--- Log opened Sat Feb 27 00:00:14 2016 | ||
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 250 seconds] | 01:24 | |
-!- aditya123 [uid148851@gateway/web/irccloud.com/x-myfufpprzirjrmaz] has joined #shogun | 05:31 | |
aditya123 | Hello all, Is Shogun going for GSoc this year ? | 05:32 |
---|---|---|
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 10:21 | |
-!- mode/#shogun [+o besser82] by ChanServ | 10:21 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 11:30 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:30 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 11:41 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 12:38 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:38 | |
@HeikoS | lisitsyn: lets chat here | 12:40 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 12:40 | |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * b4a3f17 / / (7 files): https://github.com/shogun-toolbox/shogun/commit/b4a3f1761235e5ce75d4b300fdcc60de500afff7 | 12:40 |
shogun-notifier- | shogun: hacky draft of added cpp to meta examples and the test build. Only works when make is called single-threaded | 12:40 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 08c1ab3 / examples/meta/cpp/CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/08c1ab354c9216c61e69aed6cef9be2bb9ba51ca | 12:40 |
shogun-notifier- | shogun: update comment | 12:40 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * c84cec8 / examples/meta/generator/targets/cpp.json: https://github.com/shogun-toolbox/shogun/commit/c84cec8db0d8d51baab71cd68f61138f5b192a4e | 12:40 |
shogun-notifier- | shogun: remove unnecessary whitespaces | 12:40 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * c9893bc / examples/meta/generator/targets/cpp.json: https://github.com/shogun-toolbox/shogun/commit/c9893bc10dea7e89b6731385fc2cef5bd1bb3c65 | 12:40 |
shogun-notifier- | shogun: make use of some for automatic reference counting | 12:40 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 9877631 / / (7 files): https://github.com/shogun-toolbox/shogun/commit/98776313ab055473123fe876aecb7161d7e11429 | 12:40 |
shogun-notifier- | shogun: Merge pull request #3014 from karlnapf/feature/sphinxdoc_cpp | 12:40 |
shogun-notifier- | shogun: | 12:40 |
shogun-notifier- | shogun: hacky draft of added cpp to meta examples and the test build. | 12:40 |
lisitsyn | ok | 12:40 |
lisitsyn | oops | 12:40 |
lisitsyn | ok | 12:40 |
lisitsyn | ok ok | 12:40 |
lisitsyn | :) | 12:40 |
lisitsyn | let me | 12:40 |
@HeikoS | haha | 12:40 |
@HeikoS | sorry understood you wrong | 12:40 |
lisitsyn | no worries | 12:40 |
lisitsyn | actually my changes were isolated | 12:41 |
lisitsyn | in this cmakelists | 12:41 |
lisitsyn | should be easy to do | 12:41 |
lisitsyn | HeikoS: meanwhile | 12:41 |
lisitsyn | GPs + multivariate output | 12:42 |
lisitsyn | possible? | 12:42 |
@HeikoS | yes | 12:43 |
@HeikoS | but not in sg | 12:43 |
@HeikoS | lisitsyn: but yorkerlin will do this | 12:43 |
@HeikoS | in gsoc | 12:43 |
@HeikoS | hopefully | 12:43 |
@HeikoS | lisitsyn: in fact, send him an email to ask that. He might do it for simple Gp straight away :) | 12:43 |
@HeikoS | lisitsyn: where are shoguns c++ types degfined? | 12:43 |
lisitsyn | HeikoS: what do you mean? | 12:44 |
lisitsyn | which ones? | 12:44 |
@HeikoS | float64_t | 12:44 |
lisitsyn | int_64 and stuff? | 12:44 |
lisitsyn | ah | 12:44 |
lisitsyn | I dont remember sorry, grep it ;) | 12:44 |
@HeikoS | common.h | 12:45 |
lisitsyn | HeikoS: ok testing it again | 12:46 |
lisitsyn | will push if works | 12:46 |
lisitsyn | HeikoS: bad thing | 12:48 |
lisitsyn | as we are using executable | 12:48 |
lisitsyn | we will have millions of targets | 12:48 |
lisitsyn | in cmake | 12:48 |
lisitsyn | [ 98%] Built target classifier-knn-meta-cpp | 12:48 |
lisitsyn | like that | 12:48 |
lisitsyn | here you go | 12:50 |
shogun-notifier- | shogun: Sergey Lisitsyn :feature/sphinxdoc * 4d87a00 / examples/meta/cpp/CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/4d87a0032d7eb33d7b0e9c6dda800a9f5ad24152 | 12:50 |
shogun-notifier- | shogun: Better cmake scripting of cpp meta examples | 12:50 |
lisitsyn | try that | 12:51 |
@HeikoS | lisitsyn: trying | 12:53 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 47f5983 / examples/meta/generator/targets/cpp.json: https://github.com/shogun-toolbox/shogun/commit/47f5983f36bcbcc53f800913230e6173ce1f5a1b | 12:53 |
shogun-notifier- | shogun: add all the vector and matrix types | 12:53 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * b2b6b8f / examples/meta/src/classifier/knn.sg: https://github.com/shogun-toolbox/shogun/commit/b2b6b8f5b97875843cec6692c22db586def996b6 | 12:53 |
shogun-notifier- | shogun: back to RealVector | 12:54 |
@HeikoS | lisitsyn: btw I will push the lua thing | 12:54 |
@HeikoS | and you can have a look | 12:54 |
@HeikoS | doesnt work somehow | 12:54 |
lisitsyn | HeikoS: what doesn't work? | 12:54 |
@HeikoS | https://github.com/shogun-toolbox/shogun/pull/3011 | 12:54 |
lisitsyn | ok | 12:55 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc_lua * 15e0010 / / (6 files): https://github.com/shogun-toolbox/shogun/commit/15e0010c410f15849e884dc59fefad5efdae01b4 | 12:55 |
shogun-notifier- | shogun: add lua to cookbook, meta examples, and test build. Not yet working. | 12:55 |
lisitsyn | never tried lua | 12:55 |
lisitsyn | :D | 12:55 |
lisitsyn | with shogun | 12:56 |
@HeikoS | lisitsyn: same | 12:56 |
@HeikoS | but there are some examples | 12:56 |
@HeikoS | that work | 12:56 |
@HeikoS | but meta example somehow doesnt work | 12:56 |
@HeikoS | need to compare and adjust | 12:56 |
@HeikoS | but I have no idea | 12:56 |
lisitsyn | ok I will check | 12:56 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * ffdda55 / examples/meta/generator/targets/cpp.json: https://github.com/shogun-toolbox/shogun/commit/ffdda55c40b30a76c2df445d1e46595df6b0df3d | 12:58 |
shogun-notifier- | shogun: fix copy paste error | 12:58 |
@HeikoS | lisitsyn: hey btw I have another thing I need feedback on | 12:59 |
@HeikoS | lisitsyn: or more discuss | 12:59 |
@HeikoS | so you know we disabled integration testing | 12:59 |
lisitsyn | what is this? | 12:59 |
lisitsyn | yeah | 12:59 |
@HeikoS | which is good as it was unmaintainable | 12:59 |
@HeikoS | with name changes in classes causing false positives | 13:00 |
@HeikoS | but the idea itself, making sure results do not change compared to reference, is good right? | 13:00 |
@HeikoS | so what about rather than serialising the instances, we compare vector/matrix valued output only | 13:00 |
@HeikoS | to reference | 13:00 |
@HeikoS | we could add this to meta language | 13:00 |
@HeikoS | and then compare example output across all interfaces to reference result | 13:01 |
lisitsyn | yes | 13:01 |
lisitsyn | this makes sense to me | 13:01 |
@HeikoS | and here modifying the class itself does not change the result, as one would only add result from algorithm | 13:01 |
@HeikoS | lisitsyn: so then we avoid that things slip through if not covered by unit tests | 13:01 |
lisitsyn | we just need to indicate somehow | 13:01 |
lisitsyn | what to check | 13:01 |
@HeikoS | lisitsyn: I would add a SGClass that one can pass vector/matrix | 13:01 |
@HeikoS | and it somehow writes it to file | 13:01 |
@HeikoS | just as before we pickled things | 13:02 |
@HeikoS | and then have a test script that compares these outputs | 13:02 |
@HeikoS | and meta language examples will just have to put the "last" vector with the result in there | 13:02 |
@HeikoS | lisitsyn: btw your cpp cmake tuning worked | 13:02 |
lisitsyn | I think it shouldn't be the last vector | 13:02 |
lisitsyn | there should be some lang construction | 13:03 |
lisitsyn | like 'checkpoint' | 13:03 |
lisitsyn | that is checked | 13:03 |
lisitsyn | or we can compare just any vector/matrix we get | 13:04 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 6525dc7 / examples/meta/cpp/CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/6525dc7a86c1ae96fcde1e055f767e12059cb903 | 13:04 |
shogun-notifier- | shogun: add comment to explain sergeys trick | 13:04 |
@HeikoS | lisitsyn: can you explain? | 13:04 |
lisitsyn | I think we could add a flag to generate.py | 13:04 |
@HeikoS | lisitsyn: I think it should be concious decision what is compared | 13:04 |
@HeikoS | should not compare everything | 13:05 |
@HeikoS | or dont you agree? | 13:05 |
lisitsyn | why not everything? | 13:05 |
@HeikoS | lisitsyn: you mean like all vectors and matrices that have been stored in a variable? | 13:05 |
lisitsyn | yes | 13:05 |
@HeikoS | lisitsyn: ok | 13:05 |
@HeikoS | but then | 13:05 |
lisitsyn | like generate.py --test-across-langs | 13:05 |
@HeikoS | it should not eben be in the .sg file | 13:05 |
@HeikoS | but just in the generator | 13:05 |
lisitsyn | when enabled the generate.py just takes any value it sees | 13:06 |
@HeikoS | lisitsyn: good idea, then we dont have to touch the .sg file if we change the mechanism in there | 13:06 |
lisitsyn | and stores it | 13:06 |
@HeikoS | lisitsyn: very good idea | 13:06 |
lisitsyn | ok lets put it to some issue then.. | 13:06 |
lisitsyn | before we forget haha | 13:06 |
@HeikoS | lisitsyn: I think I rather implement it straight away | 13:06 |
@HeikoS | want it before gsoc | 13:06 |
@HeikoS | because then | 13:06 |
lisitsyn | yes but I will forget about that in an hour | 13:06 |
lisitsyn | :D | 13:06 |
@HeikoS | haha | 13:07 |
@HeikoS | lisitsyn: ok feel free to write it down then | 13:07 |
lisitsyn | too many things under control | 13:07 |
@HeikoS | lisitsyn: so the generator then just adds a couple of lines at the end of file | 13:07 |
lisitsyn | HeikoS: yes | 13:07 |
@HeikoS | lisitsyn: and this is done via SG class? | 13:07 |
lisitsyn | yes | 13:07 |
@HeikoS | that stores it to file? | 13:07 |
@HeikoS | ok | 13:07 |
@HeikoS | and then we creata a python script to compare? | 13:07 |
@HeikoS | or how do we do that? | 13:07 |
@HeikoS | best would be a matrix/vector equals on it | 13:08 |
@HeikoS | so that we can put accuracy in | 13:08 |
@HeikoS | where does the comparison happen? | 13:08 |
lisitsyn | I think outputs are collected | 13:08 |
lisitsyn | to some ENV | 13:08 |
lisitsyn | directory | 13:08 |
lisitsyn | then pairwise compared | 13:08 |
@HeikoS | comparison happens in? | 13:10 |
@HeikoS | python/shogun? | 13:10 |
@HeikoS | we store results in ascii? | 13:10 |
lisitsyn | HeikoS: I think it would be best to employ shogun stuff there | 13:12 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * c220dac / examples/meta/generator/targets/cpp.json: https://github.com/shogun-toolbox/shogun/commit/c220dac005387df3a32fd374198649053a57f1fe | 13:13 |
shogun-notifier- | shogun: use types from Library.i | 13:13 |
@HeikoS | lisitsyn: agreed | 13:13 |
@HeikoS | lisitsyn: so how does this work? | 13:13 |
@HeikoS | when does the check happen? | 13:14 |
@HeikoS | lisitsyn: at example runtime? | 13:14 |
lisitsyn | HeikoS: no, after | 13:14 |
lisitsyn | once all are done | 13:14 |
lisitsyn | we compare them | 13:14 |
@HeikoS | lisitsyn: ok so them all examples produce a single file? | 13:14 |
@HeikoS | with the same name as they are called? | 13:14 |
lisitsyn | I think they store a single file per example in some example directory | 13:14 |
lisitsyn | classifier/knn/python_output | 13:14 |
lisitsyn | HeikoS: what bothers me is the speed of such a thing | 13:15 |
lisitsyn | I expect it to be pretty slow | 13:15 |
@HeikoS | lisitsyn: dont know | 13:15 |
@HeikoS | I mean it is not real examples | 13:15 |
@HeikoS | just APIU | 13:15 |
@HeikoS | API | 13:15 |
@HeikoS | so few dimensions and few instances | 13:15 |
@HeikoS | but sure | 13:15 |
lisitsyn | yes but a lot of comparisons | 13:15 |
@HeikoS | lisitsyn: so how to do it? | 13:15 |
@HeikoS | binary? | 13:16 |
@HeikoS | we have had problems with that in the past | 13:16 |
lisitsyn | HeikoS: some additional python script I guess | 13:16 |
@HeikoS | lisitsyn: ugly ;) | 13:16 |
@HeikoS | dont tell drwiking | 13:16 |
lisitsyn | why? | 13:16 |
@HeikoS | he will be mad | 13:16 |
@HeikoS | haha | 13:16 |
@HeikoS | no but I think that is the way to go | 13:16 |
@HeikoS | how would the script do it? | 13:16 |
@HeikoS | what is the format of the files? | 13:16 |
lisitsyn | nah we could just pickle numpy stuff | 13:17 |
lisitsyn | and compare | 13:17 |
@HeikoS | since I assume we only want one file per example that collects all vectors? | 13:17 |
@HeikoS | lisitsyn: problem is generating pickles from the Shogun class | 13:17 |
lisitsyn | ah yes | 13:17 |
lisitsyn | true | 13:17 |
@HeikoS | lisitsyn: I want to avoid using serialisation in there as we havent updated that yet | 13:17 |
lisitsyn | ah ok | 13:17 |
@HeikoS | but simply store the vector/matrix itself | 13:17 |
lisitsyn | lets avoid it then | 13:17 |
lisitsyn | easy | 13:17 |
-!- travis-ci [~travis-ci@ec2-54-162-136-22.compute-1.amazonaws.com] has joined #shogun | 13:17 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112227306 | 13:17 |
-!- travis-ci [~travis-ci@ec2-54-162-136-22.compute-1.amazonaws.com] has left #shogun [] | 13:17 | |
@HeikoS | how? | 13:17 |
lisitsyn | print x.columns | 13:18 |
lisitsyn | ... | 13:18 |
lisitsyn | ;) | 13:18 |
lisitsyn | HeikoS: I think it is easy to do once it is integrated | 13:18 |
lisitsyn | integration takes much more time | 13:18 |
lisitsyn | once I have some call of a dummy thing in generate.py I could iterate fast | 13:19 |
@HeikoS | lisitsyn: ok | 13:24 |
@HeikoS | lisitsyn: sure main thing is that we dont need to rouch the .sg files | 13:24 |
@HeikoS | so we can start adding examples and do this later | 13:24 |
@HeikoS | lisitsyn: in that case, see email :) | 13:24 |
@HeikoS | lisitsyn: this and the lua issue, then integration is done | 13:25 |
lisitsyn | ok cool | 13:25 |
lisitsyn | HeikoS: going to take some break now | 13:25 |
lisitsyn | HeikoS: I will check lua once I get back to it | 13:26 |
lisitsyn | then start looking at this integration stuff | 13:26 |
lisitsyn | HeikoS: issue2 could be solved with ctags or clang api | 13:27 |
lisitsyn | maybe grep could be unreliable but would work too | 13:27 |
@HeikoS | lisitsyn: sure | 13:27 |
@HeikoS | lisitsyn: can you elaborate on issue 2? | 13:27 |
@HeikoS | give me a link or so | 13:27 |
@HeikoS | ah ctags | 13:27 |
@HeikoS | of course | 13:28 |
lisitsyn | we just run ctags | 13:28 |
@HeikoS | but we dont have that yet | 13:28 |
lisitsyn | we would need them anyway | 13:28 |
@HeikoS | no but then need to fix this first | 13:28 |
lisitsyn | to support other things like auto-sphinx-highlighting | 13:28 |
@HeikoS | we will not have them in a while | 13:28 |
lisitsyn | it is very easy to add ctags target | 13:28 |
lisitsyn | why? | 13:28 |
@HeikoS | lisitsyn: mmh maybe I dont really understand what you mean | 13:29 |
lisitsyn | we just need to pass ctags file to generate.py | 13:29 |
lisitsyn | and let him understand how to handle that | 13:29 |
lisitsyn | HeikoS: the ctags file is built on every built with a custom target | 13:30 |
lisitsyn | on every build* | 13:31 |
@HeikoS | ok and ctags file contains all the includes? | 13:31 |
@HeikoS | but wait | 13:31 |
@HeikoS | generate.py only sees the .sg file | 13:31 |
lisitsyn | HeikoS: yes you can easily extract path from ctags | 13:33 |
lisitsyn | HeikoS: run 'ctags -R src/shogun' | 13:34 |
lisitsyn | and open 'tags' | 13:34 |
lisitsyn | I like it more than grep because it does some parsing | 13:35 |
lisitsyn | while grep could break at some point | 13:36 |
@HeikoS | but what is the input to ctags? | 13:36 |
@HeikoS | a cpp file? | 13:36 |
@HeikoS | that is not available when generating meta examples | 13:36 |
@HeikoS | lisitsyn: maybe I do the grep thing for now as a hack to make things work and we use ctags later? | 13:37 |
@HeikoS | lisitsyn: or you do the ctags now, and I sort out the other issues | 13:37 |
-!- travis-ci [~travis-ci@ec2-184-73-100-71.compute-1.amazonaws.com] has joined #shogun | 13:48 | |
travis-ci | it's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112228398 | 13:48 |
-!- travis-ci [~travis-ci@ec2-184-73-100-71.compute-1.amazonaws.com] has left #shogun [] | 13:48 | |
lisitsyn | HeikoS: but we want to extract locations of shogun files right? | 13:49 |
lisitsyn | they are not generated | 13:49 |
@HeikoS | lisitsyn: true | 13:50 |
@HeikoS | ah now I get it | 13:50 |
@HeikoS | so we run ctags on the shogun src tree | 13:50 |
@HeikoS | and then pass stuff | 13:50 |
lisitsyn | yes | 13:50 |
@HeikoS | lisitsyn: wanna do that? | 13:50 |
lisitsyn | can do | 13:50 |
@HeikoS | but wait | 13:50 |
@HeikoS | then generator.py has to look the types up | 13:50 |
@HeikoS | so it has a list of types that appear | 13:50 |
@HeikoS | ah yes thats fine | 13:51 |
lisitsyn | HeikoS: it already had this thing | 13:51 |
lisitsyn | remember we patched it out | 13:51 |
@HeikoS | I dont remember ;) | 13:51 |
lisitsyn | HeikoS: so they were right in the grammar | 13:56 |
lisitsyn | => impossible to write down non-shogun type | 13:56 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 2fb7887 / examples/meta/generator/targets/csharp.json: https://github.com/shogun-toolbox/shogun/commit/2fb78874fef2f9b626a379e700f8dad6707c0207 | 13:57 |
shogun-notifier- | shogun: put in complex type | 13:57 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 43c2581 / examples/meta/generator/ (2 files): https://github.com/shogun-toolbox/shogun/commit/43c258164b22e6cbfcbb678cb9f7caa46d7711b1 | 13:57 |
shogun-notifier- | shogun: remove whitespaces for nicer visuals in cookbook snippets | 13:57 |
@HeikoS | lisitsyn: I think you did that | 13:57 |
@HeikoS | but well | 13:57 |
@HeikoS | ah yes | 13:57 |
@HeikoS | the hand created file | 13:57 |
lisitsyn | so now we are getting back | 13:57 |
lisitsyn | haha | 13:57 |
@HeikoS | where he checked whether something is a valid shogun type | 13:57 |
lisitsyn | yes | 13:57 |
@HeikoS | that was more to check actually | 13:57 |
@HeikoS | so rather than having runtime error, the generator would fail | 13:57 |
@HeikoS | it is actually good to have | 13:58 |
@HeikoS | so if you add the include magic, that check should be put in too | 13:58 |
lisitsyn | HeikoS: yeah ok | 13:58 |
@HeikoS | then one can locally run make meta_examples and catches any typos in shogun class even if no modular interfaces are enabled | 13:58 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 9a198b5 / examples/meta/cpp/CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/9a198b5e580e5bcf5bb12fc62958a516e8237ecd | 14:02 |
shogun-notifier- | shogun: unify names of generated tests | 14:02 |
@HeikoS | lisitsyn: https://travis-ci.org/shogun-toolbox/shogun/builds/112228398 | 14:02 |
@HeikoS | your cmake thing worked on travis | 14:02 |
lisitsyn | good | 14:02 |
@HeikoS | lisitsyn: ok so ctags is all yours | 14:03 |
@HeikoS | I need to solve this issue 1 | 14:03 |
@HeikoS | in eht email | 14:03 |
@HeikoS | for that need to change the grammar I think | 14:03 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 8d826fa / doc/cookbook/source/conf.py,doc/cookbook/source/templates/page.html: https://github.com/shogun-toolbox/shogun/commit/8d826fa57eaa2c284cf5672ecaec33f0a26b5f8c | 14:27 |
shogun-notifier- | shogun: add customized tab titles to cookbook and move cpp down the list | 14:27 |
-!- travis-ci [~travis-ci@ec2-54-162-136-22.compute-1.amazonaws.com] has joined #shogun | 14:29 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112228772 | 14:29 |
-!- travis-ci [~travis-ci@ec2-54-162-136-22.compute-1.amazonaws.com] has left #shogun [] | 14:29 | |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 546fcb6 / examples/meta/generator/targets/csharp.json: https://github.com/shogun-toolbox/shogun/commit/546fcb63988fa81a78175769f0987e5471646842 | 14:29 |
shogun-notifier- | shogun: remove whitespaces in csharp for nicer cookbook visuals | 14:29 |
-!- travis-ci [~travis-ci@ec2-54-157-222-14.compute-1.amazonaws.com] has joined #shogun | 14:30 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112228876 | 14:30 |
-!- travis-ci [~travis-ci@ec2-54-157-222-14.compute-1.amazonaws.com] has left #shogun [] | 14:30 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 14:31 | |
-!- travis-ci [~travis-ci@ec2-184-73-100-71.compute-1.amazonaws.com] has joined #shogun | 14:32 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112229236 | 14:32 |
-!- travis-ci [~travis-ci@ec2-184-73-100-71.compute-1.amazonaws.com] has left #shogun [] | 14:32 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 14:39 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 14:39 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit] | 14:40 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 14:43 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 14:43 | |
shogun-notifier- | shogun: Sergey Lisitsyn :feature/sphinxdoc * 992eba6 / CMakeLists.txt,src/shogun/CMakeLists.txt,cmake/FindCtags.cmake: https://github.com/shogun-toolbox/shogun/commit/992eba6bee178bbba2e38bc254d405567111b247 | 14:44 |
shogun-notifier- | shogun: Integrate ctags target into the build | 14:44 |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit] | 14:45 | |
-!- travis-ci [~travis-ci@ec2-184-73-100-71.compute-1.amazonaws.com] has joined #shogun | 15:21 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112239514 | 15:21 |
-!- travis-ci [~travis-ci@ec2-184-73-100-71.compute-1.amazonaws.com] has left #shogun [] | 15:21 | |
-!- travis-ci [~travis-ci@ec2-184-73-100-71.compute-1.amazonaws.com] has joined #shogun | 15:49 | |
travis-ci | it's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112241456 | 15:49 |
-!- travis-ci [~travis-ci@ec2-184-73-100-71.compute-1.amazonaws.com] has left #shogun [] | 15:49 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 16:21 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 16:21 | |
@HeikoS | lisitsyn: jojojo | 16:22 |
@HeikoS | just saw the patch | 16:22 |
lisitsyn | HeikoS: yes ctags is being built | 16:22 |
lisitsyn | (if you ask for it) | 16:23 |
@HeikoS | cool | 16:23 |
@HeikoS | so this should be done when -DENABLE_TESTING=On | 16:23 |
@HeikoS | since we only need it for that | 16:23 |
lisitsyn | HeikoS: no, I think it is better this way | 16:24 |
@HeikoS | ok | 16:24 |
lisitsyn | it is run if you do 'make ctags' | 16:24 |
@HeikoS | ah I see | 16:24 |
@HeikoS | btw | 16:24 |
@HeikoS | auto test_labels = knn->apply(feats_test); auto output = test_labels->get_values(); | 16:24 |
lisitsyn | or if you have some dependency | 16:24 |
@HeikoS | auto test_labels = knn->apply(feats_test); | 16:24 |
@HeikoS | auto output = test_labels->get_values(); | 16:24 |
lisitsyn | yes | 16:24 |
@HeikoS | how does this work? | 16:24 |
@HeikoS | when there is abiguitiy? | 16:24 |
@HeikoS | like in get_values | 16:24 |
@HeikoS | might return SGVector or float64_t* | 16:25 |
lisitsyn | how can there be ambiguity? | 16:25 |
@HeikoS | when name is overloaded | 16:25 |
@HeikoS | then compile error? | 16:25 |
lisitsyn | it doesn't sound possible for me | 16:25 |
@HeikoS | lisitsyn: can auto be used for any statement like that? where something is copied? | 16:25 |
lisitsyn | auto can replace basically everything | 16:26 |
lisitsyn | it just infers the type | 16:26 |
@HeikoS | great | 16:26 |
@HeikoS | so issue 1 of email solved | 16:26 |
lisitsyn | HeikoS: I don't really get get_values example | 16:26 |
@HeikoS | just remains issue 2 | 16:26 |
lisitsyn | you can't have two methods returning different things | 16:27 |
lisitsyn | but the same name | 16:27 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 601ea07 / examples/meta/generator/targets/cpp.json: https://github.com/shogun-toolbox/shogun/commit/601ea070cf1adbb335d8113fa1773b8d24018148 | 16:27 |
shogun-notifier- | shogun: solve type problem in cpp meta example via 'auto' keyword | 16:27 |
shogun-notifier- | shogun: Heiko Strathmann :feature/sphinxdoc * 73634c5 / examples/meta/src/classifier/knn.sg: https://github.com/shogun-toolbox/shogun/commit/73634c5a452efd95f3af3264892f072cce64161a | 16:27 |
shogun-notifier- | shogun: re-add output as now works | 16:27 |
@HeikoS | lisitsyn: ok then, auto is new for me I was confused | 16:27 |
lisitsyn | HeikoS: it is just replaced with the return type of the method | 16:29 |
@HeikoS | lisitsyn: yep got it | 16:29 |
@HeikoS | preprocessor type of operation | 16:30 |
@HeikoS | well good | 16:30 |
lisitsyn | yeah but it is a bit more | 16:30 |
@HeikoS | if you solve the includes then we have cpp integrated | 16:30 |
lisitsyn | ok let me try now | 16:30 |
@HeikoS | ok I will do some stuff as well | 16:34 |
@HeikoS | as fixing lua | 16:34 |
@HeikoS | lisitsyn: there is a weird problem | 16:47 |
@HeikoS | you know in the conf.py of sphinx | 16:47 |
@HeikoS | we define the target languages | 16:47 |
lisitsyn | yeah | 16:47 |
@HeikoS | target_languages = ( | 16:47 |
@HeikoS | ('python', 'py'), | 16:47 |
@HeikoS | ('octave', 'm'), | 16:47 |
@HeikoS | ('java', 'java'), | 16:47 |
@HeikoS | ('ruby', 'rb'), | 16:47 |
@HeikoS | ('r', 'R'), | 16:47 |
@HeikoS | ('csharp', 'cs'), | 16:47 |
@HeikoS | ('cpp', 'cpp'), | 16:48 |
@HeikoS | ) | 16:48 |
@HeikoS | so when R appeary before ruby in there | 16:48 |
@HeikoS | then in the html, the R-tab shows the ruby example | 16:48 |
@HeikoS | must be some string matching | 16:48 |
lisitsyn | wow | 16:48 |
@HeikoS | if I swap the order, i.e. ruby before R, everything is fine | 16:48 |
@HeikoS | lisitsyn: however, the html file is fine, it contains the R code | 16:49 |
@HeikoS | it is just not shown when clicking on the tab | 16:49 |
lisitsyn | HeikoS: can you paste it somewhere? | 16:49 |
lisitsyn | it could be a matter of some wrong identifiers of elements or so | 16:49 |
lisitsyn | just the code | 16:49 |
@HeikoS | berfore this patch: https://github.com/shogun-toolbox/shogun/commit/c7991e3f478ed197237dc8fee2352a383fba0638 | 16:49 |
@HeikoS | it was wrong, after the patch it is ok | 16:49 |
@HeikoS | lisitsyn: but the ctags is more important | 16:50 |
lisitsyn | ok | 16:50 |
@HeikoS | this is fixed (even though in an ugly way ;) ) | 16:50 |
-!- travis-ci [~travis-ci@ec2-54-162-136-22.compute-1.amazonaws.com] has joined #shogun | 16:56 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112254980 | 16:56 |
-!- travis-ci [~travis-ci@ec2-54-162-136-22.compute-1.amazonaws.com] has left #shogun [] | 16:56 | |
@HeikoS | lisitsyn: it starts looking really cool | 17:06 |
@HeikoS | with all of the languages in :) | 17:06 |
lisitsyn | HeikoS: yes it is great | 17:06 |
-!- witness_ [uid10044@gateway/web/irccloud.com/x-ixkpbqmtnnlmldwv] has joined #shogun | 17:06 | |
@HeikoS | lisitsyn: hows the ctags going? | 17:27 |
lisitsyn | HeikoS: slow, switched to another thing | 17:30 |
lisitsyn | now getting back | 17:30 |
lisitsyn | man this context switching is so bad | 17:30 |
lisitsyn | HeikoS: I start to feel like a cmake ninja | 17:34 |
@HeikoS | yeah context switching no good | 17:34 |
@HeikoS | lisitsyn: ha! | 17:34 |
@HeikoS | yeah | 17:34 |
@HeikoS | I finally got the point of cmake over the last few days | 17:34 |
@HeikoS | saw this? | 17:34 |
@HeikoS | https://github.com/youssef-emad/shogun_notebooks/blob/master/Classification.ipynb | 17:34 |
lisitsyn | oh | 17:35 |
lisitsyn | that's good | 17:35 |
@HeikoS | quite nice | 17:35 |
lisitsyn | like in this old days | 17:35 |
@HeikoS | I like the notebook, needs some love but I will try to get it merged | 17:35 |
@HeikoS | would also be good to have all those in a single plot | 17:35 |
lisitsyn | oh my favourite gaussian naive bayes! | 17:35 |
lisitsyn | that's with what I started to shogun | 17:35 |
@HeikoS | for me it was svm | 17:36 |
@HeikoS | ah I got a hand on the lua thing now | 17:39 |
@HeikoS | think it is wrong environment | 17:39 |
@HeikoS | last problem that is!!! | 17:39 |
lisitsyn | haha | 17:42 |
lisitsyn | okie | 17:42 |
lisitsyn | HeikoS: ok I parsed it already now trying to understand this thing | 17:43 |
-!- travis-ci [~travis-ci@ec2-54-157-222-14.compute-1.amazonaws.com] has joined #shogun | 17:51 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112259179 | 17:51 |
-!- travis-ci [~travis-ci@ec2-54-157-222-14.compute-1.amazonaws.com] has left #shogun [] | 17:51 | |
lisitsyn | HeikoS: do we need it for any other lang? | 17:52 |
lisitsyn | java? | 17:52 |
@HeikoS | lisitsyn: nope | 17:52 |
@HeikoS | import org.shogun.*; | 17:52 |
lisitsyn | ok | 17:53 |
@HeikoS | and in Python we do explicit, but it doesnt matter and things are the same as for swig | 17:53 |
@HeikoS | from modshogun import EuclideanDistance, KNN, MulticlassLabels, CSVFile, RealFeatures | 17:53 |
@HeikoS | so doesnt matter | 17:53 |
@HeikoS | only c++ | 17:53 |
@HeikoS | lisitsyn: I think lua typemap might be borken | 17:55 |
@HeikoS | overloaded methods do not seem to be translated | 17:55 |
lisitsyn | lua had some issues with overloading | 17:55 |
lisitsyn | iirc | 17:56 |
@HeikoS | lisitsyn: yes | 18:01 |
@HeikoS | a phew so this can be fixed in hacky way | 18:01 |
lisitsyn | HeikoS: ok I am close | 18:05 |
@HeikoS | lisitsyn: same | 18:06 |
@HeikoS | lua is merged | 18:06 |
@HeikoS | for now avoid calling overloaded methods in meta examples | 18:06 |
@HeikoS | which is stupid | 18:06 |
@HeikoS | needs fix | 18:06 |
@HeikoS | but build works for now | 18:06 |
@HeikoS | changed the knn.sg example | 18:06 |
-!- 16WAAFTN5 [~travis-ci@ec2-54-157-222-14.compute-1.amazonaws.com] has joined #shogun | 18:07 | |
-!- 7YUAAFPIP [~travis-ci@ec2-54-157-222-14.compute-1.amazonaws.com] has joined #shogun | 18:07 | |
16WAAFTN5 | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112260500 | 18:07 |
-!- 16WAAFTN5 [~travis-ci@ec2-54-157-222-14.compute-1.amazonaws.com] has left #shogun [] | 18:07 | |
7YUAAFPIP | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112260500 | 18:07 |
-!- 7YUAAFPIP [~travis-ci@ec2-54-157-222-14.compute-1.amazonaws.com] has left #shogun [] | 18:07 | |
-!- travis-ci [~travis-ci@ec2-54-197-129-30.compute-1.amazonaws.com] has joined #shogun | 18:07 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112263186 | 18:07 |
-!- travis-ci [~travis-ci@ec2-54-197-129-30.compute-1.amazonaws.com] has left #shogun [] | 18:07 | |
@HeikoS | lisitsyn: let me know how it goes | 18:08 |
lisitsyn | HeikoS: works | 18:15 |
lisitsyn | polishing | 18:15 |
@HeikoS | lisitsyn: will temp. remove lua from build, otherwise the examples are too ugly | 18:19 |
lisitsyn | HeikoS: uhh I tried my best to make it less ugly | 18:34 |
lisitsyn | commit in a few minutes | 18:38 |
@HeikoS | lisitsyn: waiting :) | 18:40 |
lisitsyn | here | 18:42 |
lisitsyn | https://github.com/shogun-toolbox/shogun/commit/981fc4b3a03fcec45f50334ddfef31d425d8cf92 | 18:43 |
lisitsyn | HeikoS: ^ no irc notify for some reason | 18:43 |
lisitsyn | HeikoS: cookbook fails here | 18:49 |
lisitsyn | warns on regression clustering | 18:49 |
lisitsyn | and wiki link to knn | 18:49 |
lisitsyn | then fails | 18:49 |
@HeikoS | lisitsyn: running | 18:52 |
@HeikoS | locally | 18:52 |
@HeikoS | it works | 18:53 |
@HeikoS | on my machine | 18:53 |
@HeikoS | without any modular interfaces on | 18:54 |
@HeikoS | lisitsyn: ok gotta go now | 18:57 |
@HeikoS | really made some progress today! | 18:57 |
-!- travis-ci [~travis-ci@ec2-54-197-129-30.compute-1.amazonaws.com] has joined #shogun | 18:58 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112269016 | 18:58 |
-!- travis-ci [~travis-ci@ec2-54-197-129-30.compute-1.amazonaws.com] has left #shogun [] | 18:58 | |
lisitsyn | yeah | 18:58 |
@HeikoS | made some smallish comments in the patch | 18:58 |
lisitsyn | ok | 19:00 |
@HeikoS | lisitsyn: hey btw about the some | 19:11 |
@HeikoS | isnt there a way to do this automatically? | 19:11 |
@HeikoS | without having to write some<CKNN> | 19:11 |
@HeikoS | but just CKNN | 19:12 |
lisitsyn | HeikoS: not really | 19:13 |
@HeikoS | ok | 19:13 |
lisitsyn | HeikoS: the only alternative is to make them light-weight | 19:13 |
@HeikoS | ok next step is the testing | 19:13 |
lisitsyn | I think your last commit broke the build | 19:14 |
@HeikoS | why? | 19:14 |
lisitsyn | make[4]: *** No rule to make target `examples/meta/cpp/classifier/knn.cpp', needed by `examples/meta/cpp/CMakeFiles/build_cpp_meta_examples'. Stop. | 19:14 |
@HeikoS | works here | 19:14 |
lisitsyn | did you remove the build and run it all over again? | 19:14 |
@HeikoS | uh | 19:14 |
@HeikoS | yes | 19:14 |
@HeikoS | fixing | 19:15 |
@HeikoS | lisitsyn: where to start best with this testing thing? | 19:17 |
lisitsyn | HeikoS: integration? | 19:18 |
@HeikoS | yes | 19:18 |
lisitsyn | HeikoS: add a flag to generate.py, make it collect all the variables | 19:18 |
@HeikoS | ok cool | 19:18 |
lisitsyn | HeikoS: you want to do something on that as well? | 19:19 |
@HeikoS | think we do not need to add anything to json no? | 19:19 |
@HeikoS | dont know | 19:19 |
@HeikoS | currently cleaning up few things here and there | 19:19 |
@HeikoS | I can hepl integrate once prototype is working | 19:19 |
@HeikoS | with all the changes in data etc | 19:19 |
@HeikoS | but wanted to discuss where to start | 19:19 |
@HeikoS | so you add the flat | 19:19 |
@HeikoS | what shall I do? what after that? | 19:19 |
lisitsyn | HeikoS: once we have these 'Store(vector);' calls | 19:20 |
lisitsyn | someone would need to implement that store | 19:20 |
lisitsyn | Store(variable_name, variable_value); | 19:20 |
lisitsyn | like that | 19:20 |
@HeikoS | lisitsyn: how would it behave? | 19:21 |
@HeikoS | like Store(fname) | 19:21 |
@HeikoS | s=Store(fname) | 19:21 |
@HeikoS | s.store(SGVector v) | 19:21 |
@HeikoS | s.store(SGMatrix m) | 19:21 |
@HeikoS | s.store(basic_type b) | 19:21 |
@HeikoS | ? | 19:21 |
lisitsyn | yes | 19:21 |
@HeikoS | ok and format? | 19:23 |
@HeikoS | ordering? | 19:23 |
lisitsyn | HeikoS: what ordering? | 19:23 |
@HeikoS | so the store is a container I guess? | 19:24 |
@HeikoS | what file format? | 19:24 |
@HeikoS | should be readable from python I guess | 19:24 |
@HeikoS | to just csv? | 19:24 |
@HeikoS | or python format? | 19:24 |
@HeikoS | [[0,1],[2,3]] | 19:25 |
@HeikoS | ah I dont want to write this | 19:25 |
@HeikoS | rather use a nice csv writer | 19:25 |
lisitsyn | HeikoS: haha | 19:25 |
lisitsyn | HeikoS: I think once it is in | 19:25 |
lisitsyn | it is very easy to do | 19:25 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 19:27 | |
-!- witness_ [uid10044@gateway/web/irccloud.com/x-ixkpbqmtnnlmldwv] has quit [Quit: Connection closed for inactivity] | 19:28 | |
@HeikoS | you mean once the store is in, it is easy to do the testing? | 19:29 |
@HeikoS | lisitsyn: the store can be changed later | 19:30 |
@HeikoS | however, we want to finalize it before we start adding test reference files | 19:30 |
@HeikoS | lisitsyn: so not sure what is the best way to proceed here | 19:31 |
@HeikoS | rely on existing serialization? | 19:31 |
@HeikoS | or something else | 19:31 |
@HeikoS | serialization is overkill as we only want numerical stuff | 19:32 |
@HeikoS | but writing a store for all shogun types is suicide | 19:32 |
@HeikoS | lisitsyn: well, lets talk tomorrow or so | 19:32 |
@HeikoS | tired :) | 19:32 |
@HeikoS | bye | 19:32 |
lisitsyn | HeikoS: hmm do you really want to keep these files? | 19:32 |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 19:32 | |
lisitsyn | uh | 19:32 |
lisitsyn | ok | 19:32 |
lisitsyn | see you | 19:32 |
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 240 seconds] | 20:47 | |
-!- travis-ci [~travis-ci@ec2-54-197-129-30.compute-1.amazonaws.com] has joined #shogun | 21:12 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/112281517 | 21:12 |
-!- travis-ci [~travis-ci@ec2-54-197-129-30.compute-1.amazonaws.com] has left #shogun [] | 21:12 | |
shogun-buildbot | build #2766 of deb3 - modular_interfaces is complete: Failure [failed test libshogun] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/2766 blamelist: Bj?rn Esser <me@besser82.io> | 21:43 |
-!- Netsplit *.net <-> *.split quits: @drwiking, lupinix, sonne|work, @ChanServ | 22:45 | |
-!- Netsplit *.net <-> *.split quits: aditya123 | 22:45 | |
-!- Netsplit *.net <-> *.split quits: rcurtin | 22:45 | |
-!- Netsplit *.net <-> *.split quits: lisitsyn, octokat, shogun-buildbot, @sonney2k | 22:46 | |
-!- drwiking_ [~wiking@info2k1.hu] has joined #shogun | 22:52 | |
-!- Netsplit over, joins: lupinix, shogun-buildbot, lisitsyn | 22:52 | |
-!- sonne|work1 [~sonnenbu@ip5b414dfc.dynamic.kabel-deutschland.de] has joined #shogun | 22:52 | |
-!- Netsplit over, joins: aditya123, octokat, @ChanServ, rcurtin, @sonney2k | 22:52 | |
-!- shogun-toolbox [~shogun@7nn.de] has quit [Ping timeout: 252 seconds] | 23:19 | |
--- Log closed Sat Feb 27 23:19:32 2016 | ||
--- Log opened Sat Feb 27 23:19:40 2016 | ||
-!- shogun-toolbox [~shogun@7nn.de] has joined #shogun | 23:19 | |
-!- Irssi: #shogun: Total of 11 nicks [2 ops, 0 halfops, 0 voices, 9 normal] | 23:19 | |
-!- Irssi: Join to #shogun was synced in 6 secs | 23:19 | |
-!- sonney2k [~shogun@7nn.de] has quit [Ping timeout: 240 seconds] | 23:27 | |
-!- lisitsyn [~lisitsyn@37.139.2.75] has quit [Ping timeout: 264 seconds] | 23:32 | |
-!- lisitsyn [~lisitsyn@37.139.2.75] has joined #shogun | 23:34 | |
-!- sonney2k [~shogun@7nn.de] has joined #shogun | 23:34 | |
--- Log closed Sun Feb 28 00:00:15 2016 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!