| --- Log opened Sun Aug 18 00:00:20 2013 | ||
| -!- Heiko [5accf27c@gateway/web/freenode/ip.90.204.242.124] has joined #shogun | 00:24 | |
| Heiko | lisitsyn: hey! | 00:25 |
|---|---|---|
| lisitsyn | Heiko: hey again :) | 00:25 |
| Heiko | how is it going? | 00:25 |
| lisitsyn | still ok ;) | 00:26 |
| lisitsyn | Heiko: I was kind of researching 'parameters' thing | 00:26 |
| Heiko | you mean how shogun parameters? | 00:26 |
| lisitsyn | I'd say I have been doing that for a while though | 00:26 |
| lisitsyn | not really shogun but paraemeters in general | 00:26 |
| Heiko | ok you mean how to register etc? | 00:27 |
| lisitsyn | Heiko: what I came with is https://github.com/lisitsyn/stichwort | 00:27 |
| Heiko | getter setter | 00:27 |
| lisitsyn | yes more in this direction | 00:27 |
| Heiko | sweet stichwort :) | 00:27 |
| lisitsyn | Heiko: I broke my mind with casting idea - wanted to make it support implicit casts | 00:28 |
| lisitsyn | like you assign keyword to int and get float | 00:28 |
| Heiko | ah sweet that looks nice! | 00:28 |
| Heiko | are there some more examples? | 00:28 |
| lisitsyn | Heiko: not yet | 00:28 |
| lisitsyn | well I added some unit-tests to be sure it works | 00:28 |
| Heiko | ravis fails | 00:29 |
| lisitsyn | Heiko: the workflow with that thing is | 00:29 |
| lisitsyn | I know | 00:29 |
| lisitsyn | I have C++11 issues :D | 00:29 |
| lisitsyn | so you define keyword | 00:29 |
| lisitsyn | Keyword<double> width("width", 10.0); | 00:29 |
| lisitsyn | then you define function | 00:30 |
| hushell | wiking: Does 'sudo make install' need to be executed before ctest? | 00:30 |
| lisitsyn | hmm say | 00:30 |
| Heiko | lisitsyn: so this uses maps internally? | 00:30 |
| lisitsyn | bool train(Parameters ps); | 00:30 |
| lisitsyn | you obtain the provided value with ps[width] | 00:30 |
| lisitsyn | and call the function with | 00:30 |
| lisitsyn | train({width=3.0}) | 00:30 |
| lisitsyn | Heiko: well it uses more C++ than I ever knew about :D | 00:31 |
| Heiko | looks quite nice, why is not not yet in shogun? | 00:31 |
| lisitsyn | Heiko: I guess it is crazy slow | 00:31 |
| lisitsyn | Heiko: the main challenge with it is I try to keep the Parameter class non-templated | 00:32 |
| lisitsyn | Parameter = Keyword<T> + T | 00:32 |
| lisitsyn | i.e. width = 3.0 is the instance of Parameter | 00:33 |
| lisitsyn | that uses some internal shit underneath | 00:33 |
| lisitsyn | :D | 00:33 |
| Heiko | sweet | 00:33 |
| Heiko | speed is a concern though | 00:34 |
| Heiko | setters should be fast | 00:34 |
| lisitsyn | no that doesn't solve setters problem | 00:34 |
| lisitsyn | as it has nothing to do with swig | 00:34 |
| Heiko | ah ok | 00:35 |
| Heiko | cool stuff though | 00:35 |
| Heiko | c++11 is really cool | 00:35 |
| lisitsyn | Heiko: actually it worked with older C++ | 00:36 |
| Heiko | finally one gets the syntactic stuff that one wants and that is possible to offer | 00:36 |
| lisitsyn | the only difference was | 00:36 |
| lisitsyn | train(kwargs[width=3.0]) | 00:36 |
| lisitsyn | not a big deal | 00:36 |
| lisitsyn | Heiko: {} is possible due to std::initializer_list | 00:36 |
| lisitsyn | Heiko: you may find std::optional<T> convenient too btw | 00:37 |
| lisitsyn | but it is C++14 :D | 00:37 |
| lisitsyn | lol | 00:37 |
| lisitsyn | I pushed DIY version of it to my code though | 00:37 |
| Heiko | lisitsyn: cool! | 00:38 |
| lisitsyn | Heiko: std::optional is used to represent a value that might not be present | 00:38 |
| lisitsyn | so you don't have to deal with pointers or whatever | 00:39 |
| Heiko | lisitsyn: wow I wonder how you cam eup with all this cool stuff | 00:39 |
| Heiko | very nice | 00:39 |
| lisitsyn | Heiko: what exactly? | 00:39 |
| Heiko | so how can we get rid of pointers in shogun? | 00:39 |
| Heiko | :D | 00:40 |
| lisitsyn | Heiko: no pointers are inside of shogun due to very OOP design | 00:40 |
| Heiko | ? | 00:41 |
| lisitsyn | a lot of related objects | 00:41 |
| lisitsyn | Heiko: what's the deal with pointers btw? :) | 00:41 |
| lisitsyn | do you mean raw pointers we have? | 00:41 |
| Heiko | yeah | 00:42 |
| lisitsyn | shared_ptr :) | 00:42 |
| -!- iglesiasg [~iglesias@89-77-4-38.dynamic.chello.pl] has joined #shogun | 00:44 | |
| -!- mode/#shogun [+o iglesiasg] by ChanServ | 00:44 | |
| -!- pickle27 [~kevin@d67-193-243-174.home3.cgocable.net] has left #shogun [] | 00:44 | |
| lisitsyn | Heiko: we have to find a solution for setters/getters | 00:45 |
| Heiko | lisitsyn: would be nice | 00:45 |
| lisitsyn | Heiko: I hate them! :D | 00:46 |
| Heiko | lisitsyn: btw I would still drop the grid-search things | 00:46 |
| Heiko | yeah would be so nice | 00:46 |
| Heiko | in particular for modular bindings | 00:46 |
| lisitsyn | Heiko: each time I find a way it breaks with something else | 00:46 |
| Heiko | although I guess proper examples which illustrate real problems would cause us to write all getters we need ) | 00:46 |
| Heiko | lisitsyn: I am counting on you! | 00:47 |
| lisitsyn | Heiko: in python we might have svm.parameter(width=3) | 00:47 |
| lisitsyn | as keywords can be converted to strings | 00:47 |
| Heiko | should have something for all languages | 00:47 |
| lisitsyn | and they are not to be defined anywhere | 00:47 |
| lisitsyn | but this fails for java | 00:47 |
| Heiko | I think its not a good idea to do things specific to single languages | 00:47 |
| lisitsyn | most of things fail for java | 00:47 |
| lisitsyn | yes totally bad idea | 00:48 |
| Heiko | what about using strings? | 00:48 |
| Heiko | is this too slow? | 00:48 |
| lisitsyn | that's ok | 00:48 |
| Heiko | say we have a c backend for it | 00:48 |
| lisitsyn | well a bit slower | 00:48 |
| lisitsyn | we have other problem here | 00:48 |
| Heiko | question is does it matter if its slow? | 00:48 |
| Heiko | which is? | 00:48 |
| lisitsyn | how to get the type :D | 00:48 |
| lisitsyn | just think about the declaration of such a function | 00:49 |
| lisitsyn | ? parameter(const std::string& name); | 00:49 |
| Heiko | we have a parameter type | 00:50 |
| Heiko | TSGDatetype | 00:50 |
| lisitsyn | okay lets have | 00:50 |
| Heiko | its possible to match things maybe | 00:50 |
| lisitsyn | Parameter parameter(const std::string& name) | 00:50 |
| lisitsyn | the problem is I don't see a way to go from | 00:50 |
| lisitsyn | Parameter | 00:50 |
| lisitsyn | to say double | 00:51 |
| lisitsyn | getDouble(); | 00:51 |
| lisitsyn | that's the only way I see | 00:51 |
| lisitsyn | Heiko: that's essentially the same problem I met in that stichwort | 00:52 |
| Heiko | yeah true, I ea it could be solved with runtime errors | 00:52 |
| Heiko | user can call the method which exist for everything | 00:53 |
| Heiko | and then if types dont match there is an error | 00:53 |
| lisitsyn | I was stucked with no possible virtual template methods | 00:53 |
| lisitsyn | :D | 00:53 |
| lisitsyn | Heiko: I don't mind such way to do that | 00:54 |
| Heiko | lisitsyn: ah I dont know | 00:54 |
| lisitsyn | no I mean runtime errors | 00:55 |
| lisitsyn | I am ok with them | 00:55 |
| Heiko | lisitsyn: did you see my suggestion for the obtain from generic thing? | 00:55 |
| Heiko | lisitsyn: almost the only way (runtime errors) | 00:55 |
| lisitsyn | auto generate things? | 00:55 |
| Heiko | lisitsyn: yeah | 00:55 |
| lisitsyn | yes | 00:55 |
| lisitsyn | that's the same wall | 00:55 |
| lisitsyn | runtime vs compile-time | 00:55 |
| lisitsyn | Heiko: swig obsoletes CRTP so no static polymorphism for us :( | 00:56 |
| Heiko | lisitsyn: but the idea should work, no? | 00:57 |
| lisitsyn | yes sure | 00:58 |
| lisitsyn | Heiko: what you suggest is totally ok and should work | 00:58 |
| Heiko | lisitsyn: so should we do it? | 00:58 |
| Heiko | would at least solve all the mailing list questions ;0 | 00:59 |
| Heiko | \;) | 00:59 |
| lisitsyn | yes as we have no other solution | 00:59 |
| Heiko | lisitsyn: ok then, will push it a bit more this week | 01:00 |
| lisitsyn | Heiko: the only other solution I can think of | 01:01 |
| lisitsyn | is to preprocess headers somehow | 01:01 |
| Heiko | lisitsyn: yes I also thought about that | 01:01 |
| Heiko | but its tricky | 01:02 |
| Heiko | its a bit annoying that c doesnt allow to modify classes in a different file | 01:02 |
| Heiko | but I guess there are reasons for that | 01:02 |
| lisitsyn | shogun_class(SVM) | 01:02 |
| lisitsyn | { | 01:02 |
| lisitsyn | } | 01:02 |
| lisitsyn | Heiko: I think we could do that like that ^ | 01:02 |
| @iglesiasg | oh you guys are on! | 01:03 |
| @iglesiasg | :) | 01:03 |
| lisitsyn | the definition of such a macro is | 01:03 |
| lisitsyn | iglesiasg: yes, hey | 01:03 |
| Heiko | lisitsyn: but then we have another layer | 01:03 |
| Heiko | and everyone has to use it | 01:03 |
| Heiko | while obtain from generic is only for a few | 01:04 |
| Heiko | so I rather prefer a solution with global functions | 01:04 |
| Heiko | iglesiasg: hi! | 01:04 |
| lisitsyn | Heiko: yeah but later we can add more functions | 01:04 |
| lisitsyn | Heiko: I guess what he have is just a typical stuff in every big project :) | 01:05 |
| lisitsyn | a lot of classes and you have to sort the shit out | 01:05 |
| -!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 01:06 | |
| Heiko | lisitsyn: I guess | 01:08 |
| Heiko | lisitsyn: ok sleep | 01:08 |
| Heiko | see you Monday probably when I have my computer back :) | 01:08 |
| @iglesiasg | Heiko, you still on holidays? | 01:08 |
| lisitsyn | alright | 01:09 |
| Heiko | iglesiasg: well its weekend and my machine is at work | 01:09 |
| Heiko | btw I think we should meet in person this year again | 01:10 |
| Heiko | but ets discuss that next dev meeting | 01:10 |
| Heiko | see you, have a good night | 01:10 |
| -!- Heiko [5accf27c@gateway/web/freenode/ip.90.204.242.124] has left #shogun [] | 01:10 | |
| @iglesiasg | Heiko, ah! Just out of curiosity if you were still out somewhere holidays :) | 01:10 |
| @iglesiasg | good night | 01:10 |
| -!- iglesiasg [~iglesias@89-77-4-38.dynamic.chello.pl] has quit [Quit: Ex-Chat] | 02:15 | |
| shogun-buildbot | build #438 of nightly_none is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/438 | 03:44 |
| -!- gsomix [~gsomix@109.169.225.75] has quit [Ping timeout: 248 seconds] | 08:00 | |
| -!- lambday [67157f37@gateway/web/freenode/ip.103.21.127.55] has joined #shogun | 08:49 | |
| -!- gsomix [~gsomix@88.200.244.212] has joined #shogun | 11:05 | |
| -!- lambday [67157f37@gateway/web/freenode/ip.103.21.127.55] has quit [] | 13:27 | |
| -!- gsomix [~gsomix@88.200.244.212] has quit [Ping timeout: 256 seconds] | 14:45 | |
| -!- gsomix [~gsomix@88.200.244.212] has joined #shogun | 15:04 | |
| -!- gsomix_ [~gsomix@109.169.248.101] has joined #shogun | 15:58 | |
| -!- gsomix [~gsomix@88.200.244.212] has quit [Ping timeout: 256 seconds] | 16:01 | |
| -!- iglesiasg [~iglesias@89-77-4-38.dynamic.chello.pl] has joined #shogun | 17:20 | |
| -!- mode/#shogun [+o iglesiasg] by ChanServ | 17:20 | |
| -!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 17:22 | |
| shogun-notifier- | shogun: Roman Votyakov :develop * d60b0b1 / src/shogun/ (10 files): https://github.com/shogun-toolbox/shogun/commit/d60b0b1c364ef0eaacc07082c2c165f1701283a2 | 17:22 |
| shogun-notifier- | shogun: fix some formulas | 17:22 |
| shogun-notifier- | shogun: Fernando Iglesias :develop * 901c758 / src/shogun/ (10 files): https://github.com/shogun-toolbox/shogun/commit/901c7581e8b44177648bb1284685526d2c0f1ac5 | 17:22 |
| shogun-notifier- | shogun: Merge pull request #1436 from votjakovr/feature/gp_refactoring | 17:22 |
| shogun-notifier- | shogun: | 17:22 |
| shogun-notifier- | shogun: Fix some formulas | 17:22 |
| -!- pickle27 [~kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 17:24 | |
| shogun-buildbot | build #1431 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1431 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 17:29 |
| shogun-buildbot | build #925 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/925 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 17:29 |
| shogun-buildbot | build #1462 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1462 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 17:33 |
| shogun-buildbot | build #1550 of deb3 - modular_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1550 blamelist: Roman Votyakov <votjakovr@gmail.com> | 18:13 |
| shogun-buildbot | build #1432 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1432 blamelist: Roman Votyakov <votjakovr@gmail.com> | 18:13 |
| shogun-buildbot | build #926 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/926 blamelist: Roman Votyakov <votjakovr@gmail.com> | 18:13 |
| shogun-buildbot | build #1463 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1463 blamelist: Roman Votyakov <votjakovr@gmail.com> | 18:16 |
| @iglesiasg | pickle27, hey! | 18:42 |
| @iglesiasg | pickle27, travis failed again in the same line https://travis-ci.org/shogun-toolbox/shogun/jobs/10336090#L7046 | 18:42 |
| @iglesiasg | same test, sorry | 18:43 |
| -!- gsomix__ [~gsomix@88.200.247.36] has joined #shogun | 18:44 | |
| -!- gsomix_ [~gsomix@109.169.248.101] has quit [Ping timeout: 248 seconds] | 18:47 | |
| @sonney2k | gsomix__, ping? | 18:49 |
| @iglesiasg | sonney2k, hi! how are you doing? | 18:50 |
| @iglesiasg | sonney2k, do you think we could update a few things in the buildbot to solve the crashes ^? | 18:50 |
| @iglesiasg | one of them seems to be caused by cmake's version | 18:50 |
| gsomix__ | sonney2k, pong. | 18:51 |
| @sonney2k | iglesiasg, I am not root on rpm1 | 18:55 |
| @iglesiasg | oh I see | 18:57 |
| -!- iglesiasg [~iglesias@89-77-4-38.dynamic.chello.pl] has quit [Quit: Ex-Chat] | 19:24 | |
| shogun-notifier- | shogun: lambday :develop * 91e1a00 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/91e1a00db4db6d989c8c8cbed5321feff9e6cfb9 | 19:59 |
| shogun-notifier- | shogun: Rational approximation CG-M job unit-test modified (log-det) | 19:59 |
| shogun-notifier- | shogun: lambday :develop * 7ea907b / / (4 files): https://github.com/shogun-toolbox/shogun/commit/7ea907b75359f55f43eaa0afd0e33b2280641671 | 19:59 |
| shogun-notifier- | shogun: Log Rational approximation CG-M operator function added | 19:59 |
| shogun-notifier- | shogun: Heiko Strathmann :develop * 1fb1a95 / / (7 files): https://github.com/shogun-toolbox/shogun/commit/1fb1a956e0803417dd53cc4e11d31829f31155a7 | 19:59 |
| shogun-notifier- | shogun: Merge pull request #1437 from lambday/feature/log_determinant | 19:59 |
| shogun-notifier- | shogun: | 19:59 |
| shogun-notifier- | shogun: Log Rational approximation CG-M operator function added | 19:59 |
| shogun-buildbot | build #1433 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1433 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com> | 20:05 |
| shogun-buildbot | build #927 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/927 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com> | 20:05 |
| shogun-buildbot | build #1464 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1464 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com> | 20:08 |
| shogun-notifier- | shogun: Soeren Sonnenburg :develop * f3d7110 / src/shogun/distributions/classical/ (2 files): https://github.com/shogun-toolbox/shogun/commit/f3d7110f848c848e7c47baf57aeed4a85722d222 | 20:25 |
| shogun-notifier- | shogun: fix warning | 20:25 |
| shogun-buildbot | build #1551 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1551 | 20:40 |
| shogun-buildbot | build #928 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/928 blamelist: Soeren Sonnenburg <sonne@debian.org> | 20:44 |
| shogun-buildbot | build #1434 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1434 blamelist: Soeren Sonnenburg <sonne@debian.org> | 20:44 |
| shogun-buildbot | build #1435 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1435 blamelist: lambday <heavensdevil6909@gmail.com> | 20:45 |
| shogun-buildbot | build #929 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/929 blamelist: lambday <heavensdevil6909@gmail.com> | 20:45 |
| shogun-buildbot | build #1553 of deb3 - modular_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1553 blamelist: lambday <heavensdevil6909@gmail.com> | 20:46 |
| shogun-buildbot | build #1465 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1465 blamelist: Soeren Sonnenburg <sonne@debian.org> | 20:48 |
| shogun-buildbot | build #1466 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1466 blamelist: lambday <heavensdevil6909@gmail.com> | 20:52 |
| -!- lambday [67157e37@gateway/web/freenode/ip.103.21.126.55] has joined #shogun | 20:53 | |
| lambday | Heiko not here :( | 20:53 |
| -!- gsomix__ is now known as gsomix | 20:55 | |
| -!- lambday [67157e37@gateway/web/freenode/ip.103.21.126.55] has quit [Client Quit] | 20:57 | |
| -!- pickle27 [~kevin@d67-193-243-174.home3.cgocable.net] has left #shogun [] | 21:25 | |
| -!- iglesiasg [~iglesias@89-77-4-38.dynamic.chello.pl] has joined #shogun | 23:02 | |
| -!- mode/#shogun [+o iglesiasg] by ChanServ | 23:02 | |
| lisitsyn | iglesiasg: you must be in poland due to your hostname ;) | 23:06 |
| @iglesiasg | lisitsyn, you got me :) | 23:07 |
| lisitsyn | iglesiasg: how do you like it? | 23:07 |
| @iglesiasg | lisitsyn, it is nice | 23:07 |
| -!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 23:25 | |
| --- Log closed Mon Aug 19 00:00:21 2013 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!