blackburn | shogun-buildbot: status | 15:42 |
shogun-buildbot | cmdline_static: idle | 15:42 |
shogun-buildbot | csharp_modular: idle | 15:42 |
shogun-buildbot | java_modular: idle | 15:42 |
shogun-buildbot | libshogun: idle, last build 46m46s ago: failed test | 15:42 |
shogun-buildbot | lua_modular: idle | 15:42 |
shogun-buildbot | matlab_static: idle | 15:42 |
shogun-buildbot | octave_modular: idle | 15:42 |
shogun-buildbot | octave_static: idle | 15:42 |
shogun-buildbot | python_modular: idle | 15:42 |
shogun-buildbot | python_static: idle | 15:42 |
shogun-buildbot | r_modular: idle | 15:42 |
shogun-buildbot | r_static: idle | 15:42 |
shogun-buildbot | ruby_modular: idle | 15:42 |
blackburn | shogun-buildbot: help | 15:42 |
shogun-buildbot | Get help on what? (try 'help <foo>', or 'commands' for a command list) | 15:42 |
blackburn | shogun-buildbot: commands | 15:42 |
shogun-buildbot | buildbot commands: commands, dance, destroy, excited, force, hello, help, last, list, mute, notify, source, status, stop, unmute, version, watch | 15:42 |
@sonney2k | blackburn, I am still configuring the bot... | 16:09 |
@sonney2k | seems like it works more or less | 16:09 |
@sonney2k | shogun-buildbot status libshogun | 16:10 |
CIA-87 | shogun: Soeren Sonnenburg master * rfc12543 / (3 files in 2 dirs): fix remaining warnigns in classifier - http://bit.ly/p2Zc0l | 16:18 |
@sonney2k | shogun-buildbot, status libshogun | 16:21 |
shogun-buildbot | libshogun: building(compiling) | 16:21 |
@sonney2k | blackburn, did you get an email from the buildbot? | 16:22 |
shogun-buildbot | build #1 of libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/libshogun/builds/1 | 16:24 |
@sonney2k | hurray! | 16:25 |
blackburn | sonney2k: no, can't see any email | 16:26 |
@sonney2k | blackburn, not even on failure? | 16:27 |
blackburn | sonney2k: nope | 16:28 |
@sonney2k | hmmh | 16:29 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 16:36 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 16:36 | |
CIA-87 | shogun: Soeren Sonnenburg master * r99d18cc / src/shogun/classifier/svm/WDSVMOcas.cpp : remove unused 'end' variable - http://bit.ly/nZM1hl | 16:37 |
CIA-87 | shogun: Soeren Sonnenburg master * r0b64595 / (3 files in 2 dirs): | 16:41 |
CIA-87 | shogun: Merge pull request #290 from sploving/master | 16:41 |
CIA-87 | shogun: add ruby and csharp docs - http://bit.ly/n2aUHX | 16:41 |
CIA-87 | shogun: Soeren Sonnenburg master * r4c34008 / (6 files in 3 dirs): | 16:42 |
CIA-87 | shogun: Merge pull request #291 from frx/streaming_vw | 16:42 |
CIA-87 | shogun: Documented and renamed a few functions in parser/ring for readability. - http://bit.ly/nWH2Fo | 16:42 |
blackburn | sonney2k: what's up with memory usage? | 16:44 |
blackburn | did it get lower? | 16:44 |
@sonney2k | no idea | 16:45 |
@sonney2k | the buildbot didn't seem to pick up the build... | 16:45 |
@sonney2k | *sigh* | 16:45 |
@sonney2k | hmmhh now it does | 16:47 |
blackburn | sonney2k: it seems java serialization works | 16:51 |
blackburn | I'll check carefully a little bit later | 16:52 |
@sonney2k | blackburn, yay! now the buildbot is in super action http://shogun-toolbox.org:5347/waterfall | 16:55 |
blackburn | nice | 17:15 |
blackburn | sonney2k: hey much less warnings now | 17:19 |
-!- in3xes [~in3xes@49.201.194.134] has quit [Quit: Leaving] | 17:44 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 20:00 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 20:00 | |
-!- in3xes [~in3xes@49.202.137.17] has joined #shogun | 20:03 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 20:06 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 20:07 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Client Quit] | 20:07 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 20:08 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 20:26 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 20:27 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Client Quit] | 20:27 | |
-!- shogun-buildbot [~shogun-bu@87.118.92.43] has joined #shogun | 20:28 | |
-!- in3xes [~in3xes@49.202.137.17] has quit [Ping timeout: 260 seconds] | 20:28 | |
@sonney2k | blackburn, memory requirements are down to 300 MB? | 20:55 |
@sonney2k | could that be? | 20:55 |
@sonney2k | ahh no | 20:56 |
blackburn | sonney2k: not possible | 20:56 |
blackburn | I would believe ~600 | 20:56 |
@sonney2k | 1.6G | 20:59 |
blackburn | bad | 20:59 |
@sonney2k | but will full optimization | 21:00 |
blackburn | sonney2k: no idea how to lower it <1g | 21:02 |
@sonney2k | well lets see how much octave_modular requires | 21:05 |
@sonney2k | 1.9G already | 21:05 |
@sonney2k | 2.1 | 21:06 |
@sonney2k | 2.4 | 21:06 |
@sonney2k | <out_of_memory> | 21:07 |
@sonney2k | blackburn, look failures when build with all opts on and with all disabled http://shogun-toolbox.org:5347/waterfall | 21:08 |
@sonney2k | now we need to also run make check-tests / check-examples at some point and all good | 21:09 |
@sonney2k | and osx | 21:09 |
@sonney2k | and cygwin support | 21:09 |
@sonney2k | blackburn, btw everyone could install a buildbot-slave on his machine | 21:09 |
@sonney2k | then we can run things on more machines - so if you know s.o. with a mac / windows machine that is on all the time anyways - ask him :) | 21:10 |
blackburn | okay | 21:10 |
@sonney2k | blackburn, ahh and btw. I added links to github - e.g. if you click on http://shogun-toolbox.org:5347/console | 21:11 |
blackburn | sonney2k: how 'big' should be serialized SVMLight? | 21:11 |
@sonney2k | and then fc12543ea.. | 21:11 |
blackburn | nice work | 21:11 |
@sonney2k | then you will see the diff of this patch :) | 21:12 |
@sonney2k | we could even have back links to trac if we were really using it :) | 21:12 |
blackburn | sonney2k: what is nightly? | 21:12 |
@sonney2k | nightly is running at 19:00 UTC | 21:12 |
@sonney2k | and running once with everything in configure --enable-*'d | 21:13 |
blackburn | hm ok | 21:13 |
@sonney2k | and once everything disabled | 21:13 |
@sonney2k | as you can see both fail | 21:13 |
blackburn | arpack test fails | 21:13 |
@sonney2k | yeah, didn't have it installed on the build slave | 21:14 |
blackburn | but configure shouldn't fail if no arpack.. | 21:14 |
@sonney2k | it should of course | 21:14 |
@sonney2k | it is enable-arapch'd | 21:14 |
@sonney2k | arpack | 21:14 |
blackburn | ah | 21:14 |
@sonney2k | blackburn, the cool thing is that we could even create daily snapshots - but only when all tests pass... e.g. .tar.gz and .deb pkg's | 21:18 |
blackburn | sonney2k: seen mail from serialhex? | 21:19 |
@sonney2k | no | 21:19 |
@sonney2k | blackburn, now yes | 21:22 |
@sonney2k | perfect timeing :) | 21:22 |
@sonney2k | blackburn, what is mythical 15 milliseconds in fibheap? | 21:23 |
blackburn | sonney2k: reducing memory allocation rate | 21:23 |
blackburn | hrr | 21:23 |
blackburn | forgot to merge eugeniy's patch | 21:23 |
CIA-87 | shogun: Evgeniy Andreev master * rac3d853 / (246 files in 23 dirs): Merge branch 'master' of git://github.com/shogun-toolbox/shogun - http://bit.ly/n8h61M | 21:25 |
CIA-87 | shogun: Evgeniy Andreev master * rcb5c195 / (4 files): Implemented Hash set and added 'mythical 15 milliseconds' for Fibonacci heap - http://bit.ly/pedzQ3 | 21:25 |
CIA-87 | shogun: Soeren Sonnenburg master * r850010a / (4 files): | 21:25 |
CIA-87 | shogun: Merge pull request #292 from gsomix/master | 21:25 |
CIA-87 | shogun: Implemented Hash set and added 'mythical 15 milliseconds' for Fibonacci heap - http://bit.ly/pGJ80e | 21:25 |
@sonney2k | blackburn, sry | 21:25 |
@sonney2k | was reading his patch... | 21:25 |
blackburn | hmm he has already implemented hashset | 21:26 |
blackburn | nice | 21:26 |
@sonney2k | blackburn, problem is I don't knwo how reliable it is - | 21:26 |
@sonney2k | if we use it for mission critical stuff | 21:26 |
blackburn | sonney2k: fibonacci heap is pretty reliable | 21:26 |
blackburn | sonney2k: uh.. just noticed very bad thing with java serialization | 21:28 |
@sonney2k | ? | 21:28 |
blackburn | e.g. it is called for SVMLight first | 21:29 |
blackburn | then for SVM | 21:29 |
blackburn | then for SGObject, so on | 21:29 |
@sonney2k | hmmhh, why that? so what we would need is to only implement the serializable interface for the class deepest in the hierarchy | 21:31 |
@sonney2k | dammed | 21:31 |
@sonney2k | my home machine just rebooted | 21:31 |
@sonney2k | no idea why | 21:31 |
blackburn | I can't understand why it is called for each successor.. | 21:31 |
blackburn | but as I can see there are 5 temporary files | 21:31 |
@sonney2k | blackburn, is there any way that you can make it only call it once for the child class? | 21:32 |
blackburn | I'll try to investigate | 21:33 |
-!- gsomix [~gsomix@88.200.215.245] has joined #shogun | 21:35 | |
gsomix | hi all | 21:35 |
gsomix | sonney2k, collisions are solved by means of lists (chains). | 21:35 |
@sonney2k | blackburn, I mean could you somehow figure out that your object is in fact some 'base' class | 21:36 |
@sonney2k | gsomix, ahh great | 21:36 |
@sonney2k | gsomix, did you test it? | 21:36 |
gsomix | sonney2k, i tested a little bit. Tomorrow will continue. | 21:36 |
blackburn | sonney2k: yeah I'll try | 21:36 |
blackburn | I lost jdk sources again | 21:37 |
blackburn | will check how objectoutputstream works | 21:37 |
@sonney2k | gsomix, it is just before we replace the *crappy* CSet and replace it with yours we should know that it works :) | 21:37 |
blackburn | sonney2k: false alarm I'd say | 21:50 |
blackburn | but why called 5 times.. | 21:50 |
@sonney2k | :) | 21:54 |
gsomix | sonney2k, ok. I'm going to use valgrind. | 21:56 |
blackburn | always 5 times ?! | 21:56 |
gsomix | And to explore vim. Hehe. However, it does not matter. | 21:56 |
blackburn | sonney2k: your fault! ;) | 22:02 |
@sonney2k | ahh of course | 22:02 |
@sonney2k | I knew it :D | 22:02 |
@sonney2k | blackburn, so enlighten me | 22:02 |
@sonney2k | what did I do? | 22:02 |
blackburn | you forgot throws in writeObject signature :) | 22:02 |
blackburn | but I'm not sure it would help | 22:04 |
@sonney2k | aha | 22:05 |
blackburn | yeah didn't help | 22:07 |
@sonney2k | blackburn, ok even if you try some very simple example? | 22:07 |
@sonney2k | e.g. serializae | 22:07 |
@sonney2k | CLabels? | 22:07 |
blackburn | sonney2k: yeah 4 times for RealFeatures | 22:08 |
@sonney2k | blackburn, try Labels | 22:08 |
blackburn | sonney2k: 2 times for Labels?? | 22:09 |
@sonney2k | blackburn, can you think of anything simpler than labels? | 22:10 |
blackburn | sonney2k: SGObject I guess | 22:10 |
@sonney2k | blackburn, ok then try it | 22:11 |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 22:11 | |
shogun-buildbot | build #1 of octave_modular is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org:5347/builders/octave_modular/builds/1 blamelist: gsomix@gmail.com | 22:13 |
gsomix | O_o | 22:13 |
-!- in3xes [~in3xes@180.149.49.227] has quit [Client Quit] | 22:14 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 22:16 | |
@sonney2k | gsomix, hmmhh internal compiler error? | 22:19 |
@sonney2k | maybe my machine has memory issues | 22:19 |
gsomix | sonney2k, it's not me. о_О | 22:20 |
CIA-87 | shogun: Soeren Sonnenburg master * rdd165c5 / src/shogun/io/SerializableXmlReader00.cpp : fix further warning in io - http://bit.ly/qF2MiJ | 22:30 |
shogun-buildbot | build #3 of octave_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/octave_modular/builds/3 | 22:30 |
@sonney2k | gsomix, see everything good ^ | 22:33 |
gsomix | ^_^ | 22:33 |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 276 seconds] | 22:37 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 22:37 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 22:45 | |
blackburn | still didn't find an answer | 22:50 |
@sonney2k | blackburn, so when you serilaize sgobject | 22:50 |
@sonney2k | you have 1 ? | 22:50 |
blackburn | I can't instantiate SGObject | 22:51 |
@sonney2k | blackburn, because of get_name | 22:51 |
@sonney2k | ? | 22:51 |
blackburn | don't know, may be yes | 22:51 |
@sonney2k | so how about adding a dummy DummyObject derived from sgobject that doesn't do anything? | 22:51 |
blackburn | I'm pretty sure it is related | 22:51 |
blackburn | if SGObject - Kernel - GaussianKernel then 3 or so | 22:52 |
@sonney2k | but then your initial guess was correct - or? | 22:52 |
blackburn | yes but can't see any solution | 22:53 |
blackburn | SGObject writeObject | 22:55 |
blackburn | class org.shogun.Labels | 22:55 |
blackburn | /tmp/-1568551569shogun.tmp | 22:55 |
blackburn | Labels writeObject | 22:55 |
blackburn | class org.shogun.Labels | 22:55 |
blackburn | /tmp/-1706591360shogun.tmp | 22:55 |
blackburn | yes, I've modified SGObject writeObject and Labels writeObject | 22:55 |
@sonney2k | blackburn, so java's serialization works by telling all parental classes to serialize first and then serializing the current class right? | 23:03 |
blackburn | seems so | 23:03 |
@sonney2k | if so - then it would explain what we see | 23:03 |
@sonney2k | and then all we could do is some hack | 23:04 |
@sonney2k | like set a flag i_am_already_serialized in SGObject | 23:04 |
-!- f-x [~user@117.192.192.210] has joined #shogun | 23:06 | |
blackburn | sonney2k: nono, we should do it such way | 23:06 |
blackburn | I'll try to make it Externizable, not Serializable | 23:06 |
blackburn | Externilizable | 23:07 |
blackburn | shit | 23:07 |
blackburn | Externalizable | 23:07 |
blackburn | sonney2k: Externalizable fits much better to this task | 23:10 |
blackburn | really full control of serialization process | 23:10 |
@sonney2k | blackburn, how is that different? | 23:10 |
blackburn | there are some structure of serialization process | 23:10 |
blackburn | defaultWriteObject and the other things | 23:11 |
blackburn | in the Externalizable we only override write/read and voi-la | 23:11 |
@sonney2k | blackburn, yeah but what if write/read are overridden in all (parental) classes? | 23:12 |
@sonney2k | I mean don't we have the same problem? | 23:12 |
blackburn | there is no overriding here | 23:12 |
blackburn | read/write are private | 23:12 |
blackburn | Serialization uses reflection process to sort out the things here | 23:13 |
blackburn | if it is externalizable all should work exactly how we want | 23:13 |
@sonney2k | blackburn, so it means the class will be externalized w/o recursing to parental classes? | 23:14 |
blackburn | should be so | 23:14 |
blackburn | oh | 23:15 |
blackburn | I got wrong | 23:15 |
blackburn | in the serialization there are public read/write | 23:15 |
blackburn | err in externalization | 23:15 |
@sonney2k | yeah an in serialization they are private | 23:15 |
blackburn | sonney2k: but we should definitely use externalization | 23:16 |
blackburn | i.e. custom serialization is if we are serializing only half of array and 3/4 integer values | 23:16 |
blackburn | and externalization if we output object to say some .pdf | 23:17 |
@sonney2k | ok as you say :) | 23:18 |
blackburn | sonney2k: hooray! | 23:18 |
@sonney2k | ? | 23:18 |
blackburn | works | 23:18 |
blackburn | called 1 time | 23:19 |
@sonney2k | great! | 23:19 |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 23:20 | |
@sonney2k | and the buildbot is happy too http://shogun-toolbox.org:5347/console | 23:20 |
@sonney2k | alright happy end for today - cu tomorrow | 23:20 |
blackburn | sonney2k: is sparse bug fixed? | 23:21 |
blackburn | see you | 23:21 |
-!- in3xes [~in3xes@180.149.49.227] has quit [Remote host closed the connection] | 23:29 | |
-!- gsomix [~gsomix@88.200.215.245] has quit [Quit: Ухожу я от вас (xchat 2.4.5 или старше)] | 23:49 | |
--- Log closed Sat Aug 13 00:00:18 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!