| --- Log opened Thu Aug 11 00:00:59 2011 | ||
| blackburn | sonney2k: what's up im Westen Front? :D | 00:03 |
|---|---|---|
| @bettyboo | strange | 00:03 |
| @sonney2k | blackburn, looking at how to send out blame emails | 00:03 |
| @sonney2k | (for buildbot) | 00:03 |
| -!- shogun-irclog [~shogun@7nn.de] has quit [Ping timeout: 260 seconds] | 00:03 | |
| --- Log closed Thu Aug 11 00:03:51 2011 | ||
| --- Log opened Thu Aug 11 00:03:55 2011 | ||
| -!- shogun-irclog [~shogun@7nn.de] has joined #shogun | 00:03 | |
| -!- Irssi: #shogun: Total of 10 nicks [3 ops, 0 halfops, 0 voices, 7 normal] | 00:03 | |
| -!- Irssi: Join to #shogun was synced in 4 secs | 00:03 | |
| blackburn | sonney2k: what about sparse? | 00:04 |
| @sonney2k | blackburn, hmmhh it seems we could have some irc thingy too | 00:04 |
| blackburn | for what? | 00:04 |
| @sonney2k | blackburn, won't work on this now | 00:04 |
| @sonney2k | for buildbot | 00:04 |
| @sonney2k | seems like a bot to query build status etc | 00:05 |
| blackburn | sonney2k: let it send sms to my mobile phone ahaha | 00:05 |
| @sonney2k | blackburn, even better lets call you and read out the bug report aloud :D | 00:05 |
| blackburn | yeah | 00:06 |
| blackburn | no, let it sing that | 00:06 |
| blackburn | with frank sinatra's voice | 00:06 |
| @sonney2k | I think darth vaders voice would be more convincing | 00:06 |
| blackburn | singing darth vader could be more strange for sure :D | 00:07 |
| blackburn | or even darth maul | 00:07 |
| blackburn | hrrhrhr | 00:07 |
| blackburn | sonney2k: can't get %extend working :( | 00:14 |
| @sonney2k | what is going on? | 00:14 |
| blackburn | I tried to add macro | 00:14 |
| blackburn | but don't know what to do with template param | 00:14 |
| blackburn | %define SERIALIZABLE_DUMMY(SWIGCLASS) | 00:15 |
| blackburn | %extend SWIGCLASS { | 00:15 |
| blackburn | bool save_serializable(CSerializableFile* file, const char* prefix="") { return false; }; | 00:15 |
| blackburn | bool load_serializable(CSerializableFIle* file, const char* prefix="") { return false; }; | 00:15 |
| blackburn | }; | 00:15 |
| blackburn | %enddef | 00:15 |
| blackburn | that's how I did that | 00:15 |
| blackburn | sonney2k: SERIALIZABLE_DUMMY(SGMatrix) or SERIALIZABLE_DUMM(SG_MATRIX<class T>) or ...? | 00:16 |
| @sonney2k | blackburn, first of all is it working w/o templates? | 00:17 |
| blackburn | sonney2k: without %template things? | 00:18 |
| @sonney2k | blackburn, no | 00:18 |
| blackburn | ah without<>? | 00:18 |
| @sonney2k | when you try that with some class that is not templated | 00:18 |
| blackburn | ah | 00:18 |
| blackburn | will try to check | 00:18 |
| blackburn | Since the C++ compiler will hopefully check this when it compiles the resulting wrapper file, there is no practical reason for SWIG to duplicate this functionality (besides, none of the SWIG developers are masochistic enough to want to implement this right now). | 00:24 |
| blackburn | :D | 00:24 |
| blackburn | HOW SLOW IT IS | 00:31 |
| @sonney2k | blackburn, what is slow? | 00:31 |
| blackburn | recompiling the whole modular .i | 00:31 |
| blackburn | sonney2k: yeah it works for non-template | 00:33 |
| @sonney2k | very good | 00:35 |
| CIA-87 | shogun: Soeren Sonnenburg master * r1537fd1 / src/shogun/regression/svr/SVRLight.cpp : fix further unused variable warnings - http://bit.ly/plBWYf | 00:35 |
| CIA-87 | shogun: Soeren Sonnenburg master * r1baaf47 / src/shogun/base/Parameter.cpp : fix case not handled in switch - http://bit.ly/mXW4dT | 00:35 |
| @sonney2k | blackburn, for templated classes you really need to use the | 00:37 |
| @sonney2k | template<class T> class SGVector | 00:37 |
| @sonney2k | syntax | 00:37 |
| @sonney2k | so just a class Foo | 00:38 |
| @sonney2k | won't work | 00:38 |
| blackburn | %extend template<class T> class SGVector { ... };?? | 00:38 |
| @sonney2k | yes | 00:38 |
| blackburn | are you sure? | 00:38 |
| @sonney2k | no :) | 00:38 |
| blackburn | I really want to finish it today | 00:39 |
| blackburn | sonney2k: wrong | 00:40 |
| blackburn | who wants to be a millionaire | 00:40 |
| @sonney2k | ok and what does SERIALIZABLE_DUMMY(SGMatrix<float64_t>) do? | 00:41 |
| blackburn | %extend thing | 00:41 |
| @sonney2k | might be that it won't work for generic types | 00:41 |
| blackburn | I'll try SGStringList<T> thing | 00:42 |
| @sonney2k | the build bot is supposed to send out email this time | 00:45 |
| @sonney2k | we will see... | 00:45 |
| blackburn | sonney2k: hm yes my macro is broken | 00:49 |
| blackburn | sonney2k: do you know any math 'construction' when e.g. DNA string is reconstructured with k another? | 01:02 |
| @sonney2k | ? | 01:04 |
| @sonney2k | sorry but I am too sleepy now anyways | 01:04 |
| blackburn | okaay | 01:04 |
| @sonney2k | cu tomorrow :) | 01:04 |
| blackburn | well I have to sleep too | 01:04 |
| @sonney2k | good idea | 01:04 |
| blackburn | will finish it tomorrow's night | 01:04 |
| blackburn | my macro still not working for some reason | 01:05 |
| blackburn | but I believe there is a way to do it with templates :) | 01:05 |
| blackburn | see you | 01:05 |
| @sonney2k | I don't think there is a way w/ templates | 01:05 |
| blackburn | bad if so | 01:05 |
| @sonney2k | try whether SERIALIZABLE_DUMMY(SGMatrix<float64_t>) works | 01:05 |
| @sonney2k | g'night | 01:06 |
| -!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 01:06 | |
| -!- mlsec [~mlsec@bane.ml.tu-berlin.de] has quit [Ping timeout: 240 seconds] | 01:34 | |
| -!- bettyboo [~bettyboo@bane.ml.tu-berlin.de] has quit [Ping timeout: 260 seconds] | 01:35 | |
| -!- f-x [~user@117.192.208.105] has quit [Remote host closed the connection] | 02:09 | |
| CIA-87 | shogun: Soeren Sonnenburg master * r3adb67d / src/shogun/base/Parameter.cpp : use SG_SNOTIMPLEMENTED to fix compile error in non-SGObject derived class - http://bit.ly/n0MYCc | 07:11 |
| CIA-87 | shogun: Soeren Sonnenburg master * r6a23b05 / src/shogun/structure/DynProg.cpp : fix warnign in DynProg about unused variable - http://bit.ly/ru2asy | 07:11 |
| CIA-87 | shogun: Soeren Sonnenburg master * r2042a6d / src/shogun/base/Parameter.cpp : fix remaining warnings in parameter - http://bit.ly/oNndkx | 07:11 |
| -!- in3xes [~in3xes@49.201.227.57] has joined #shogun | 07:29 | |
| -!- shogun-buildbot [~shogun-bu@f053035243.adsl.alicedsl.de] has joined #shogun | 07:32 | |
| @sonney2k | shogun-buildbot, dance | 07:34 |
| shogun-buildbot | <(^.^<) | 07:34 |
| shogun-buildbot | <(^.^)> | 07:34 |
| shogun-buildbot | (>^.^)> | 07:34 |
| shogun-buildbot | (7^.^)7 | 07:34 |
| shogun-buildbot | (>^.^<) | 07:34 |
| @sonney2k | shogun-buildbot, status all | 07:35 |
| shogun-buildbot | runtests: building(compiling) [ETA 1 seconds], (compiling) [ETA 2 seconds] | 07:35 |
| shogun-buildbot | build #18 of runtests is complete: Success [build successful] Build details are at http://fortknox:8010/builders/runtests/builds/18 | 07:36 |
| @sonney2k | shogun-buildbot, last | 07:37 |
| shogun-buildbot | last build [runtests]: 38 seconds ago: build successful | 07:37 |
| in3xes | you have been using it? or just started using? | 07:38 |
| in3xes | btw, I work for buildbot as part of GSoC :) | 07:38 |
| in3xes | shogun-buildbot: dance | 07:39 |
| shogun-buildbot | <(^.^<) | 07:39 |
| shogun-buildbot | <(^.^)> | 07:39 |
| shogun-buildbot | (>^.^)> | 07:39 |
| shogun-buildbot | (7^.^)7 | 07:39 |
| shogun-buildbot | (>^.^<) | 07:39 |
| in3xes | sorry for spamming. I was checking it takes command from everyone | 07:39 |
| -!- shogun-buildbot [~shogun-bu@f053035243.adsl.alicedsl.de] has quit [Ping timeout: 252 seconds] | 07:43 | |
| -!- in3xes [~in3xes@49.201.227.57] has quit [Remote host closed the connection] | 10:11 | |
| -!- in3xes [~in3xes@49.201.227.57] has joined #shogun | 11:20 | |
| @sonney2k | in3xes, I am just setting it up | 12:36 |
| @sonney2k | in3xes, do you have any hints on how to get email notifications to work? | 12:36 |
| @sonney2k | (too bad the build-machine just rebooted) | 12:37 |
| -!- shogun-buildbot [~shogun-bu@f053035243.adsl.alicedsl.de] has joined #shogun | 12:48 | |
| @sonney2k | shogun-buildbot, back to work right? | 12:48 |
| -!- alesis-novik [~alesis@vpn2-170.vpn.net.ed.ac.uk] has quit [Quit: Leaving] | 12:51 | |
| -!- heiko [~heiko@134.91.52.173] has joined #shogun | 12:52 | |
| @sonney2k | shogun-buildbot, dance for heiko please :) | 12:54 |
| shogun-buildbot | <(^.^<) | 12:54 |
| shogun-buildbot | <(^.^)> | 12:54 |
| shogun-buildbot | (>^.^)> | 12:55 |
| shogun-buildbot | (7^.^)7 | 12:55 |
| shogun-buildbot | (>^.^<) | 12:55 |
| @sonney2k | shogun-buildbot, status all | 12:55 |
| shogun-buildbot | runtests: idle, last build 5h18m33s ago: build successful | 12:55 |
| -!- heiko [~heiko@134.91.52.173] has quit [Ping timeout: 258 seconds] | 12:59 | |
| -!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun | 13:09 | |
| CIA-87 | shogun: Soeren Sonnenburg master * rec41ad2 / (3 files in 3 dirs): fix further compiler warnings - http://bit.ly/mYPzMp | 13:13 |
| -!- f-x [~user@117.192.211.222] has joined #shogun | 13:16 | |
| heiko | sonney2k, around? | 13:16 |
| @sonney2k | heiko, I let the buildbot dance for you and you are still asking? | 13:17 |
| heiko | hehe :) what? | 13:17 |
| @sonney2k | shogun-buildbot, dance for heiko again please :) | 13:17 |
| shogun-buildbot | <(^.^<) | 13:17 |
| shogun-buildbot | <(^.^)> | 13:17 |
| shogun-buildbot | (>^.^)> | 13:17 |
| shogun-buildbot | (7^.^)7 | 13:17 |
| shogun-buildbot | (>^.^<) | 13:17 |
| @sonney2k | shogun-buildbot, status all | 13:18 |
| shogun-buildbot | runtests: building(configuring) | 13:18 |
| heiko | lol ;) | 13:18 |
| @sonney2k | shogun-buildbot, status all | 13:18 |
| shogun-buildbot | runtests: building(compiling) | 13:18 |
| heiko | shogun-buidbot, status all | 13:18 |
| heiko | shogun-buildbot, status all | 13:18 |
| shogun-buildbot | runtests: building(compiling) | 13:18 |
| heiko | nice one | 13:18 |
| @sonney2k | it just catched my commit above | 13:18 |
| @sonney2k | polls every few minutes | 13:18 |
| heiko | cool thing | 13:18 |
| @sonney2k | and then rebuilds ... | 13:19 |
| heiko | and what does it do exactly? | 13:19 |
| heiko | also tests? | 13:19 |
| @sonney2k | tests not yet | 13:19 |
| @sonney2k | I mean we don't have them yet so it doesn't make sense | 13:19 |
| heiko | all interfaces? | 13:19 |
| @sonney2k | it compiles for all interfaces | 13:19 |
| heiko | cool, how long does it take? | 13:19 |
| @sonney2k | ~20 min | 13:19 |
| @sonney2k | err all interfaces that are enabled by *default* | 13:19 |
| heiko | ok | 13:19 |
| heiko | is it running on your computer or in university? | 13:20 |
| @sonney2k | I think I will add a build run once per day that does *all* interfaces with all options | 13:20 |
| @sonney2k | and then with as few options as possible | 13:20 |
| @sonney2k | at home curretnly | 13:20 |
| heiko | ok cool | 13:21 |
| heiko | really cool | 13:21 |
| @sonney2k | too bad that I don't have enough memory to run 4 of these buildbots | 13:21 |
| heiko | 4? | 13:21 |
| @sonney2k | and I failed to configure the email notification | 13:21 |
| @sonney2k | heiko, builder daemons | 13:21 |
| @sonney2k | then it could build on e.g. osx, cygwin etc | 13:21 |
| heiko | ah ok | 13:21 |
| heiko | would be nice :) | 13:22 |
| heiko | but hey, its a big step already | 13:22 |
| heiko | sonney2k, about the parameter versions... | 13:22 |
| heiko | http://pastebin.com/C3iwcLnH | 13:22 |
| heiko | have a look at this | 13:22 |
| @sonney2k | heiko, sorry for what? you did it just right | 13:22 |
| heiko | huh? sorry ? what? | 13:23 |
| @sonney2k | hmmhh I guess I need sleep | 13:23 |
| heiko | how long has it been? | 13:24 |
| heiko | sonney2k, if you are too tired now, we can talk later | 13:24 |
| heiko | just a few comments on this pseudo-code | 13:25 |
| heiko | the parameter map needs to be a variable of SGObject, therefore may not be an SGObject itself | 13:25 |
| heiko | and i want to do one migrate method that switches over name/version for migration | 13:25 |
| @sonney2k | uhh | 13:25 |
| heiko | or do you think one for each verison is better? | 13:26 |
| @sonney2k | sry migrate method? | 13:26 |
| @sonney2k | I thought we are doing this in the post hook? | 13:26 |
| heiko | yes to transfer from one version to another | 13:26 |
| heiko | well ok or there | 13:26 |
| heiko | it does not matter | 13:27 |
| @sonney2k | ok | 13:27 |
| heiko | but do you think it is better to just add code there in every new version, or create a new method which is called from there every version? | 13:27 |
| @sonney2k | heiko, I would do the migration only in the *post function (and if that gets too complex via private helper functions) | 13:29 |
| heiko | ok then | 13:29 |
| @sonney2k | I mean in the end a migration is far too general | 13:29 |
| heiko | true, just a name i chose | 13:29 |
| @sonney2k | so it can combine everything with everything | 13:30 |
| heiko | for getting a more concrete conecept down | 13:30 |
| @sonney2k | and it might be more efficient to do the conversion from version 1 -> version 100 directly too (if needed) | 13:30 |
| heiko | ok | 13:30 |
| heiko | oh | 13:30 |
| heiko | ehm | 13:30 |
| heiko | wait | 13:30 |
| @sonney2k | heiko, so all you need to take care of is that the map is loaded and data is in there | 13:31 |
| heiko | the parameter migration is done recursively | 13:31 |
| heiko | well ok, could be done iteratively in the post method | 13:31 |
| @sonney2k | heiko, it really is up to the user ... whatever he wants to do in the post method | 13:32 |
| @sonney2k | not really your concern so far :) | 13:32 |
| heiko | so would you use the save_serializable_post method or create a new one? | 13:34 |
| @sonney2k | heiko, I think it should be safe to use the save_serializable_post method. | 13:36 |
| @sonney2k | Hmmhh, limitations might be that parameters get moved from class a to derived class b or some dependency between base and its subclasses | 13:37 |
| heiko | I think this would work | 13:38 |
| heiko | all parameters are loaded in the beginning and put into some list | 13:38 |
| heiko | the post method may access this list | 13:38 |
| @sonney2k | and if some base class has a private variable? | 13:39 |
| heiko | mmh | 13:40 |
| CIA-87 | shogun: Baozeng Ding master * rbc6cd6c / src/interfaces/csharp_modular/swig_typemaps.i : support stringlist for csharp - http://bit.ly/nqXT8G | 13:41 |
| CIA-87 | shogun: Baozeng Ding master * r0367d91 / (3 files): add some csharp examples - http://bit.ly/rcx1Gk | 13:41 |
| CIA-87 | shogun: Soeren Sonnenburg master * rd6b9d89 / (4 files in 2 dirs): | 13:41 |
| CIA-87 | shogun: Merge pull request #286 from sploving/master | 13:41 |
| CIA-87 | shogun: support stringlist for csharp and add some examples - http://bit.ly/q4XO8G | 13:41 |
| @sonney2k | heiko, in the worst case we can only work with protected varialbes | 13:43 |
| heiko | yes | 13:43 |
| heiko | I dont know, I think perhaps I should get it working the way we diskussed it first. Should work for most cases | 13:44 |
| heiko | then we can think of these cases | 13:44 |
| heiko | Its complicated enough already | 13:44 |
| @sonney2k | of course there might be corner cases where a total refactoring (splitting up a class into several) etc might not work - but hey better 99% working than not at all | 13:44 |
| @sonney2k | agreed | 13:44 |
| shogun-buildbot | build #21 of runtests is complete: Failure [failed compile] Build details are at http://fortknox:8010/builders/runtests/builds/21 blamelist: sploving1@gmail.com | 13:45 |
| heiko | oh cool, buildbot hat a blame list :) | 13:46 |
| @sonney2k | shogun-buildbot, status all | 13:48 |
| shogun-buildbot | runtests: building(compiling) [ETA 7m04s] | 13:48 |
| @sonney2k | heiko, I don't understand how it could have possibly failed there?! | 13:48 |
| heiko | sonney2k, how to see the logs? | 13:48 |
| heiko | the url is an empty page here | 13:49 |
| @sonney2k | heiko, not possible yet | 13:49 |
| @sonney2k | I have to move the build master to some public server | 13:49 |
| heiko | ok | 13:49 |
| @sonney2k | still >2500 warnings... | 13:52 |
| @sonney2k | ohh well | 13:52 |
| @sonney2k | way to go | 13:52 |
| shogun-buildbot | build #22 of runtests is complete: Success [build successful] Build details are at http://fortknox:8010/builders/runtests/builds/22 | 14:05 |
| -!- in3xes [~in3xes@49.201.227.57] has quit [Quit: Leaving] | 14:10 | |
| CIA-87 | shogun: Heiko Strathmann master * rf95d3a0 / (2 files): made get_array const - http://bit.ly/nxbizl | 15:05 |
| CIA-87 | shogun: Soeren Sonnenburg master * r7dcf548 / (6 files in 3 dirs): | 15:05 |
| CIA-87 | shogun: Merge pull request #287 from karlnapf/master | 15:05 |
| CIA-87 | shogun: ParameterMap is now not a CSGObject anymore - http://bit.ly/oElV2Z | 15:05 |
| CIA-87 | shogun: Heiko Strathmann master * r73e5515 / (2 files): parameter map will be a variable of CSGObject, so must not be a CSGObject itself. switched to manual memory handling and removed inheritance. - http://bit.ly/pguLLO | 15:05 |
| CIA-87 | shogun: Heiko Strathmann master * r42f532d / examples/undocumented/libshogun/base_parameter_map.cpp : applied changes of ParameterMap - http://bit.ly/qZPa9p | 15:05 |
| CIA-87 | shogun: Heiko Strathmann master * r4be098c / src/shogun/base/DynArray.h : made get_array const - http://bit.ly/mZMnCN | 15:05 |
| -!- in3xes [~in3xes@180.149.49.230] has joined #shogun | 16:10 | |
| -!- in3xes [~in3xes@180.149.49.230] has quit [Ping timeout: 250 seconds] | 16:15 | |
| -!- in3xes [~in3xes@182.156.163.142] has joined #shogun | 16:32 | |
| heiko | shogun-buildbot status all | 16:42 |
| heiko | shogun-buildbot status | 16:43 |
| heiko | shogun-buildbot status dance | 16:43 |
| heiko | shogun-buildbot, status all | 16:43 |
| shogun-buildbot | runtests: idle, last build 1h12m39s ago: build successful | 16:43 |
| heiko | shogun-buildbot, dance | 16:43 |
| shogun-buildbot | <(^.^<) | 16:43 |
| shogun-buildbot | <(^.^)> | 16:43 |
| shogun-buildbot | (>^.^)> | 16:43 |
| shogun-buildbot | (7^.^)7 | 16:43 |
| shogun-buildbot | (>^.^<) | 16:43 |
| heiko | shogun-buildbot, help | 16:43 |
| shogun-buildbot | Get help on what? (try 'help <foo>', or 'commands' for a command list) | 16:43 |
| heiko | shogun-buildbot, commands | 16:43 |
| shogun-buildbot | buildbot commands: commands, dance, destroy, excited, force, hello, help, last, list, mute, notify, source, status, stop, unmute, version, watch | 16:43 |
| -!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds] | 17:25 | |
| -!- mrsrikanth [~mrsrikant@59.92.36.47] has joined #shogun | 17:39 | |
| -!- shogun-buildbot_ [~shogun-bu@7nn.de] has joined #shogun | 17:50 | |
| -!- shogun-buildbot_ [~shogun-bu@7nn.de] has quit [Client Quit] | 17:51 | |
| -!- shogun-buildbot [~shogun-bu@f053035243.adsl.alicedsl.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 17:51 | |
| -!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 17:51 | |
| -!- mrsrikanth [~mrsrikant@59.92.36.47] has quit [Quit: Leaving] | 18:19 | |
| -!- srikanth [~mrsrikant@59.92.36.47] has joined #shogun | 18:22 | |
| -!- srikanth [~mrsrikant@59.92.36.47] has quit [Read error: Connection reset by peer] | 18:24 | |
| -!- mrsrikanth [~mrsrikant@59.92.36.47] has joined #shogun | 18:25 | |
| -!- blackburn [~blackburn@188.122.238.50] has joined #shogun | 19:12 | |
| blackburn | wow shogun-buildbot here | 19:13 |
| blackburn | shogun-buildbot: dance | 19:13 |
| shogun-buildbot | <(^.^<) | 19:13 |
| shogun-buildbot | <(^.^)> | 19:13 |
| shogun-buildbot | (>^.^)> | 19:13 |
| shogun-buildbot | (7^.^)7 | 19:13 |
| shogun-buildbot | (>^.^<) | 19:13 |
| blackburn | shogun-buildbot: status | 19:27 |
| blackburn | sonney2k: I will get memory usage lower | 19:27 |
| blackburn | soon :) | 19:27 |
| shogun-buildbot | runtests: idle, last build 1h05m13s ago: build successful | 19:27 |
| -!- Netsplit *.net <-> *.split quits: blackburn | 19:38 | |
| -!- Netsplit over, joins: blackburn | 19:38 | |
| -!- in3xes [~in3xes@182.156.163.142] has quit [Ping timeout: 276 seconds] | 19:47 | |
| @sonney2k | blackburn, but http://shogun-toolbox.org:5347/builders/runtests/builds/24/steps/compile/logs/warnings%20(2527)/text | 19:52 |
| blackburn | why but? | 19:52 |
| @sonney2k | >2500 warnings | 19:53 |
| blackburn | I'm sure we will get rid of them | 19:55 |
| f-x | sonney2k: using VW's parsing/tokenizing stuff for reading from ascii files really speeds things up | 19:57 |
| f-x | do you think i should modify our code to use that? | 19:57 |
| @sonney2k | f-x, that would be great - question now is how much time do you want / will you commit after gsoc... | 19:57 |
| f-x | i mean i'm getting 6.2s -> 3.2s after the transformation | 19:57 |
| @sonney2k | I mean you only have 1.5 more weeks until the final deadline | 19:58 |
| f-x | sonney2k: this would hardly take any time, within tonight | 19:58 |
| @sonney2k | ok | 19:58 |
| @sonney2k | then go go go :) | 19:58 |
| f-x | ok :) | 19:58 |
| f-x | but i'll be overloading the float64_t/float32_t readers only for now - is that ok? | 19:58 |
| @sonney2k | f-x, that said - it would be great if you would continue to extend the online learning part of shogun | 19:58 |
| f-x | sonney2k: even i think so | 19:59 |
| @sonney2k | you are now the expert there | 19:59 |
| @sonney2k | f-x, Re float32* ? | 19:59 |
| @sonney2k | I don't understand what you mean | 19:59 |
| f-x | sonney2k: in AsciiFile/StreamingAsciiFile etc | 19:59 |
| @sonney2k | you mean the get/set functions? | 19:59 |
| f-x | we have #defined everything for float32_t/float64_t etc, right? | 19:59 |
| f-x | yeah | 19:59 |
| @sonney2k | ahh sorry I misread what you said above | 20:00 |
| @sonney2k | yes sure | 20:00 |
| f-x | so for now i'll just overload the float32_t/float64_t functions since they're the easiest to do | 20:00 |
| @sonney2k | that is a good start | 20:00 |
| @sonney2k | it is trivial to extend to integer etc though | 20:00 |
| f-x | sonney2k: there's a tokenize() function which will be used - where should i put it? SGIO? | 20:00 |
| f-x | sonney2k: yeah, once we write the proper functions for *_from_substring | 20:01 |
| @sonney2k | f-x, if it is very general yes - otherwise put it in asciifile | 20:01 |
| f-x | sonney2k: very general, i don't know - but it will be used by every get_* if we do this transformation. ok so it goes into AsciiFile for the time being | 20:02 |
| @sonney2k | f-x, I mean if you make it static it can be called from anywhere ... | 20:03 |
| @sonney2k | so no prob | 20:03 |
| f-x | yes correct, i won't meddle with SGIO currently | 20:03 |
| @sonney2k | blackburn, it would be soo cool if you get the thing to compile requiring <500MB :) | 20:06 |
| @sonney2k | I am going to fix a few warnings in the meantime | 20:06 |
| blackburn | I'd say it is not possible with various SGMatrices and other things | 20:06 |
| CIA-87 | shogun: Soeren Sonnenburg master * rb829fcc / src/shogun/base/Parameter.cpp : initialize source and dest with NULL to fix uninited warnings - http://bit.ly/pS2i5P | 20:08 |
| @sonney2k | blackburn, well currently it can require up to 3G | 20:08 |
| @sonney2k | shogun-buildbot, status all | 20:10 |
| shogun-buildbot | runtests: building(compiling) [ETA 21m40s] | 20:10 |
| @sonney2k | kk | 20:10 |
| blackburn | http://www.visual-literacy.org/periodic_table/periodic_table.html | 20:23 |
| @sonney2k | he | 20:27 |
| @sonney2k | f-x, any ideas about classifier/vw/cache/VwNativeCacheReader.cpp:83:36: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing] ? | 20:29 |
| @sonney2k | shouldn't the varialbe be defined size_t from the beginning? | 20:30 |
| f-x | sonney2k: variable p? | 20:30 |
| @sonney2k | f-x, yes p | 20:31 |
| @sonney2k | I mean it is a local variable and ends there | 20:31 |
| @sonney2k | the function ends afterwards I meant | 20:31 |
| @sonney2k | so one can safely do a char* cast in the buf.read* function call | 20:32 |
| f-x | buf.read_file((char*)p, total) ? | 20:32 |
| @sonney2k | yes | 20:33 |
| f-x | sonney2k: that should eliminate the warning | 20:33 |
| CIA-87 | shogun: Soeren Sonnenburg master * r081405b / (4 files): add SG_NOTIMPLEMENTED to fix switch CT_NDARRAY warning - http://bit.ly/rfTGSd | 20:34 |
| CIA-87 | shogun: Soeren Sonnenburg master * r7fb2ee0 / (2 files): fix warnings in classifiers - http://bit.ly/nko2Yi | 20:34 |
| @sonney2k | shogun-buildbot, status all | 20:34 |
| shogun-buildbot | runtests: idle, last build 1m48s ago: build successful | 20:34 |
| @sonney2k | hmmhh why didn't it say so... | 20:34 |
| f-x | sonney2k: aren't you getting the same warning for line 127 too? | 20:35 |
| f-x | or is that valid? | 20:35 |
| -!- mrsrikanth [~mrsrikant@59.92.36.47] has quit [Quit: Leaving] | 20:35 | |
| shogun-buildbot | build #26 of runtests is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org:5347/builders/runtests/builds/26 blamelist: sonne@debian.org | 20:36 |
| @sonney2k | f-x, http://shogun-toolbox.org:5347/builders/runtests/builds/25/steps/compile/logs/warnings%20(2525)/text | 20:36 |
| @sonney2k | doesn't say so | 20:36 |
| @sonney2k | shogun-buildbot, thanks for pointing that out | 20:38 |
| @sonney2k | blackburn, hurray email notifictaions work | 20:40 |
| blackburn | sonney2k: nice | 20:40 |
| @sonney2k | I got an email that things don't compile :) | 20:40 |
| blackburn | I just finished SERIALIZABLE_DUMMY things | 20:41 |
| blackburn | serialization should work soon | 20:42 |
| CIA-87 | shogun: Soeren Sonnenburg master * r50d60be / src/shogun/io/SerializableHdf5File.cpp : use SG_S* function for static function - http://bit.ly/oLKfvP | 20:47 |
| @sonney2k | blackburn, what do you think - should the buildbot do some quick tests first | 20:51 |
| @sonney2k | blackburn, I mean like just compile w/ debug enabled only libshogun | 20:51 |
| @sonney2k | and then static interfaces | 20:51 |
| @sonney2k | and then modular ones one by one? | 20:51 |
| blackburn | yeah seems to be good way | 20:52 |
| @sonney2k | so we need ccache | 20:52 |
| blackburn | hooray! | 20:52 |
| blackburn | got java to compile with my ignores | 20:53 |
| @sonney2k | yay! | 20:53 |
| @sonney2k | great | 20:53 |
| @sonney2k | you should test csharp, lua,ruby too though | 20:53 |
| @sonney2k | if these compile & examples still run (mostly) we are good | 20:53 |
| blackburn | can we use buildbot for this task? | 20:54 |
| blackburn | I'll check python now | 20:55 |
| @sonney2k | blackburn, not yet | 20:55 |
| blackburn | sonney2k: is sparse bug already fixed? | 20:56 |
| @sonney2k | blackburn, no | 20:56 |
| blackburn | uhoh | 20:56 |
| @sonney2k | I am still working on buildbot... | 20:56 |
| @sonney2k | blackburn, that is impressive http://build.chromium.org/p/chromium/console | 20:59 |
| blackburn | hard to comphrehend | 21:00 |
| @sonney2k | we have this too btw | 21:00 |
| @sonney2k | but not yet as beautiful | 21:00 |
| shogun-buildbot | build #28 of runtests is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/runtests/builds/28 | 21:12 |
| blackburn | wow | 21:13 |
| blackburn | sonney2k: about Q I asked you yesterday | 21:29 |
| blackburn | is there any method/etc to reconstruct *string* by other *strings* using weights? | 21:30 |
| blackburn | i.e. string A is 0.2 of B and 0.7 of C and 0.1 of D | 21:30 |
| blackburn | not really 'reconstruct' but get these weights | 21:31 |
| @sonney2k | sorry but I don't understand what you mean | 21:34 |
| blackburn | sonney2k: I'm trying to find analogue for euclidean barycentric coordinates | 21:34 |
| blackburn | it is used in LLE | 21:34 |
| @sonney2k | you mean you want a vectorial representation for strings? | 21:35 |
| blackburn | not really | 21:35 |
| blackburn | I just want weights | 21:35 |
| @sonney2k | weights for what? | 21:35 |
| blackburn | for each string I want to compute some weights by other strings | 21:35 |
| blackburn | essentially for DNA seqs | 21:36 |
| @sonney2k | ?!?!?!?!? | 21:36 |
| blackburn | uh oh sorry I'm stupid :D | 21:36 |
| blackburn | in LLE every point is reconstructured with linear combination of its neighbors | 21:37 |
| blackburn | and I'm trying to get analogue for strings | 21:37 |
| blackburn | I don't really want linear combination of strings (non-sense) | 21:37 |
| blackburn | but weights | 21:37 |
| @sonney2k | blackburn, linear combination of strings is sth I can give you | 21:38 |
| @sonney2k | I mean when each string is embedded into some feature space | 21:38 |
| @sonney2k | then you can do this | 21:38 |
| blackburn | in euclidean space? | 21:38 |
| @sonney2k | so what you are looking for is either a vectorial representation of strings or string kernels | 21:38 |
| @sonney2k | yes | 21:38 |
| blackburn | no I don't want to find embedding because it could be a part of embedding algorithm | 21:39 |
| blackburn | sonney2k: http://www.cs.nyu.edu/~roweis/lle/images/llef2med.gif that's how it is done for already euclidean space | 21:40 |
| blackburn | and the idea is to find an analogue for this 'reconstruction' | 21:40 |
| @sonney2k | yeah exactly | 21:42 |
| blackburn | exactly what? | 21:42 |
| @sonney2k | so you want an embedding of strings | 21:42 |
| blackburn | yes | 21:42 |
| @sonney2k | and you want to tune that embedding using your algorithm later | 21:42 |
| blackburn | I had an idea to modify LLE | 21:42 |
| @sonney2k | so take the embedding you can get and good | 21:42 |
| blackburn | all the things should stay but this reconstruction | 21:43 |
| blackburn | have you any ideas how to compute these weights for strings? | 21:43 |
| blackburn | sure it could be done for data already embedded into euclidean space | 21:43 |
| blackburn | but if string-space is input or so.. | 21:44 |
| @sonney2k | blackburn, you only compute dot-products between example_i and j right? | 21:52 |
| blackburn | no, no dot products | 21:52 |
| blackburn | in LLE exact coordinates are required | 21:52 |
| @sonney2k | blackburn, btw kernelpca... | 21:52 |
| -!- in3xes [~in3xes@182.156.166.217] has joined #shogun | 21:52 | |
| @sonney2k | blackburn, yeah but then you need to embed the string | 21:52 |
| blackburn | sonney2k: yes, I remember | 21:52 |
| @sonney2k | otherwise you won't get 'weights' | 21:52 |
| blackburn | that's the problem | 21:53 |
| @sonney2k | so e.g. use DotFeatures | 21:53 |
| blackburn | how to get weights without embedding | 21:53 |
| @sonney2k | impossible | 21:53 |
| blackburn | not programming issue, math | 21:53 |
| @sonney2k | yeah but e.g. wdfeautres do such transofrmation: string -> Phi(str) | 21:54 |
| @sonney2k | some with weightedspectrum | 21:54 |
| blackburn | string input | 21:54 |
| blackburn | and vector output? | 21:55 |
| @sonney2k | yes | 21:59 |
| blackburn | sonney2k: what is the lengths of output vectors? | 22:00 |
| @sonney2k | 10**20 | 22:01 |
| @sonney2k | but sparse | 22:01 |
| @sonney2k | so say few hundred nonzero elements | 22:01 |
| @sonney2k | blackburn, I have to continue work on buildbot | 22:01 |
| blackburn | uh | 22:01 |
| blackburn | okay sorry | 22:01 |
| f-x | sonney2k: when it's possible i'd like to know your thoughts on the StreamingVwFeatures patch | 22:59 |
| blackburn | sonney2k: got | 23:30 |
| blackburn | [WARN] `/tmp/shogun.tmp' has file-version `(unkown)', which is not supported! | 23:30 |
| blackburn | [WARN] `/tmp/shogun.tmp' has file-version `(unkown)', which is not supported! | 23:30 |
| blackburn | [WARN] `/tmp/shogun.tmp' has file-version `(unkown)', which is not supported! | 23:30 |
| blackburn | [WARN] `/tmp/shogun.tmp' has file-version `(unkown)', which is not supported! | 23:30 |
| blackburn | and nothing serialized | 23:30 |
| -!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 23:40 | |
| --- Log closed Fri Aug 12 00:00:07 2011 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!
