--- Log opened Wed Mar 15 00:00:05 2017 | ||
OXPHOS | wiking: I might have fixed the viennacl error on buildbot. How could I test it? | 01:21 |
---|---|---|
@wiking | mmm | 01:21 |
@wiking | the only way | 01:21 |
@wiking | atm | 01:21 |
@wiking | is to push it into develop | 01:22 |
@wiking | :) | 01:22 |
@wiking | i mean | 01:22 |
@wiking | you can reproduce the runtime env | 01:22 |
@wiking | with docker of course | 01:22 |
OXPHOS | wiking: may I ask why travis doesn't test viennacl operations? Sounds like it can be setup in docker | 01:37 |
@wiking | OXPHOS, in the travis docker image | 01:39 |
@wiking | we have not enabled viennacl | 01:39 |
@wiking | because those instances | 01:39 |
@wiking | are some virtual amazon instances | 01:39 |
@wiking | so it actually would be a NOP | 01:39 |
@wiking | they dont have gpu | 01:39 |
@wiking | nothing | 01:39 |
@wiking | barely support multithreading :) | 01:40 |
@wiking | like for example all our openmp tests are actually passing there | 01:40 |
@wiking | although some of them should fail | 01:40 |
@wiking | :) | 01:40 |
OXPHOS | wiking: I see. So I can still test viennacl and multi-thread if I am running docker on my own machine? | 01:43 |
@wiking | of course | 01:43 |
@wiking | docker is not a virtual machine | 01:43 |
@wiking | it's just a container on kernel level | 01:44 |
@wiking | UNLESS!! | 01:44 |
@wiking | you use osx | 01:44 |
@wiking | which you do i know | 01:44 |
OXPHOS | .. | 01:44 |
@wiking | hence then it's a virtual machine | 01:44 |
@wiking | :))))) | 01:44 |
@wiking | ok look | 01:44 |
@wiking | is it one commit? | 01:44 |
@wiking | or can you make it one commit? | 01:44 |
@wiking | and send the patch to my mail | 01:44 |
@wiking | git format-patch -1 | 01:44 |
@wiking | and i can test it on that machine | 01:44 |
@wiking | like now | 01:44 |
OXPHOS | wiking: kk thanks a lot! I think it is some kind of weird header issue. Because all I did was moving things around and now ViennaCL cannot find one method within itself | 01:46 |
OXPHOS | but since it's kinda troublesome, let me do a little more research | 01:46 |
@wiking | no worries | 01:46 |
OXPHOS | wiking: I'll fix the warnings for now. | 01:47 |
@wiking | ok look | 01:48 |
@wiking | if you can send anything | 01:48 |
@wiking | regarding the fix | 01:48 |
@wiking | for those errors | 01:48 |
@wiking | (not warnings) | 01:49 |
@wiking | that'd be awesome | 01:49 |
@wiking | because now this is blocking the whole testing infra :) | 01:49 |
@wiking | so having the compiler errors | 01:49 |
@wiking | fixe | 01:49 |
@wiking | d | 01:49 |
@wiking | should be a prio | 01:49 |
OXPHOS | wiking: Sure I'm on it. I already fixed the warnings. Just need to push it to dev. | 01:50 |
@wiking | k | 01:52 |
@wiking | cool | 01:52 |
@sukey | New Commit "fix buildbot warnings" to shogun-toolbox/shogun by OXPHOS: https://github.com/shogun-toolbox/shogun/commit/15db5756172ae90ef33a591b0747ef5b4d3ca7c2 | 01:52 |
@wiking | OXPHOS, ok lemme test this patch | 02:31 |
OXPHOS | wiking: thx! | 02:32 |
@wiking | /usr/include/viennacl/matrix.hpp:1948:11: error: 'ambm_m' is not a member of 'viennacl::linalg' | 02:34 |
@wiking | where is this coming from | 02:34 |
@wiking | but this | 02:37 |
@wiking | seems to be some viennacl shit | 02:37 |
OXPHOS | wiking: so the same errors? https://github.com/viennacl/viennacl-dev/blob/master/viennacl/linalg/matrix_operations.hpp#L174 | 02:38 |
-!- mikeling|afk [uid89706@gateway/web/irccloud.com/x-uzcsyziwrtraokjz] has joined #shogun | 02:38 | |
@wiking | yeah | 02:39 |
@wiking | /usr/include/viennacl/matrix.hpp: In static member function 'static void viennacl::linalg::detail::op_executor<viennacl::matrix_base<T, F>, viennacl::op_inplace_sub, viennacl::matrix_expression<const LHS, const RHS, viennacl::op_sub> >::apply(viennacl::matrix_base<T, F>&, const viennacl::matrix_expression<const viennacl::matrix_expression<const viennacl::matrix_base<T, F>, const ScalarType, viennacl::op_div>, const viennacl::matrix_expression<const vi | 02:39 |
@wiking | ennacl::matrix_base<T, F>, const ScalarType2, viennacl::op_div>, viennacl::op_sub>&)': | 02:39 |
@wiking | /usr/include/viennacl/matrix.hpp:2454:11: error: 'ambm_m' is not a member of 'viennacl::linalg' | 02:39 |
@wiking | viennacl::linalg::ambm_m(lhs, | 02:39 |
OXPHOS | wiking: I moved the header from linalg/**/apply.h to gpumatrix in this pr: https://github.com/shogun-toolbox/shogun/commit/75e77f192c4c9c64874c53aa652883be512929fe | 02:39 |
@wiking | ok but why should this cause this trouble? :D | 02:40 |
@wiking | and i suppose you are having viennacl 1.7.x | 02:40 |
@wiking | and all's good | 02:40 |
@wiking | right? | 02:40 |
-!- nagdev [dce39546@gateway/web/freenode/ip.220.227.149.70] has joined #shogun | 02:40 | |
OXPHOS | wiking: no idea at all..yep it worked good on my machine | 02:40 |
OXPHOS | thinking about weird header order stuff | 02:41 |
@wiking | lol | 02:41 |
@wiking | suddenly | 02:41 |
@wiking | src/shogun/classifier/svm/SubGradientSVM.cpp | 02:41 |
@wiking | is causing shit to me | 02:41 |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has joined #shogun | 02:46 | |
travis-ci | it's Pan Deng / Zora'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/211182181 | 02:46 |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has left #shogun [] | 02:46 | |
@wiking | OXPHOS, | 02:51 |
@wiking | ok i think i have a fix | 02:51 |
@wiking | still need to run a unit test for it | 02:51 |
@wiking | will push it to develop directly if it works out | 02:51 |
OXPHOS | all ears | 02:51 |
@wiking | ok? | 02:51 |
OXPHOS | okay | 02:51 |
OXPHOS | thx | 02:51 |
@wiking | +#include <viennacl/linalg/matrix_operations.hpp> | 02:51 |
@wiking | +#include <viennacl/linalg/vector_operations.hpp> | 02:51 |
@wiking | +#include <viennacl/matrix.hpp> | 02:51 |
@wiking | instead of this | 02:51 |
@wiking | +#include <viennacl/matrix.hpp> | 02:52 |
@wiking | +#include <viennacl/linalg/matrix_operations.hpp> | 02:52 |
@wiking | +#include <viennacl/linalg/vector_operations.hpp> | 02:52 |
@wiking | dunno if it work actually on 1.7.x | 02:52 |
@wiking | but will try to test it | 02:52 |
OXPHOS | viennacl/matrix.hpp | 02:52 |
OXPHOS | includes | 02:52 |
OXPHOS | viennacl/linalg/matrix_operations.hpp | 02:53 |
OXPHOS | so | 02:53 |
OXPHOS | +#include <viennacl/matrix.hpp> | 02:53 |
OXPHOS | +#include <viennacl/linalg/vector_operations.hpp> | 02:53 |
OXPHOS | might work | 02:53 |
OXPHOS | I had no problem at all with just +#include <viennacl/matrix.hpp> | 02:53 |
@wiking | k | 02:55 |
@wiking | i'm about to do the ctest | 02:57 |
@wiking | if it's ok | 02:57 |
@wiking | then i'll push | 02:57 |
OXPHOS | finger crossed | 02:58 |
@wiking | oh man | 02:59 |
@wiking | what a shit ../src/shogun/base/SGObject.h:389:2: error: control may reach end of non-void function [-Werror,-Wreturn-type] | 02:59 |
@wiking | but ok | 02:59 |
@wiking | unrelated | 02:59 |
@wiking | /home/buildslave/trusty_-_libshogun_-_viennacl/build/src/shogun/mathematics/linalg/LinalgBackendViennaCL.h:44:35: fatal error: viennacl/linalg/sum.hpp: No such file or directory | 03:00 |
@wiking | #include <viennacl/linalg/sum.hpp> | 03:00 |
@wiking | ^ | 03:00 |
@wiking | :D | 03:00 |
@wiking | ok so lemme find out | 03:01 |
@wiking | which viennacl had the first time sum | 03:01 |
@wiking | https://github.com/viennacl/viennacl-dev/commit/4063c941235d46804cd448db7ddecf0c3238548f#diff-9737fc4ea031985d194e70b3e9fbacd9 | 03:01 |
@wiking | ok this is pretty funky | 03:01 |
OXPHOS | a mess | 03:01 |
@wiking | release-1.7.0 | 03:01 |
@wiking | ok | 03:02 |
@wiking | so we need to do an if there for the versioning | 03:02 |
@wiking | OXPHOS, what should we do if we dont have sum? | 03:03 |
@wiking | :) | 03:03 |
@wiking | can we force fallback on eigne? | 03:03 |
@wiking | eigen? | 03:03 |
OXPHOS | wiking: lemme see how the old linalg does sum | 03:03 |
OXPHOS | it doens't make sense | 03:03 |
OXPHOS | wiking: okay it was the verbose kernel. we can still do that | 03:04 |
@sukey | Pull Request #3694 "Fix wrong behaviour of SGIO::progress" merged by vigsterkr - https://github.com/shogun-toolbox/shogun/pull/3694 | 03:05 |
@wiking | ok look | 03:05 |
@sukey | New Commit "Merge pull request #3694 from geektoni/patch-12 | 03:05 |
@sukey | Fix wrong behaviour of SGIO::progress" to shogun-toolbox/shogun by vigsterkr: https://github.com/shogun-toolbox/shogun/commit/2599ce58e1d80330a6eb691f8d155aa3600ffa84 | 03:05 |
@wiking | i'll push to develop | 03:05 |
@wiking | the mini header fix | 03:05 |
@wiking | then what you need to do is | 03:05 |
@wiking | check for VIENNACL_VERSION macro | 03:05 |
@wiking | so #if 10700 <= VIENNACL_VERSION | 03:06 |
@wiking | then you can have sum | 03:06 |
@sukey | New Commit "move matrix_operations header under linalg.h" to shogun-toolbox/shogun by vigsterkr: https://github.com/shogun-toolbox/shogun/commit/4d9979f80b3336322b5a26f23ce1d06ab39bd718 | 03:06 |
OXPHOS | wiking: kk. | 03:07 |
-!- mikeling|afk is now known as mikeling | 03:13 | |
mikeling | wiking: do we have existing environment variable in Shogun? I only found we query variable like "SHOGUN_GPU_WARNINGS" but failed to find where we define or add it | 03:30 |
@wiking | mikeling, it's there | 03:32 |
@wiking | and there are 2 others | 03:32 |
@wiking | SHOGUN_NUM_THREADS | 03:32 |
@wiking | and the one with LOGLEVEL | 03:32 |
@wiking | i mean | 03:32 |
@wiking | if you would like to write a wrapper class | 03:33 |
@wiking | that actually gets a (ENV VAR NAME, FUNCTION POINTER that handles the logic for the ENV VAR setting, DOCUMENTATION) triplet | 03:33 |
@wiking | and generates the code | 03:33 |
@wiking | that would be good | 03:33 |
nagdev | Are there any entrance tasks for the "Application to real world" project? I couldn't find anything in the issues or the wiki. | 03:47 |
nagdev | Should I just come up with my own data and problem statement? | 03:47 |
@wiking | nagdev, yes you should come up with your own story there if the ones there are not something you'd like to work on | 03:51 |
@wiking | but | 03:51 |
@wiking | regardless of things | 03:51 |
@wiking | we need you do do some entrance tasks | 03:51 |
@wiking | because we need to see how you can develop in an opensource project | 03:52 |
@wiking | and there are no such things as entrance tasks for "Application to real world" | 03:52 |
@wiking | just take any entrance task - does not need to be related with your project proposal | 03:52 |
@wiking | and fix some of them | 03:52 |
@wiking | there are many entrance tasks that does not really require any serious c++ or coding skills | 03:53 |
nagdev | ok thanks | 03:56 |
nagdev | should I discuss my project proposal (once I have it) here or just in my application? | 03:57 |
mikeling | wiking: oh, I think I got the point. So I need to write a class like "dataPath", and define 'set_data_path()', 'get_data_path()'. Then init that class in init.cpp and try to query SHOGUN_DATA_DIR by getenv and call 'set_data_path()' if the getenv return false | 03:57 |
mikeling | right? | 03:57 |
@wiking | mmmm | 03:58 |
@wiking | wait | 03:58 |
@wiking | you are now talking about https://github.com/shogun-toolbox/shogun/issues/3697 | 03:58 |
@wiking | ? | 03:58 |
mikeling | yep | 03:59 |
@wiking | ok why do you want to make this part of shogun itself? | 03:59 |
@wiking | or your idea is | 03:59 |
@wiking | that there's a common shogun fucntion | 03:59 |
@wiking | that one can query from any interface | 03:59 |
@wiking | and that has the correct path | 04:00 |
@wiking | of the data dir? | 04:00 |
@wiking | because i had something else in my mind | 04:01 |
@wiking | but we can talk about this | 04:01 |
mikeling | yeah, or we should only make a reference in notebooks? | 04:01 |
@wiking | because the fact about | 04:01 |
@wiking | the | 04:01 |
@wiking | notebooks | 04:01 |
@wiking | it's easy to just fix it there | 04:01 |
@wiking | right? | 04:01 |
mikeling | yep | 04:01 |
@wiking | os.getenv('SHOGUN_DATA_DIR') | 04:01 |
@wiking | using simple python functionality | 04:01 |
@wiking | like this you dont need to touch the library itself | 04:02 |
@wiking | just the sources | 04:02 |
@wiking | of the ipython notebooks | 04:02 |
@wiking | but | 04:02 |
@wiking | indeed | 04:02 |
@wiking | it might be beneficial | 04:02 |
@wiking | to have a common way to query this variable | 04:02 |
@wiking | in every interface | 04:02 |
@wiking | (r, jvm, python etc) | 04:02 |
@wiking | if you wanna do that | 04:02 |
@wiking | let's discuss your proposal there in the issue | 04:02 |
@wiking | because i want others to give input on this | 04:02 |
@wiking | ok? | 04:02 |
mikeling | hmmm, but before we call os.getenv('SHOGUN_DATA_DIR'), we need create it(add it) in the environment variable right? | 04:03 |
@wiking | yeah... but that's the user's/runtime environment's task | 04:03 |
@wiking | not the libraries | 04:03 |
@wiking | we dont need to be able to set it per se | 04:04 |
@wiking | just read it | 04:04 |
@wiking | because we have no idea where shogun will ever be placed | 04:04 |
@wiking | let along the data dir | 04:04 |
@wiking | hence it's an env var | 04:04 |
@wiking | anyhow | 04:05 |
@wiking | propose your idea in the issue | 04:05 |
@wiking | and let's discuss it there further | 04:05 |
mikeling | but actually, if someone like me query the whole git rep with its submodule | 04:05 |
mikeling | the 'data' actually been putted in the same folder | 04:06 |
mikeling | with shogun codebase | 04:06 |
mikeling | ok | 04:06 |
@sukey | Issue #3605 "port parallel code using pthread"- https://github.com/shogun-toolbox/shogun/issues/3605 | 04:06 |
@wiking | yes | 04:07 |
@wiking | mikeling, but imagine you take the notebook | 04:07 |
@wiking | somewhere else | 04:07 |
@wiking | like copy it | 04:07 |
@wiking | then the only way currently to fix it (because it does a relative reference on the data) | 04:07 |
@wiking | is to replace all those paths in the notebook | 04:08 |
@wiking | instead of just setting a variable :) | 04:08 |
@sukey | Issue #1991 "Possible CSVFile bug with high number of feature vectors"- https://github.com/shogun-toolbox/shogun/issues/1991 | 04:10 |
@sukey | Issue #2824 "Use an external library for reading files"- https://github.com/shogun-toolbox/shogun/issues/2824 | 04:10 |
@sukey | Issue #3018 "SWIG lua does not support overloaded methods"- https://github.com/shogun-toolbox/shogun/issues/3018 | 04:10 |
@sukey | Issue #2569 "Clean up CFile interface"- https://github.com/shogun-toolbox/shogun/issues/2569 | 04:10 |
@sukey | Issue #2824 "Use an external library for reading files"- https://github.com/shogun-toolbox/shogun/issues/2824 | 04:10 |
@sukey | Issue #2824 "Use an external library for reading files"- https://github.com/shogun-toolbox/shogun/issues/2824 | 04:10 |
@sukey | Issue #1991 "Possible CSVFile bug with high number of feature vectors"- https://github.com/shogun-toolbox/shogun/issues/1991 | 04:10 |
@sukey | Issue #3018 "SWIG lua does not support overloaded methods"- https://github.com/shogun-toolbox/shogun/issues/3018 | 04:10 |
@sukey | Issue #1991 "Possible CSVFile bug with high number of feature vectors"- https://github.com/shogun-toolbox/shogun/issues/1991 | 04:10 |
@sukey | Issue #2569 "Clean up CFile interface"- https://github.com/shogun-toolbox/shogun/issues/2569 | 04:10 |
@sukey | Issue #2569 "Clean up CFile interface"- https://github.com/shogun-toolbox/shogun/issues/2569 | 04:10 |
@sukey | Issue #3131 "Create nightly binary packages for Fedora"- https://github.com/shogun-toolbox/shogun/issues/3131 | 04:10 |
@sukey | Issue #3131 "Create nightly binary packages for Fedora"- https://github.com/shogun-toolbox/shogun/issues/3131 | 04:10 |
@sukey | Issue #3018 "SWIG lua does not support overloaded methods"- https://github.com/shogun-toolbox/shogun/issues/3018 | 04:10 |
@sukey | Issue #2569 "Clean up CFile interface"- https://github.com/shogun-toolbox/shogun/issues/2569 | 04:10 |
@sukey | Issue #3131 "Create nightly binary packages for Fedora"- https://github.com/shogun-toolbox/shogun/issues/3131 | 04:10 |
@sukey | Issue #3131 "Create nightly binary packages for Fedora"- https://github.com/shogun-toolbox/shogun/issues/3131 | 04:10 |
@sukey | Issue #3448 "CStringFeatures<char>::clone() fails"- https://github.com/shogun-toolbox/shogun/issues/3448 | 04:10 |
@sukey | Issue #3448 "CStringFeatures<char>::clone() fails"- https://github.com/shogun-toolbox/shogun/issues/3448 | 04:10 |
@sukey | Issue #3018 "SWIG lua does not support overloaded methods"- https://github.com/shogun-toolbox/shogun/issues/3018 | 04:10 |
@sukey | Issue #2824 "Use an external library for reading files"- https://github.com/shogun-toolbox/shogun/issues/2824 | 04:10 |
@sukey | Issue #1991 "Possible CSVFile bug with high number of feature vectors"- https://github.com/shogun-toolbox/shogun/issues/1991 | 04:10 |
@sukey | Issue #3448 "CStringFeatures<char>::clone() fails"- https://github.com/shogun-toolbox/shogun/issues/3448 | 04:10 |
@sukey | Issue #3448 "CStringFeatures<char>::clone() fails"- https://github.com/shogun-toolbox/shogun/issues/3448 | 04:10 |
@sukey | Issue #3452 "C# has no bool vector/matrix typemap"- https://github.com/shogun-toolbox/shogun/issues/3452 | 04:10 |
@sukey | Issue #3452 "C# has no bool vector/matrix typemap"- https://github.com/shogun-toolbox/shogun/issues/3452 | 04:10 |
@sukey | Issue #3452 "C# has no bool vector/matrix typemap"- https://github.com/shogun-toolbox/shogun/issues/3452 | 04:10 |
@sukey | Issue #3452 "C# has no bool vector/matrix typemap"- https://github.com/shogun-toolbox/shogun/issues/3452 | 04:10 |
@sukey | Issue #3467 "Add scalar cases to WrappedObjectArray"- https://github.com/shogun-toolbox/shogun/issues/3467 | 04:10 |
@sukey | Issue #3466 "Fix and enable R meta examples (and tests)"- https://github.com/shogun-toolbox/shogun/issues/3466 | 04:10 |
@sukey | Issue #3466 "Fix and enable R meta examples (and tests)"- https://github.com/shogun-toolbox/shogun/issues/3466 | 04:10 |
@sukey | Issue #3466 "Fix and enable R meta examples (and tests)"- https://github.com/shogun-toolbox/shogun/issues/3466 | 04:10 |
@sukey | Issue #3466 "Fix and enable R meta examples (and tests)"- https://github.com/shogun-toolbox/shogun/issues/3466 | 04:10 |
@sukey | Issue #3467 "Add scalar cases to WrappedObjectArray"- https://github.com/shogun-toolbox/shogun/issues/3467 | 04:10 |
@sukey | Issue #3467 "Add scalar cases to WrappedObjectArray"- https://github.com/shogun-toolbox/shogun/issues/3467 | 04:10 |
@sukey | Issue #3467 "Add scalar cases to WrappedObjectArray"- https://github.com/shogun-toolbox/shogun/issues/3467 | 04:10 |
@sukey | Issue #3485 "add scala interface to docs and tests"- https://github.com/shogun-toolbox/shogun/issues/3485 | 04:10 |
@sukey | Issue #3485 "add scala interface to docs and tests"- https://github.com/shogun-toolbox/shogun/issues/3485 | 04:10 |
@sukey | Issue #3486 "eigen3 operator bug workaround fix"- https://github.com/shogun-toolbox/shogun/issues/3486 | 04:10 |
@sukey | Issue #3485 "add scala interface to docs and tests"- https://github.com/shogun-toolbox/shogun/issues/3485 | 04:10 |
@sukey | Issue #3485 "add scala interface to docs and tests"- https://github.com/shogun-toolbox/shogun/issues/3485 | 04:10 |
@sukey | Issue #3486 "eigen3 operator bug workaround fix"- https://github.com/shogun-toolbox/shogun/issues/3486 | 04:10 |
@sukey | Issue #3489 "fix modular interface make install with custom install prefix"- https://github.com/shogun-toolbox/shogun/issues/3489 | 04:10 |
@sukey | Issue #3486 "eigen3 operator bug workaround fix"- https://github.com/shogun-toolbox/shogun/issues/3486 | 04:10 |
@sukey | Issue #3486 "eigen3 operator bug workaround fix"- https://github.com/shogun-toolbox/shogun/issues/3486 | 04:10 |
@sukey | Issue #3489 "fix modular interface make install with custom install prefix"- https://github.com/shogun-toolbox/shogun/issues/3489 | 04:10 |
@sukey | Issue #3489 "fix modular interface make install with custom install prefix"- https://github.com/shogun-toolbox/shogun/issues/3489 | 04:10 |
@sukey | Issue #3489 "fix modular interface make install with custom install prefix"- https://github.com/shogun-toolbox/shogun/issues/3489 | 04:10 |
@sukey | Issue #3497 "make cookbook fails on guarded meta examples"- https://github.com/shogun-toolbox/shogun/issues/3497 | 04:10 |
@sukey | Issue #3497 "make cookbook fails on guarded meta examples"- https://github.com/shogun-toolbox/shogun/issues/3497 | 04:10 |
@sukey | Issue #3497 "make cookbook fails on guarded meta examples"- https://github.com/shogun-toolbox/shogun/issues/3497 | 04:10 |
@sukey | Issue #3497 "make cookbook fails on guarded meta examples"- https://github.com/shogun-toolbox/shogun/issues/3497 | 04:10 |
@sukey | Issue #3500 "integrate linalg gsoc2016 work"- https://github.com/shogun-toolbox/shogun/issues/3500 | 04:10 |
@sukey | Issue #3500 "integrate linalg gsoc2016 work"- https://github.com/shogun-toolbox/shogun/issues/3500 | 04:10 |
@sukey | Issue #3500 "integrate linalg gsoc2016 work"- https://github.com/shogun-toolbox/shogun/issues/3500 | 04:10 |
@sukey | Issue #3500 "integrate linalg gsoc2016 work"- https://github.com/shogun-toolbox/shogun/issues/3500 | 04:10 |
@sukey | Issue #3501 "merge all gsoc2016 PRs"- https://github.com/shogun-toolbox/shogun/issues/3501 | 04:10 |
@sukey | Issue #3501 "merge all gsoc2016 PRs"- https://github.com/shogun-toolbox/shogun/issues/3501 | 04:10 |
@sukey | Issue #3501 "merge all gsoc2016 PRs"- https://github.com/shogun-toolbox/shogun/issues/3501 | 04:11 |
@sukey | Issue #3503 "Resolve all swig warnings"- https://github.com/shogun-toolbox/shogun/issues/3503 | 04:11 |
@sukey | Issue #3501 "merge all gsoc2016 PRs"- https://github.com/shogun-toolbox/shogun/issues/3501 | 04:11 |
@sukey | Issue #3503 "Resolve all swig warnings"- https://github.com/shogun-toolbox/shogun/issues/3503 | 04:11 |
@sukey | Issue #3503 "Resolve all swig warnings"- https://github.com/shogun-toolbox/shogun/issues/3503 | 04:11 |
@sukey | Issue #3503 "Resolve all swig warnings"- https://github.com/shogun-toolbox/shogun/issues/3503 | 04:11 |
@sukey | Issue #3504 "fix kernel normalisers for GaussianKernel"- https://github.com/shogun-toolbox/shogun/issues/3504 | 04:11 |
@sukey | Issue #3504 "fix kernel normalisers for GaussianKernel"- https://github.com/shogun-toolbox/shogun/issues/3504 | 04:11 |
@sukey | Issue #3505 "fix all ipython notebooks"- https://github.com/shogun-toolbox/shogun/issues/3505 | 04:11 |
@sukey | Issue #3505 "fix all ipython notebooks"- https://github.com/shogun-toolbox/shogun/issues/3505 | 04:11 |
@sukey | Issue #3504 "fix kernel normalisers for GaussianKernel"- https://github.com/shogun-toolbox/shogun/issues/3504 | 04:11 |
@sukey | Issue #3510 "Investigate and fix docstrings generated by swig"- https://github.com/shogun-toolbox/shogun/issues/3510 | 04:11 |
@sukey | Issue #3504 "fix kernel normalisers for GaussianKernel"- https://github.com/shogun-toolbox/shogun/issues/3504 | 04:11 |
@sukey | Issue #3505 "fix all ipython notebooks"- https://github.com/shogun-toolbox/shogun/issues/3505 | 04:11 |
@sukey | Issue #3505 "fix all ipython notebooks"- https://github.com/shogun-toolbox/shogun/issues/3505 | 04:11 |
@sukey | Issue #3510 "Investigate and fix docstrings generated by swig"- https://github.com/shogun-toolbox/shogun/issues/3510 | 04:11 |
@sukey | Issue #3510 "Investigate and fix docstrings generated by swig"- https://github.com/shogun-toolbox/shogun/issues/3510 | 04:11 |
@sukey | Issue #3517 "clean up Notebook description"- https://github.com/shogun-toolbox/shogun/issues/3517 | 04:11 |
@sukey | Issue #3510 "Investigate and fix docstrings generated by swig"- https://github.com/shogun-toolbox/shogun/issues/3510 | 04:11 |
@sukey | Issue #3517 "clean up Notebook description"- https://github.com/shogun-toolbox/shogun/issues/3517 | 04:11 |
@sukey | Issue #3517 "clean up Notebook description"- https://github.com/shogun-toolbox/shogun/issues/3517 | 04:11 |
@sukey | Issue #3517 "clean up Notebook description"- https://github.com/shogun-toolbox/shogun/issues/3517 | 04:11 |
@sukey | Issue #3519 "write a NOTEBOOKS.md readme"- https://github.com/shogun-toolbox/shogun/issues/3519 | 04:11 |
@sukey | Issue #3519 "write a NOTEBOOKS.md readme"- https://github.com/shogun-toolbox/shogun/issues/3519 | 04:11 |
@sukey | Issue #3519 "write a NOTEBOOKS.md readme"- https://github.com/shogun-toolbox/shogun/issues/3519 | 04:11 |
@sukey | Issue #3522 "locate broken links and format conversion errors in notebooks"- https://github.com/shogun-toolbox/shogun/issues/3522 | 04:11 |
@sukey | Issue #3519 "write a NOTEBOOKS.md readme"- https://github.com/shogun-toolbox/shogun/issues/3519 | 04:11 |
@sukey | Issue #3522 "locate broken links and format conversion errors in notebooks"- https://github.com/shogun-toolbox/shogun/issues/3522 | 04:11 |
@sukey | Issue #3522 "locate broken links and format conversion errors in notebooks"- https://github.com/shogun-toolbox/shogun/issues/3522 | 04:11 |
@sukey | Issue #3522 "locate broken links and format conversion errors in notebooks"- https://github.com/shogun-toolbox/shogun/issues/3522 | 04:11 |
@sukey | Issue #3537 "Systematically test serialization of trained models"- https://github.com/shogun-toolbox/shogun/issues/3537 | 04:11 |
@sukey | Issue #3537 "Systematically test serialization of trained models"- https://github.com/shogun-toolbox/shogun/issues/3537 | 04:11 |
@sukey | Issue #3537 "Systematically test serialization of trained models"- https://github.com/shogun-toolbox/shogun/issues/3537 | 04:11 |
@sukey | Issue #3548 "Make `meta_examples` make target depend on generator jsons"- https://github.com/shogun-toolbox/shogun/issues/3548 | 04:11 |
@sukey | Issue #3548 "Make `meta_examples` make target depend on generator jsons"- https://github.com/shogun-toolbox/shogun/issues/3548 | 04:11 |
@sukey | Issue #3548 "Make `meta_examples` make target depend on generator jsons"- https://github.com/shogun-toolbox/shogun/issues/3548 | 04:11 |
@sukey | Issue #3537 "Systematically test serialization of trained models"- https://github.com/shogun-toolbox/shogun/issues/3537 | 04:11 |
@sukey | Issue #3548 "Make `meta_examples` make target depend on generator jsons"- https://github.com/shogun-toolbox/shogun/issues/3548 | 04:11 |
@sukey | Issue #3549 "changes in SWIG interface requires make clean"- https://github.com/shogun-toolbox/shogun/issues/3549 | 04:11 |
@sukey | Issue #3549 "changes in SWIG interface requires make clean"- https://github.com/shogun-toolbox/shogun/issues/3549 | 04:11 |
@sukey | Issue #3549 "changes in SWIG interface requires make clean"- https://github.com/shogun-toolbox/shogun/issues/3549 | 04:11 |
@sukey | Issue #3549 "changes in SWIG interface requires make clean"- https://github.com/shogun-toolbox/shogun/issues/3549 | 04:11 |
@sukey | Issue #3552 "Octave interface doesnt have boolean arrays"- https://github.com/shogun-toolbox/shogun/issues/3552 | 04:11 |
@sukey | Issue #3552 "Octave interface doesnt have boolean arrays"- https://github.com/shogun-toolbox/shogun/issues/3552 | 04:11 |
@sukey | Issue #3552 "Octave interface doesnt have boolean arrays"- https://github.com/shogun-toolbox/shogun/issues/3552 | 04:11 |
@sukey | Issue #3553 "Fix meta example when ´-DENABLE_TESTING=OFF`"- https://github.com/shogun-toolbox/shogun/issues/3553 | 04:11 |
@sukey | Issue #3553 "Fix meta example when ´-DENABLE_TESTING=OFF`"- https://github.com/shogun-toolbox/shogun/issues/3553 | 04:11 |
@sukey | Issue #3580 "Add integration testing data for all uncovered meta examples"- https://github.com/shogun-toolbox/shogun/issues/3580 | 04:11 |
@sukey | Issue #3553 "Fix meta example when ´-DENABLE_TESTING=OFF`"- https://github.com/shogun-toolbox/shogun/issues/3553 | 04:11 |
@sukey | Issue #3553 "Fix meta example when ´-DENABLE_TESTING=OFF`"- https://github.com/shogun-toolbox/shogun/issues/3553 | 04:11 |
@sukey | Issue #3580 "Add integration testing data for all uncovered meta examples"- https://github.com/shogun-toolbox/shogun/issues/3580 | 04:11 |
@sukey | Issue #3580 "Add integration testing data for all uncovered meta examples"- https://github.com/shogun-toolbox/shogun/issues/3580 | 04:11 |
@sukey | Issue #3580 "Add integration testing data for all uncovered meta examples"- https://github.com/shogun-toolbox/shogun/issues/3580 | 04:11 |
@sukey | Issue #3596 "make doc fails when building from release-tarball…"- https://github.com/shogun-toolbox/shogun/issues/3596 | 04:11 |
@sukey | Issue #3596 "make doc fails when building from release-tarball…"- https://github.com/shogun-toolbox/shogun/issues/3596 | 04:11 |
@sukey | Issue #3596 "make doc fails when building from release-tarball…"- https://github.com/shogun-toolbox/shogun/issues/3596 | 04:11 |
@sukey | Issue #3612 "Stratified Cross Validation with Combined Kernels using Custom Kernels"- https://github.com/shogun-toolbox/shogun/issues/3612 | 04:11 |
@sukey | Issue #3596 "make doc fails when building from release-tarball…"- https://github.com/shogun-toolbox/shogun/issues/3596 | 04:11 |
@sukey | Issue #3612 "Stratified Cross Validation with Combined Kernels using Custom Kernels"- https://github.com/shogun-toolbox/shogun/issues/3612 | 04:11 |
@sukey | Issue #3612 "Stratified Cross Validation with Combined Kernels using Custom Kernels"- https://github.com/shogun-toolbox/shogun/issues/3612 | 04:11 |
@sukey | Issue #3672 "octave-modular build failed again"- https://github.com/shogun-toolbox/shogun/issues/3672 | 04:12 |
@sukey | Issue #3672 "octave-modular build failed again"- https://github.com/shogun-toolbox/shogun/issues/3672 | 04:12 |
@sukey | Issue #3672 "octave-modular build failed again"- https://github.com/shogun-toolbox/shogun/issues/3672 | 04:12 |
@sukey | Issue #3612 "Stratified Cross Validation with Combined Kernels using Custom Kernels"- https://github.com/shogun-toolbox/shogun/issues/3612 | 04:12 |
@sukey | Issue #3672 "octave-modular build failed again"- https://github.com/shogun-toolbox/shogun/issues/3672 | 04:12 |
@sukey | Issue #3683 "Memcheck KMeans"- https://github.com/shogun-toolbox/shogun/issues/3683 | 04:12 |
@sukey | Issue #3683 "Memcheck KMeans"- https://github.com/shogun-toolbox/shogun/issues/3683 | 04:12 |
@sukey | Issue #3683 "Memcheck KMeans"- https://github.com/shogun-toolbox/shogun/issues/3683 | 04:12 |
@sukey | Issue #3683 "Memcheck KMeans"- https://github.com/shogun-toolbox/shogun/issues/3683 | 04:12 |
@sukey | Issue #3697 "replace relative paths to data dir in notebooks"- https://github.com/shogun-toolbox/shogun/issues/3697 | 04:12 |
@sukey | Issue #3668 "CrossValidation_multithread unit test fails"- https://github.com/shogun-toolbox/shogun/issues/3668 | 04:12 |
mikeling | wiking: alright, let's back to the starting pointer of that issue, let's assume we only care about query SHOGUN_DATA_DIR on notebooks, do i need to add that variable in travis or other ci tool as a part of that issue? | 04:12 |
mikeling | or I just need make some change to my environment variable | 04:12 |
@wiking | no, because it'll have some default value | 04:12 |
@wiking | i mean like if you would just do this change in the notbooks | 04:13 |
@wiking | then you would do something like | 04:13 |
mikeling | ok | 04:13 |
mikeling | I see | 04:13 |
@wiking | os.env.get("SHOGUN_DATA_PATH", "../../../data") | 04:13 |
@wiking | this way if the env doesn't have it set | 04:13 |
@wiking | then it's probably in a place where it is by default | 04:13 |
@wiking | and then things should work | 04:13 |
mikeling | make sense ;) | 04:13 |
@wiking | if they dont work | 04:13 |
@wiking | then you set the env var | 04:13 |
@wiking | to make it work | 04:13 |
@wiking | but again | 04:14 |
@sukey | New Commit "fix buildbot warnings - SGVector" to shogun-toolbox/shogun by OXPHOS: https://github.com/shogun-toolbox/shogun/commit/6400d84731663153aea6fabcfdd1148d2fe610e1 | 04:14 |
@wiking | if we create a helper thing | 04:14 |
mikeling | BTW, what's these issues for? | 04:14 |
mikeling | been kee | 04:14 |
@wiking | mikeling, when we deploy shogun in cluod | 04:14 |
@wiking | *cloud | 04:14 |
mikeling | * keeping pope up | 04:14 |
@wiking | and people use the notebooks | 04:14 |
@wiking | we currently have to follow a specific | 04:14 |
@wiking | directory structure | 04:14 |
@wiking | that all notebook assumes | 04:14 |
@wiking | which is shit | 04:15 |
@wiking | because you cannot copy notebook around | 04:15 |
@wiking | etc | 04:15 |
mikeling | yeah, I know. However, I just wondering about the 'issues' been reported by 'sukey' . Why so many issues been popped up all at once in irc? | 04:17 |
mikeling | what's these issues for? | 04:17 |
mikeling | wiking: ^ | 04:20 |
OXPHOS | mikeling: "vigsterkr modified the milestone: Shogun 5.1, Shogun 6.0 11 minutes ago" | 04:21 |
OXPHOS | I think that's it | 04:21 |
mikeling | OXPHOS: ah | 04:21 |
mikeling | thank you | 04:21 |
OXPHOS | np | 04:21 |
@wiking | mikeling, moved the mileston | 04:22 |
@wiking | e | 04:22 |
@wiking | sukey, is talking a lot | 04:22 |
@wiking | :) | 04:22 |
@wiking | sukey: help | 04:22 |
@sukey | wiking: ship it - Display a motivation squirrel | 04:22 |
@sukey | sukey adapter - Reply with the adapter | 04:22 |
@sukey | sukey animate me <query> - The same thing as `image me`, except adds a few parameters to try to return an animated GIF instead. | 04:22 |
@sukey | sukey calculate <expression> - Calculate the given math expression. | 04:22 |
@sukey | sukey convert <expression> in <units> - Convert expression to given units. | 04:22 |
@sukey | sukey echo <text> - Reply back with <text> | 04:22 |
@sukey | sukey flip - Hubot flips a table | 04:22 |
@sukey | sukey help - Displays all of the help commands that Hubot knows about. | 04:22 |
@sukey | sukey help <query> - Displays all help commands that match <query>. | 04:22 |
@sukey | sukey image me <query> - The Original. Queries Google Images for <query> and returns a random top result. | 04:22 |
@sukey | sukey map me <query> - Returns a map view of the area returned by `query`. | 04:22 |
@sukey | sukey mustache me <url|query> - Adds a mustache to the specified URL or query result. | 04:22 |
@sukey | sukey ping - Reply with pong | 04:22 |
@sukey | sukey pug bomb N - get N pugs | 04:22 |
@sukey | sukey pug me - Receive a pug | 04:22 |
@sukey | sukey tell <recipients> <some message> - tell <recipients> <some message> next time they are present. | 04:22 |
@sukey | sukey the rules - Make sure hubot still knows the rules. | 04:22 |
@sukey | sukey time - Reply with current time | 04:22 |
@sukey | sukey translate me <phrase> - Searches for a translation for the <phrase> and then prints that bad boy out. | 04:22 |
mikeling | wiking: the GSoC page seems like went wrong somehow | 04:22 |
@sukey | sukey translate me from <source> into <target> <phrase> - Translates <phrase> from <source> into <target>. Both <source> and <target> are optional | 04:23 |
@wiking | sukey animate me chatty | 04:23 |
@sukey | Google Image Search API is not longer available. Please [setup up Custom Search Engine API](https://github.com/hubot-scripts/hubot-google-images#cse-setup-details). | 04:23 |
@sukey | http://i.imgur.com/CzFTOkI.png | 04:23 |
mikeling | lost the link to the projects page | 04:23 |
@wiking | heheh poor sukey is dumb | 04:23 |
@wiking | mikeling, where? | 04:23 |
mikeling | wiking: https://github.com/shogun-toolbox/shogun/wiki lost the link to projects page | 04:23 |
@wiking | [2017 projects](Google Summer of Code 2017 Projects) | 04:23 |
@wiking | who changed this the last time | 04:23 |
@sukey | New Commit "change viennacl matrix_operations header to matrix header | 04:24 |
@sukey | Attempt to fix buildbot error #3" to shogun-toolbox/shogun by OXPHOS: https://github.com/shogun-toolbox/shogun/commit/20f03478e8713ff14cb08396bb6a878297e7e943 | 04:24 |
@wiking | wtf | 04:25 |
@wiking | when and how the fuck did this happen | 04:25 |
mikeling | wiking: it's good yesterday I guess. I just check it before | 04:26 |
@wiking | i need to see the history | 04:27 |
@wiking | mmm | 04:28 |
@wiking | i think there was an update in the backend | 04:28 |
@wiking | because nothing changed in our stuff | 04:28 |
@wiking | since november | 04:28 |
@sukey | Wiki page: Home edited on shogun-toolbox/shogun by vigsterkr | 04:28 |
@wiking | now it should be ok | 04:29 |
-!- nagdev [dce39546@gateway/web/freenode/ip.220.227.149.70] has quit [Ping timeout: 260 seconds] | 04:30 | |
mikeling | wiking: thank you! | 04:31 |
mikeling | how many student we are going accept for Shogun detox 2 project? | 04:31 |
@wiking | dunno | 04:32 |
mikeling | If I found anything worth to improve and haven't been listed | 04:32 |
mikeling | could I add it to the proposal? | 04:33 |
@wiking | yes yes | 04:37 |
@wiking | these are just suggestions | 04:37 |
@wiking | in your proposal | 04:37 |
@wiking | you can put anything | 04:37 |
@wiking | of course you can send us proposals | 04:37 |
@wiking | prior the applications | 04:37 |
@wiking | so we can give feedback etc. | 04:37 |
-!- goksinen [~goksinen@2604:2000:c591:8400:d1ed:7b48:76cf:bb15] has quit [Remote host closed the connection] | 04:38 | |
@sukey | New Commit "change viennacl matrix_operations header to matrix header | 04:39 |
@sukey | Attempt to fix buildbot error #3" to shogun-toolbox/shogun by OXPHOS: https://github.com/shogun-toolbox/shogun/commit/cdbf2e840502fd70974fa6f98a9165a2d9444571 | 04:39 |
mikeling | wiking: do we have template for proposal? like what's the necessary thing I should say in propsal | 04:40 |
-!- travis-ci [~travis-ci@ec2-23-22-249-67.compute-1.amazonaws.com] has joined #shogun | 04:47 | |
travis-ci | it's Viktor Gal'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/211196814 | 04:47 |
-!- travis-ci [~travis-ci@ec2-23-22-249-67.compute-1.amazonaws.com] has left #shogun [] | 04:47 | |
@wiking | mikeling, good uestion | 04:52 |
@wiking | i think we used to | 04:52 |
@sukey | Pull Request #3690 "Remove apply_locked* methods from SWIG interfaces #3687" synchronized by tdjogi010 - https://github.com/shogun-toolbox/shogun/pull/3690 | 05:02 |
@wiking | OXPHOS, big error | 05:18 |
-!- lambday [6a3311dc@gateway/web/freenode/ip.106.51.17.220] has joined #shogun | 05:18 | |
-!- mode/#shogun [+o lambday] by ChanServ | 05:18 | |
@lambday | wiking: yo | 05:18 |
@wiking | OXPHOS, /home/buildslave/trusty_-_libshogun_-_viennacl/build/src/shogun/mathematics/linalg/LinalgBackendViennaCL.h:298:62: required from 'void shogun::LinalgBackendViennaCL::matrix_prod_impl(shogun::SGMatrix<T>&, shogun::SGMatrix<T>&, shogun::SGMatrix<T>&, bool, bool) const [with T = unsigned int]' | 05:18 |
@wiking | /home/buildslave/trusty_-_libshogun_-_viennacl/build/src/shogun/mathematics/linalg/LinalgBackendViennaCL.h:111:1199: required from here | 05:18 |
@wiking | /usr/include/viennacl/scheduler/forwards.h:670:69: error: call of overloaded 'add_element(viennacl::vcl_size_t&, viennacl::scheduler::lhs_rhs_element&, const unsigned int&)' is ambiguous | 05:18 |
@wiking | return add_element(next_free, array_[current_index].rhs, t); | 05:18 |
@wiking | ^ | 05:18 |
@wiking | /usr/include/viennacl/scheduler/forwards.h:670:69: note: candidates are: | 05:18 |
@wiking | /usr/include/viennacl/scheduler/forwards.h:511:20: note: viennacl::vcl_size_t viennacl::scheduler::statement::add_element(viennacl::vcl_size_t, viennacl::scheduler::lhs_rhs_element&, const float&) | 05:18 |
@wiking | vcl_size_t add_element(vcl_size_t next_free, | 05:18 |
@wiking | ^ | 05:19 |
@wiking | /usr/include/viennacl/scheduler/forwards.h:522:20: note: viennacl::vcl_size_t viennacl::scheduler::statement::add_element(viennacl::vcl_size_t, viennacl::scheduler::lhs_rhs_element&, const double&) | 05:19 |
@wiking | vcl_size_t add_element(vcl_size_t next_free, | 05:19 |
@wiking | lambday, y0 | 05:19 |
@wiking | got the email | 05:19 |
@wiking | thnx heaps | 05:19 |
@lambday | wiking: cool... was those the results you needed? | 05:22 |
@wiking | yep | 05:23 |
@wiking | sounds good | 05:23 |
@lambday | wiking: presenting shogun somewhere? | 05:24 |
@wiking | lambday, http://2017.fossasia.org/tracks.html#2843 | 05:26 |
@wiking | fuck | 05:27 |
@wiking | they moved me! | 05:27 |
@wiking | :<<<< | 05:27 |
@wiking | we used to be on the sunday session | 05:28 |
@wiking | *saturday session | 05:28 |
@lambday | nice.. you have one more day :) | 05:35 |
@lambday | soooo many indian guys | 05:35 |
@lambday | apart from you and 2 others, everyone in the AI track are from india :/ | 05:36 |
@wiking | lambday, ehhe the saturday session was much better | 05:36 |
@wiking | ;( | 05:36 |
@wiking | would have been much better to stay there | 05:36 |
@wiking | wonder for whom i got ditched | 05:36 |
@lambday | drumpf? | 05:36 |
@lambday | :D | 05:36 |
@lambday | btw rebasing the matrix patch.. I agree, doing these kind of changes in a feature branch is much more effective | 05:37 |
@sukey | Pull Request #3685 "SGMatrix cleanups and speedups" synchronized by lambday - https://github.com/shogun-toolbox/shogun/pull/3685 | 05:38 |
@wiking | lambday, in this case https://github.com/shogun-toolbox/shogun/pull/3685/files#diff-2def2de5f7f35b7c1f0e40a4f3110abaR353 | 05:45 |
@wiking | how the resolving of namespace is going to happen | 05:45 |
@wiking | as there are couple of memcpy functions :) | 05:46 |
@lambday | ah crap.. | 05:47 |
@lambday | but wait.. | 05:48 |
@lambday | in many other places we simply have memcpy uses with using shogun namespace.... what happens to those | 05:48 |
@lambday | in fact, I noticed the modular interface complaining about my memcpy not returning anything.. so I changed it to void*... | 05:49 |
@lambday | maybe an entrance task? | 05:49 |
@lambday | first of all, no using namespace shogun please... | 05:49 |
@lambday | using shogun::stuffs instead | 05:49 |
@lambday | what do you say? | 05:50 |
@lambday | secondly, no #include <string.h> | 05:53 |
@lambday | with <cstring>, we can differentiate between std::memcpy and shogun::memcpy | 05:55 |
@lambday | wiking: ^ | 05:55 |
@wiking | yeah | 05:56 |
@wiking | but i mean what happens when | 05:56 |
@wiking | using namespace std; | 05:56 |
@wiking | :) | 05:56 |
@lambday | yeah we're screwed, unless we specify shogun::memcpy... | 05:57 |
@wiking | so yeah | 05:57 |
@wiking | i would explicitly | 05:57 |
@wiking | oooooor | 05:57 |
@lambday | but point is, we should not be using `using namespace std;` | 05:58 |
@wiking | could we do the trick? | 05:58 |
@wiking | that we do for | 05:58 |
@wiking | new operator? | 05:58 |
@lambday | wiking: overloading? | 05:58 |
@wiking | yeah | 05:58 |
@wiking | in memory.h we do that right? | 05:58 |
@wiking | lambday, i mean about "we should not be using `using namespace std;`" | 06:00 |
@wiking | this we cannot assure | 06:00 |
@wiking | there's always gonna be somebody | 06:00 |
@wiking | whos' gonna come with a patch | 06:00 |
@wiking | :) | 06:00 |
@wiking | and there's always gonna be somebody who's gonna be tired enough | 06:01 |
@wiking | not to catch that in a PR to merge it | 06:01 |
-!- lambday [6a3311dc@gateway/web/freenode/ip.106.51.17.220] has quit [Ping timeout: 260 seconds] | 06:03 | |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has joined #shogun | 06:19 | |
travis-ci | it's Pan Deng / Zora'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/211210581 | 06:19 |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has left #shogun [] | 06:19 | |
-!- lambday|work [c40f1706@gateway/web/freenode/ip.196.15.23.6] has joined #shogun | 06:27 | |
-!- mode/#shogun [+o lambday|work] by ChanServ | 06:27 | |
@lambday|work | wiking: sorry I got disconnected.. can't see the log as it gives internal service error.. | 06:47 |
@lambday|work | wiking: what were you saying regarding new operator ? | 06:47 |
@wiking | well | 06:48 |
@wiking | that we override that right? | 06:48 |
@wiking | in memory.h/cpp | 06:48 |
@wiking | so that the new operator is actually using shogun's malloc | 06:48 |
@wiking | so this way you can actually use a different memory backend consistently within shogun | 06:49 |
@lambday|work | wiking: ummm.. I am not sure you can do that for normal functions.. they will have the same signature, right? works for operator new and delete but not sure it will work for memcpy | 06:58 |
@wiking | yeah that's what i thoguth as well | 07:00 |
@wiking | *thought | 07:00 |
@wiking | i mean the only way then then | 07:00 |
@wiking | is to actually change the name of ours | 07:01 |
@wiking | because really | 07:01 |
@wiking | we dont wanna rely | 07:01 |
@wiking | on people being concious here | 07:01 |
@wiking | i mean we can do a wrapper | 07:01 |
@wiking | like | 07:01 |
@wiking | shogun::memcpy | 07:01 |
@wiking | but that ctually calls something like | 07:01 |
@wiking | sg_memcpy | 07:01 |
@lambday|work | if not, however, I think a systematic approach will be to (a) replace #include <string.h> with #include <cstring> (b) remove `using namespace std;`s in shogun, IIRC there aren't many (c) grep for memcpy (d) replace with shogun::memcpy | 07:01 |
@wiking | or someting like thist | 07:01 |
@wiking | so that in case you want | 07:01 |
@wiking | you can explicitly call that function | 07:01 |
@wiking | yeah but do you see what i mean | 07:02 |
@wiking | you kind of rely | 07:02 |
@wiking | on people being | 07:02 |
@wiking | super concious there | 07:02 |
@wiking | *conscious | 07:02 |
@wiking | i mean yeah it's not a big deal | 07:02 |
@wiking | but think about external PRs | 07:02 |
@wiking | and then just one lazy merge | 07:02 |
@wiking | and you dont even realise | 07:03 |
@wiking | because there's no way one would realise this right? | 07:03 |
@wiking | other than actually grepping th ecode | 07:03 |
@wiking | and if we use explicitly an sg_memcpy | 07:03 |
@wiking | or something similar | 07:03 |
@wiking | then we can regex grep for std::memcpy or simple memcpy | 07:03 |
@wiking | that way we can see where things go south | 07:03 |
-!- lambday|work_ [c40f1706@gateway/web/freenode/ip.196.15.23.6] has joined #shogun | 07:04 | |
-!- mode/#shogun [+o lambday|work_] by ChanServ | 07:04 | |
@lambday|work_ | wiking: umm so we should name it sg_memcpy.. no wrappers cause that creates the same problem when someone uses both shogun and std namespaces | 07:04 |
@lambday|work_ | then the entrance task is to grep all memcpy and replace it with sg_memcpy | 07:04 |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has joined #shogun | 07:05 | |
travis-ci | it's Pan Deng / Zora'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/211212836 | 07:05 |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has left #shogun [] | 07:05 | |
@lambday|work_ | using namespace std is anyway bad.. and we should discourage such code to be a part of shogun, just saying.. | 07:06 |
-!- lambday|work [c40f1706@gateway/web/freenode/ip.196.15.23.6] has quit [Ping timeout: 260 seconds] | 07:06 | |
-!- lambday|work_ is now known as lambday|work | 07:06 | |
@wiking | lambday|work_, yeah something like that | 07:06 |
@wiking | lambday|work, yeah i totally agree with you | 07:06 |
@wiking | i'm just seeing the past things | 07:07 |
@wiking | and you know as well | 07:07 |
@wiking | that when we merge stuff | 07:07 |
@wiking | sometimes things go through | 07:07 |
@lambday|work | yeah... true that | 07:07 |
@lambday|work | wiking: cool.. I'll create an entrance task for that then.. | 07:08 |
@wiking | tbhnx | 07:08 |
@wiking | *thnx | 07:08 |
@sukey | Issue #3700 "Refactoring for `shogun::memcpy`" lambday added label: "Cleanups" - https://github.com/shogun-toolbox/shogun/issues/3700 | 07:19 |
@sukey | Issue #3700 "Refactoring for `shogun::memcpy`" opened by lambday - https://github.com/shogun-toolbox/shogun/issues/3700 | 07:19 |
@sukey | Issue #3700 "Refactoring for `shogun::memcpy`" lambday added label: "entrance" - https://github.com/shogun-toolbox/shogun/issues/3700 | 07:19 |
mikeling | wiking: the metadata of each cell will been changed to '"editable": true...', how could I avoid that? At least avoid include these changes during I make a pr to github | 07:26 |
@wiking | edit it with a simple text editor :) | 07:26 |
mikeling | wiking: But it means I need to edit every ipython notebook file for https://github.com/shogun-toolbox/shogun/issues/3697issue :/ | 07:28 |
@wiking | mikeling, exactly | 07:28 |
@wiking | this is why it's an entrance task | 07:28 |
@wiking | :) | 07:28 |
@wiking | it's simple | 07:29 |
mikeling | alright...I see | 07:29 |
@wiking | even people w/o coding skills can do it | 07:29 |
mikeling | wiking: I would like to write more unit test for shogun. The first place I want to look into is modules under the sum folder | 08:23 |
mikeling | Is that ok? | 08:23 |
@wiking | sum folder? | 08:23 |
mikeling | Could I file an issue for mark my progress ? | 08:23 |
@wiking | you mean svm? | 08:23 |
mikeling | svm | 08:23 |
mikeling | yep | 08:23 |
@wiking | its' ok | 08:23 |
mikeling | typo ;/ | 08:23 |
@wiking | no need for an issue | 08:23 |
@wiking | just start working | 08:23 |
@wiking | once you have something to show | 08:23 |
@wiking | just send the PR | 08:23 |
@wiking | we track these things ourselves dont worry ;) | 08:24 |
mikeling | ok, I see | 08:24 |
mikeling | thanks | 08:24 |
@wiking | nw | 08:24 |
mikeling | where could I read shogun cookbook? | 08:51 |
mikeling | Because I'm wondering what's the 'svm_file' looks like and how to create one | 08:52 |
@wiking | http://stats.stackexchange.com/questions/61328/libsvm-data-format | 08:57 |
@sukey | Pull Request #3701 "use get and set in SVM class" opened by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3701 | 09:05 |
mikeling | wiking: mmm, I think the link is about the form of training data. But the svm_file in here https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/classifier/svm/SVM.cpp#L86 means the configuration of svm | 09:10 |
mikeling | I just don't know what's that file should looks like | 09:10 |
shogun-buildbot | build #41 of nightly_all is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_all/builds/41 blamelist: lambday <heavensdevil6909@gmail.com>, Pan Deng / Zora <engelzora@gmail.com>, Pan Deng / Zora <OXPHOS@users.noreply.github.com>, Viktor Gal <vigsterkr@gmail.com>, Soumyajit De <heavensdevil6909@gmail.com>, | 09:11 |
shogun-buildbot | Heiko Strathmann <heiko.strathmann@gmail.com>, abhinavrai44 <abhinavrai44@gmail.com> | 09:11 |
-!- micmn [4e0d90fe@gateway/web/freenode/ip.78.13.144.254] has joined #shogun | 09:16 | |
@sukey | Pull Request #3685 "SGMatrix cleanups and speedups" merged by lambday - https://github.com/shogun-toolbox/shogun/pull/3685 | 09:33 |
@sukey | New Commit "Merge pull request #3685 from lambday/feature/sgmatrix | 09:33 |
@sukey | SGMatrix cleanups and speedups" to shogun-toolbox/shogun by lambday: https://github.com/shogun-toolbox/shogun/commit/0968ee6206ed88011a7fe2cf1c1ad302cb63989c | 09:33 |
-!- iRmantou [adefc6ce@gateway/web/freenode/ip.173.239.198.206] has joined #shogun | 09:46 | |
-!- goksinen [~goksinen@2604:2000:c591:8400:d1ed:7b48:76cf:bb15] has joined #shogun | 10:06 | |
-!- goksinen [~goksinen@2604:2000:c591:8400:d1ed:7b48:76cf:bb15] has quit [Ping timeout: 246 seconds] | 10:10 | |
-!- micmn2 [~micmn@5.90.109.157] has joined #shogun | 10:12 | |
-!- micmn2 [~micmn@5.90.109.157] has quit [Client Quit] | 10:12 | |
-!- micmn [4e0d90fe@gateway/web/freenode/ip.78.13.144.254] has quit [Quit: Page closed] | 10:12 | |
-!- micmn [~micmn@5.90.109.157] has joined #shogun | 10:12 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 10:16 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 10:16 | |
@wiking | HeikoS, mastah | 10:16 |
@HeikoS | wiking: jojo | 10:17 |
@HeikoS | man | 10:17 |
@HeikoS | this wiki thing | 10:17 |
lisitsyn | hey | 10:17 |
@HeikoS | phews! | 10:17 |
@sukey | Pull Request #3666 "added check before ref_count method to avoid buildbot failure" - https://github.com/shogun-toolbox/shogun/pull/3666 | 10:26 |
-!- micmn [~micmn@5.90.109.157] has quit [Ping timeout: 258 seconds] | 10:28 | |
@sukey | Wiki page: Home edited on shogun-toolbox/shogun by karlnapf | 10:29 |
@sukey | Wiki page: _Sidebar edited on shogun-toolbox/shogun by karlnapf | 10:29 |
@sukey | Issue #3702 "Broken links in notebooks." opened by kirstymcnaught - https://github.com/shogun-toolbox/shogun/issues/3702 | 10:33 |
@sukey | Issue #3683 "Memcheck KMeans" closed by lambday - https://github.com/shogun-toolbox/shogun/issues/3683 | 10:36 |
@sukey | Issue #2719 "Make C++11 support mandatory for Shogun 4.x.x" closed by lambday - https://github.com/shogun-toolbox/shogun/issues/2719 | 10:39 |
@sukey | Issue #3499 "merge bigtest branch" closed by lambday - https://github.com/shogun-toolbox/shogun/issues/3499 | 10:44 |
@sukey | Issue #2919 "fully support C++11 ?" closed by lambday - https://github.com/shogun-toolbox/shogun/issues/2919 | 10:45 |
-!- travis-ci [~travis-ci@ec2-54-198-186-122.compute-1.amazonaws.com] has joined #shogun | 10:46 | |
travis-ci | it's Soumyajit De'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/211265223 | 10:46 |
-!- travis-ci [~travis-ci@ec2-54-198-186-122.compute-1.amazonaws.com] has left #shogun [] | 10:46 | |
@sukey | Issue #2527 "Add linear solves to linalg framework" assigned to: OXPHOS by lambday - https://github.com/shogun-toolbox/shogun/issues/2527 | 10:47 |
@sukey | Issue #2527 "Add linear solves to linalg framework" unassigned lambday by lambday - https://github.com/shogun-toolbox/shogun/issues/2527 | 10:47 |
@sukey | Issue #2526 "Add matrix factorisations to linalg" assigned to: OXPHOS by lambday - https://github.com/shogun-toolbox/shogun/issues/2526 | 10:48 |
@sukey | Issue #2526 "Add matrix factorisations to linalg" unassigned lambday by lambday - https://github.com/shogun-toolbox/shogun/issues/2526 | 10:48 |
-!- goksinen [~goksinen@2604:2000:c591:8400:d1ed:7b48:76cf:bb15] has joined #shogun | 11:00 | |
-!- goksinen [~goksinen@2604:2000:c591:8400:d1ed:7b48:76cf:bb15] has quit [Ping timeout: 246 seconds] | 11:05 | |
-!- geektoni [~geektoni@93-34-234-212.ip52.fastwebnet.it] has joined #shogun | 11:30 | |
@sukey | Wiki page: GSoC_2017_applications_webdemos edited on shogun-toolbox/shogun by karlnapf | 11:39 |
@sukey | Pull Request #3699 "Minor fix for cookbook" merged by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3699 | 11:48 |
@sukey | New Commit "Merge pull request #3699 from abhinavrai44/fix | 11:48 |
@sukey | Minor fix for cookbook" to shogun-toolbox/shogun by karlnapf: https://github.com/shogun-toolbox/shogun/commit/fd67ee5647402828e242f968bf0092301d5af7e6 | 11:48 |
@sukey | Issue #3500 "integrate linalg gsoc2016 work" closed by karlnapf - https://github.com/shogun-toolbox/shogun/issues/3500 | 11:51 |
@HeikoS | lisitsyn: jo | 11:53 |
lisitsyn | hey | 11:54 |
-!- goksinen [~goksinen@2604:2000:c591:8400:d1ed:7b48:76cf:bb15] has joined #shogun | 11:55 | |
@sukey | Pull Request #3690 "Remove apply_locked* methods from SWIG interfaces #3687" merged by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3690 | 11:56 |
@sukey | New Commit "Merge pull request #3690 from tdjogi010/clean_up_swig_interface | 11:56 |
@sukey | Remove apply_locked* methods from SWIG interfaces #3687" to shogun-toolbox/shogun by karlnapf: https://github.com/shogun-toolbox/shogun/commit/05777e03b03b4764abe00f47f31cfdf4408b380e | 11:56 |
@sukey | Issue #3687 "Remove `apply_locked*` methods from SWIG interfaces" closed by karlnapf - https://github.com/shogun-toolbox/shogun/issues/3687 | 11:57 |
lisitsyn | HeikoS: hey | 11:57 |
-!- goksinen [~goksinen@2604:2000:c591:8400:d1ed:7b48:76cf:bb15] has quit [Ping timeout: 246 seconds] | 11:59 | |
@sukey | Issue #3702 "Broken links in notebooks." karlnapf added label: "entrance" - https://github.com/shogun-toolbox/shogun/issues/3702 | 12:00 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 260 seconds] | 12:20 | |
-!- geektoni [~geektoni@93-34-234-212.ip52.fastwebnet.it] has left #shogun [] | 12:21 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 12:22 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:22 | |
@HeikoS | lisitsyn: sorry | 12:23 |
@HeikoS | now again | 12:23 |
@HeikoS | you there? | 12:23 |
@HeikoS | lisitsyn: so about tags, the way I understand, we can also register member variables in there, no? | 12:24 |
@HeikoS | wiking: jo | 12:25 |
@HeikoS | Check file MockObject.h | 12:25 |
@HeikoS | line 26 | 12:26 |
@HeikoS | registers a member variable as a tag | 12:26 |
@HeikoS | and SGOBject_unittest.cc covers this | 12:26 |
@HeikoS | SGObject::register_param | 12:26 |
-!- iRmantou [adefc6ce@gateway/web/freenode/ip.173.239.198.206] has quit [Ping timeout: 260 seconds] | 12:26 | |
@HeikoS | http://shogun.ml/api/latest/classshogun_1_1CSGObject.html#ac0fdc76cf24d242d99d8d862475131a6 | 12:27 |
@HeikoS | There is no need to remove all class variables imo | 12:27 |
@HeikoS | but lisitsyn needs to comment | 12:27 |
@sukey | Issue #3703 "Automate coobkook version" karlnapf added label: "Documentation" - https://github.com/shogun-toolbox/shogun/issues/3703 | 12:30 |
@sukey | Issue #3703 "Automate coobkook version" karlnapf added label: "entrance" - https://github.com/shogun-toolbox/shogun/issues/3703 | 12:30 |
@sukey | Issue #3703 "Automate coobkook version" opened by karlnapf - https://github.com/shogun-toolbox/shogun/issues/3703 | 12:30 |
@sukey | Issue #3703 "Automate coobkook version" karlnapf added label: "cmake" - https://github.com/shogun-toolbox/shogun/issues/3703 | 12:30 |
@sukey | Issue #3703 "Automate coobkook version"- https://github.com/shogun-toolbox/shogun/issues/3703 | 12:30 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 260 seconds] | 12:45 | |
-!- travis-ci [~travis-ci@ec2-54-198-186-122.compute-1.amazonaws.com] has joined #shogun | 12:45 | |
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/211300688 | 12:45 |
-!- travis-ci [~travis-ci@ec2-54-198-186-122.compute-1.amazonaws.com] has left #shogun [] | 12:45 | |
lisitsyn | HeikoS: yes we can | 13:17 |
-!- nagdev [dce39546@gateway/web/freenode/ip.220.227.149.70] has joined #shogun | 13:20 | |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has joined #shogun | 13:31 | |
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/211303071 | 13:31 |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has left #shogun [] | 13:31 | |
@sukey | Pull Request #3701 "use get and set in SVM class" synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3701 | 13:35 |
mikeling | lisitsyn: hi, for this comment https://github.com/shogun-toolbox/shogun/pull/3698#discussion_r106023708 how about rename it to 'esp' | 13:43 |
lisitsyn | mikeling: what about 'the_epsilon'? | 13:44 |
lisitsyn | :D | 13:44 |
mikeling | lisitsyn: got it ;) | 13:44 |
-!- lambday|work [c40f1706@gateway/web/freenode/ip.196.15.23.6] has quit [Quit: Page closed] | 14:07 | |
OXPHOS | wiking: any idea where I could find the version info about viennacl on buildbot? I cannot even find the line with error in viennacl | 14:23 |
@wiking | it's whatever you have in | 14:23 |
@wiking | trusty by default | 14:23 |
@wiking | it's a 1.5.x | 14:23 |
OXPHOS | thx | 14:24 |
OXPHOS | i'm confused | 14:25 |
@wiking | OXPHOS, http://packages.ubuntu.com/trusty/libviennacl-dev | 14:25 |
@wiking | that's the one installed there | 14:25 |
OXPHOS | but my gut told me it's another header issue | 14:25 |
OXPHOS | thank you | 14:25 |
@wiking | again the easiest is | 14:25 |
OXPHOS | guard it? | 14:25 |
@wiking | that you start off a docker image | 14:25 |
@wiking | but | 14:25 |
@wiking | again | 14:25 |
OXPHOS | yep.. | 14:26 |
@wiking | sum is not existing | 14:26 |
@wiking | :) | 14:26 |
@wiking | you want me to grep? | 14:26 |
OXPHOS | If you have time, for sure. It's gonna be easier | 14:26 |
@wiking | /usr/include/viennacl/linalg/cuda/vector_operations.hpp: // sums the array 'vec1' and writes to result. Makes use of a single work-group only. | 14:26 |
@wiking | i seen this | 14:27 |
@wiking | // sums the array 'vec1' and writes to result. Makes use of a single work-group only. | 14:28 |
@wiking | template <typename T> | 14:28 |
@wiking | __global__ void vector_sum_kernel_floats( | 14:28 |
@wiking | you need a sum for vector/ | 14:28 |
@wiking | right? | 14:28 |
@sukey | Pull Request #3698 "use get and set in LibLinear class" synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3698 | 14:29 |
@wiking | OXPHOS, interestingly | 14:29 |
@wiking | this only exists | 14:29 |
@wiking | under | 14:29 |
@wiking | /usr/include/viennacl/linalg/cuda/vector_operations.hpp | 14:29 |
mikeling | lisitsyn: I just update that pr ;) | 14:29 |
@wiking | OXPHOS, want me to grep anything else? | 14:29 |
mikeling | please take a look at it after the ci build turn green | 14:30 |
mikeling | thank yop | 14:30 |
mikeling | u | 14:30 |
OXPHOS | wiking: there's more problem? yes please | 14:30 |
@wiking | OXPHOS, what problems i mean | 14:30 |
OXPHOS | if yes, please | 14:30 |
@wiking | the patch you sent | 14:31 |
@wiking | made everything worse :) | 14:31 |
OXPHOS | why : ( | 14:31 |
@wiking | w/o that sum was missing | 14:31 |
@wiking | and that's all | 14:31 |
@wiking | ) | 14:31 |
@wiking | :) | 14:31 |
OXPHOS | wiking: so, the sum should work before version 1.7? because I cast it back to matrix sum | 14:32 |
@wiking | OXPHOS, should it? | 14:32 |
@wiking | dunno | 14:32 |
@wiking | that patch should do it? | 14:32 |
OXPHOS | wiking: that sum patch should do it | 14:33 |
@wiking | ok lemme see it again | 14:33 |
OXPHOS | but i am using viennacl::linalg::sum(vector) for version > 1.7 | 14:33 |
@wiking | mmm | 14:34 |
@wiking | wait | 14:34 |
@wiking | i think | 14:34 |
@wiking | i'm missing stuff from patch | 14:34 |
@wiking | because i only have this in the patch | 14:34 |
@wiking | #include <viennacl/vector.hpp> | 14:34 |
@wiking | -#include <viennacl/linalg/inner_prod.hpp> | 14:34 |
@wiking | #include <viennacl/matrix.hpp> | 14:34 |
@wiking | +#include <viennacl/linalg/inner_prod.hpp> | 14:34 |
@wiking | +#include <viennacl/linalg/prod.hpp> | 14:34 |
@wiking | #include <shogun/mathematics/linalg/GPUMemoryViennaCL.h> | 14:34 |
@wiking | is that all? | 14:34 |
OXPHOS | wiking: ohh I pushed something to dev directly | 14:35 |
OXPHOS | i guarded sum | 14:35 |
@wiking | mmm | 14:35 |
@wiking | where is it | 14:35 |
@wiking | ? | 14:35 |
@wiking | :) | 14:35 |
@wiking | ok lemme check | 14:35 |
@wiking | wait | 14:35 |
@wiking | OXPHOS, can you give me the sha? | 14:36 |
OXPHOS | 0151f26bb42d0307998f42872cce29f4f36838ad | 14:36 |
OXPHOS | wiking: and cdbf2e840502fd70974fa6f98a9165a2d9444571 | 14:37 |
OXPHOS | you'll need it later | 14:37 |
@wiking | ok i see now | 14:38 |
@wiking | but ok lemme see wtf is happening here | 14:38 |
OXPHOS | I need to do some bench work. but I will be around. | 14:39 |
@wiking | ok now i see | 14:40 |
@wiking | now it's the unit tests | 14:40 |
@wiking | that are making the shits | 14:40 |
-!- HeikoS [~heiko@eduroam-int-pat-8-248.ucl.ac.uk] has joined #shogun | 14:52 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 14:52 | |
-!- lambday [6a3314a2@gateway/web/freenode/ip.106.51.20.162] has joined #shogun | 14:57 | |
-!- mode/#shogun [+o lambday] by ChanServ | 14:57 | |
-!- HeikoS [~heiko@eduroam-int-pat-8-248.ucl.ac.uk] has quit [Remote host closed the connection] | 15:01 | |
@sukey | Issue #3108 "add Cholesky update/downdate to linalg" closed by lambday - https://github.com/shogun-toolbox/shogun/issues/3108 | 15:03 |
OXPHOS | wiking: what is the new error now? :) | 15:07 |
@wiking | mmm same | 15:08 |
@wiking | as in buildbot | 15:08 |
OXPHOS | wiking: I was hoping the "prod" error would be solved by the last patch i sent u yesterday | 15:09 |
@wiking | looooooool | 15:10 |
@wiking | http://gee.cs.oswego.edu/dl/html/j9mm.html | 15:10 |
@wiking | lisitsyn, ^ jdk9 copies c++11 :) | 15:10 |
@wiking | OXPHOS, - | 15:10 |
@wiking | that one | 15:11 |
@wiking | OXPHOS, http://pastebin.com/F4QrfMNr | 15:13 |
OXPHOS | bummer! | 15:15 |
-!- HeikoS [~heiko@eduroam-int-pat-8-248.ucl.ac.uk] has joined #shogun | 15:33 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 15:33 | |
@wiking | lisitsyn, y0 | 15:37 |
@wiking | lisitsyn, so i had some time now to write some of the servables for serving .... but | 15:37 |
@wiking | for the case you wanted | 15:38 |
@wiking | lisitsyn, do you basically i guess just want to be able to do TFRegression(Features) | 15:39 |
@wiking | right? | 15:39 |
@sukey | Issue #3108 "add Cholesky update/downdate to linalg" reopened by karlnapf - https://github.com/shogun-toolbox/shogun/issues/3108 | 15:39 |
@HeikoS | lambday: ^ | 15:39 |
@lambday | HeikoS: yoo | 15:40 |
@HeikoS | lambday: jojo | 15:40 |
@HeikoS | think the up/downdates are not there yet | 15:40 |
lisitsyn | wiking: I am not sure I get it | 15:40 |
@wiking | lisitsyn, ok so you wanted tf_Serving | 15:41 |
@lambday | HeikoS: huh? :/ | 15:41 |
@lambday | oh that was for sergey | 15:41 |
@wiking | to be able to use tf models from shogun | 15:41 |
@wiking | that's what we gathered the other time | 15:41 |
@wiking | becuase i wanted the other way round | 15:41 |
@wiking | so i started to implement the shogun servables | 15:41 |
@wiking | that is bascially to be able to serve shogun models in tf_serving | 15:42 |
@wiking | but | 15:42 |
lisitsyn | wiking: ok so TFRegression serves some shogun linear model? | 15:42 |
@HeikoS | wiking: I am trying to build ubuntu on trusty .... | 15:42 |
@wiking | lisitsyn, no TFRegression is actually serving a TF model for regression | 15:42 |
@wiking | HeikoS, ubuntu on trusty? what does that mena? :) | 15:42 |
@wiking | *mean | 15:42 |
@HeikoS | my workstation is trusty | 15:42 |
@HeikoS | and I would like to use it | 15:42 |
lisitsyn | wiking: gotit | 15:42 |
@HeikoS | since it has 12 cores | 15:43 |
@wiking | lisitsyn, so am i right that you woudl like to use | 15:43 |
@HeikoS | but I guess I should say goodbye to that ;) | 15:43 |
@wiking | TF models from shogun namespace | 15:43 |
@wiking | i..e. that to pass a Features obj | 15:43 |
@wiking | and get classification/regression | 15:43 |
lisitsyn | yes | 15:43 |
@wiking | in the right fromat | 15:43 |
@wiking | okok | 15:43 |
lisitsyn | I think that's cool | 15:43 |
@wiking | that's even easier | 15:43 |
@wiking | than the stuff i'm doing | 15:43 |
@HeikoS | wait we have trusty on buildbot ... | 15:44 |
@HeikoS | mmh | 15:44 |
@wiking | HeikoS, yes we do | 15:44 |
@wiking | HeikoS, all the errors are there :) | 15:44 |
@HeikoS | /nfs/nhome/live/ucabhst/git/shogun/shogun_develop/shogun/src/shogun/statistical_testing/kernelselection/internals/OptimizationSolver.cpp:35:26: error: ‘libqp_state_T’ has not been declared | 15:44 |
@HeikoS | static void print_state(libqp_state_T state); | 15:44 |
@HeikoS | ^ | 15:44 |
@HeikoS | currently get this | 15:44 |
@HeikoS | guards ... | 15:44 |
@wiking | but with what? | 15:44 |
@wiking | i mean man | 15:44 |
@wiking | i'm out of context | 15:44 |
@HeikoS | sorry | 15:44 |
@wiking | r u trying to debug | 15:44 |
@wiking | for OXPHOS ? | 15:44 |
@HeikoS | just trying to compile shogun on my workstation and it doesnt | 15:44 |
@wiking | because for that make sure you have libvienncl-dev installed | 15:44 |
@HeikoS | just for my research | 15:44 |
@wiking | HeikoS, ah | 15:44 |
@HeikoS | no vienna | 15:44 |
@wiking | HeikoS, r u sure you are using | 15:45 |
@wiking | latest develop? | 15:45 |
@HeikoS | 5b126ce90ed6d1f7706e1f25c69b8b5922c73fc2 | 15:45 |
@wiking | shogun-buildbot: force build 'trusty - libshogun' | 15:45 |
shogun-buildbot | build forced [ETA 4m09s] | 15:45 |
shogun-buildbot | I'll give a shout when the build finishes | 15:45 |
@HeikoS | yeah | 15:45 |
@wiking | HeikoS, ^ lets see :) | 15:45 |
@HeikoS | no its all green | 15:45 |
@HeikoS | checked the page | 15:45 |
@wiking | HeikoS, it's tricky | 15:45 |
@wiking | HeikoS, it's green because it wasn't build | 15:45 |
@wiking | *built since | 15:45 |
@wiking | OXPHOS, merged the shit | 15:46 |
@HeikoS | ah i am before that | 15:46 |
@HeikoS | I am last week or so | 15:46 |
@HeikoS | feature branch that I spawned then | 15:46 |
@HeikoS | /nfs/nhome/live/ucabhst/git/shogun/shogun_develop/shogun/src/shogun/classifier/svm/SVMLight.cpp: In member function ‘float64_t* shogun::CSVMLight::optimize_qp(shogun::QP*, float64_t*, int32_t, float64_t*, int32_t&)’: | 15:46 |
@HeikoS | /nfs/nhome/live/ucabhst/git/shogun/shogun_develop/shogun/src/shogun/classifier/svm/SVMLight.cpp:2444:9: error: ‘STILL_RUNNING’ was not declared in this scope | 15:46 |
@HeikoS | result=STILL_RUNNING; | 15:46 |
@wiking | mmm then what can i do/ | 15:46 |
@wiking | ? | 15:46 |
@wiking | :) | 15:46 |
@HeikoS | just asking whether you can make sense of these errors | 15:46 |
@wiking | mm never seen them | 15:46 |
@HeikoS | I mean that is guard errors | 15:47 |
@HeikoS | I am puzzled | 15:47 |
@HeikoS | well yeah nevermind I will check | 15:47 |
@wiking | i guess yeah | 15:47 |
@wiking | u have smoe weird shitto with some ifdefs | 15:47 |
@HeikoS | I had to install custom cmake | 15:47 |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has joined #shogun | 15:47 | |
@HeikoS | maybe its that | 15:47 |
@wiking | HeikoS, sure not | 15:47 |
@wiking | ) | 15:47 |
@wiking | if cmake runs | 15:47 |
@wiking | then it must be something else | 15:48 |
@wiking | btw that shogun buildbot | 15:48 |
@wiking | trusty | 15:48 |
@wiking | has a custom cmake as well | 15:48 |
@wiking | obviously | 15:48 |
@HeikoS | yeah | 15:48 |
@HeikoS | Ill check | 15:48 |
@wiking | it'll be another 10 mins | 15:48 |
@wiking | and u'll get your answer | 15:48 |
@wiking | from shogun-buildbot | 15:48 |
@wiking | HeikoS, man | 15:51 |
@wiking | just a sidenote | 15:51 |
@wiking | HeikoS, you know that when we launch things with jupyterhub and you have your own volume... it takes 60s to get the notebook | 15:52 |
@wiking | so now i'm testing some stuff... and now launching manually the very same type of container | 15:52 |
@wiking | but not via jupyterhub, but manually | 15:52 |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has quit [Ping timeout: 260 seconds] | 15:52 | |
@wiking | HeikoS, the container is being created and mounted with all the fucking volumes within 2s | 15:53 |
@wiking | ... | 15:53 |
@wiking | so this fucking jupyterhub | 15:53 |
@wiking | i'm getting really fucking angry here... | 15:53 |
@HeikoS | ah | 15:53 |
@HeikoS | is it worth reporting that back and help them fixing it? | 15:54 |
@wiking | HeikoS, dunno | 15:54 |
@wiking | i've tried reaching out | 15:54 |
@wiking | nobody answered | 15:54 |
@wiking | but man | 15:54 |
@sukey | Pull Request #3666 "Remove FeaturesUtil class, use CFeatures methods instead" synchronized by lambday - https://github.com/shogun-toolbox/shogun/pull/3666 | 15:54 |
@wiking | wtf is this | 15:54 |
@wiking | :D | 15:54 |
@HeikoS | when did you ask? | 15:54 |
@wiking | days ago | 15:54 |
@wiking | :) | 15:54 |
@HeikoS | weird | 15:55 |
@wiking | like reaching out :) | 15:55 |
@wiking | nvm | 15:55 |
@wiking | i'm looking into shit | 15:55 |
@wiking | and once i have the root cause | 15:55 |
@wiking | i'll fire an issue | 15:55 |
shogun-buildbot | build #113 of trusty - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun/builds/113 | 15:55 |
@wiking | on gh | 15:55 |
@wiking | HeikoS, ^ | 15:55 |
@wiking | mmm i wonder | 15:55 |
@HeikoS | wiking: mmh | 15:55 |
@wiking | why the fuck did it pick up viennacl | 15:55 |
@wiking | :D | 15:55 |
@wiking | seems i need to do shiat with the cmake flags | 15:56 |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has joined #shogun | 15:57 | |
@wiking | to actually | 15:59 |
@wiking | mmm | 15:59 |
@wiking | fuck | 15:59 |
@wiking | lisitsyn, so the best demo for that module could be a word2vec model's output being fed to a shogun model | 15:59 |
lisitsyn | or keras | 16:09 |
lisitsyn | exported keras model run within shogun | 16:09 |
-!- catwoman [ca8e6c9b@gateway/web/freenode/ip.202.142.108.155] has joined #shogun | 16:11 | |
@wiking | lisitsyn, you mean keras to build tf model that u can use in shogun | 16:11 |
@wiking | :) | 16:11 |
@wiking | or? | 16:12 |
lisitsyn | yes | 16:12 |
@wiking | yeye sure | 16:12 |
lisitsyn | train in keras | 16:12 |
lisitsyn | export | 16:12 |
@wiking | i'm just | 16:12 |
@wiking | wondering | 16:12 |
@wiking | what example | 16:12 |
@wiking | i should put together | 16:12 |
@wiking | :) | 16:12 |
@wiking | sure sure | 16:12 |
@wiking | that is rather easy | 16:12 |
@wiking | the only part is swig :P | 16:13 |
catwoman | hello everyone! I'm Neha and i'm a pre final year student of B.Tech(ICE). and i was looking forward to participate in GSoC'17 as a student of shogun. I looked at the issues on the github repo. How can i get the open issues to contribute in!? Thanks in advance. | 16:13 |
@wiking | catwoman, read the wiki pages how to contribut | 16:14 |
@wiking | and get started | 16:14 |
@wiking | with shogun | 16:14 |
@HeikoS | lambday: https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/statistical_testing/kernelselection/internals/OptimizationSolver.cpp#L21 | 16:18 |
@HeikoS | whats that? | 16:18 |
@wiking | lisitsyn, have u seen that http://openjdk.java.net/jeps/193 | 16:19 |
lisitsyn | wiking: too low level | 16:19 |
@wiking | lisitsyn, :D | 16:19 |
lisitsyn | :D | 16:19 |
@wiking | HeikoS, that is my friend style | 16:19 |
@wiking | :) | 16:19 |
@wiking | LIKE A BOSS | 16:19 |
@wiking | :>>> | 16:19 |
@HeikoS | break buils if GPL is off | 16:20 |
lisitsyn | I've seen some of my code is GPLed | 16:21 |
@wiking | HeikoS, B) | 16:21 |
lisitsyn | https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/transfer/multitask/MultitaskClusteredLogisticRegression.cpp | 16:21 |
lisitsyn | like this | 16:21 |
@HeikoS | lisitsyn: yes | 16:21 |
@HeikoS | feel free to change it :D | 16:21 |
lisitsyn | alright | 16:21 |
@HeikoS | BUT the functionality is not available anyways when GPL is away | 16:22 |
@HeikoS | since malsar is not available | 16:22 |
@wiking | lisitsyn, i dont know who's the bigger boss here :) | 16:22 |
lisitsyn | HeikoS: I am not sure what's the malsar status | 16:22 |
lisitsyn | IIRC I asked for gpl | 16:23 |
@sukey | Issue #3704 "USE_GPL_SHOGUN=OFF build error" opened by karlnapf - https://github.com/shogun-toolbox/shogun/issues/3704 | 16:23 |
@sukey | Issue #3704 "USE_GPL_SHOGUN=OFF build error" karlnapf added label: "BUG" - https://github.com/shogun-toolbox/shogun/issues/3704 | 16:23 |
@sukey | Issue #3704 "USE_GPL_SHOGUN=OFF build error" karlnapf added label: "cmake" - https://github.com/shogun-toolbox/shogun/issues/3704 | 16:23 |
lisitsyn | and they said ok if you acknowledge | 16:23 |
@sukey | Issue #3704 "USE_GPL_SHOGUN=OFF build error" assigned to: lambday by karlnapf - https://github.com/shogun-toolbox/shogun/issues/3704 | 16:23 |
lisitsyn | but its rewrite of code | 16:23 |
lisitsyn | it was in matlab | 16:23 |
@HeikoS | would be cool to get that | 16:23 |
@HeikoS | hurrah | 16:23 |
@HeikoS | managed to compile | 16:23 |
@HeikoS | on trusty | 16:23 |
@HeikoS | kind of a nightmare | 16:23 |
@HeikoS | ok next swig | 16:24 |
@wiking | HeikoS, just because | 16:24 |
@wiking | u r hipster and want gpl off | 16:24 |
@wiking | ;) | 16:24 |
lisitsyn | HeikoS: my subordinate at job is going to implement factorized machine in shogun | 16:24 |
lisitsyn | :D | 16:24 |
@wiking | :> | 16:24 |
@HeikoS | no i wanted to compile less and had errors in GPL classes :D | 16:24 |
lisitsyn | factorization sorry | 16:24 |
@wiking | :)))))))))))))))) | 16:24 |
@HeikoS | lisitsyn: lol nice | 16:24 |
@wiking | HeikoS, will he contribute the code? :) | 16:25 |
@wiking | * lisitsyn ^ | 16:25 |
@wiking | or just keep it internal ?:> | 16:25 |
lisitsyn | wiking: yes contribute as a student project | 16:25 |
@wiking | \o/ | 16:25 |
@wiking | perfecto | 16:25 |
@wiking | HeikoS, btw it came out | 16:26 |
@wiking | HeikoS, that fucking dcos doesn't do a real virtalization of resources | 16:27 |
@wiking | HeikoS, that number i give for cpus | 16:27 |
@wiking | it's just a weight | 16:27 |
@wiking | ..... | 16:27 |
@HeikoS | ................................................. | 16:27 |
@wiking | yep | 16:27 |
@HeikoS | whoooooooooooo | 16:27 |
@wiking | so fuck that as well | 16:28 |
@wiking | so wait | 16:29 |
@wiking | now that we have a code of conduct | 16:29 |
@wiking | me swearing is basically gonna get me on a blacklist | 16:29 |
@wiking | ? :D | 16:29 |
@HeikoS | yep | 16:31 |
@HeikoS | I already reported a few instances | 16:31 |
@HeikoS | .... | 16:31 |
@HeikoS | to myself | 16:31 |
@wiking | :> | 16:31 |
@HeikoS | Could NOT find PythonLibs (missing: PYTHON_LIBRARIES) (found suitable exact | 16:34 |
@HeikoS | version "2.7.6") | 16:34 |
@HeikoS | ??? | 16:34 |
@wiking | well i guess u have maybe more than one python runtimes? | 16:34 |
@HeikoS | conda shiat | 16:35 |
@wiking | mmm | 16:35 |
@wiking | yeah have fun | 16:35 |
@wiking | :) | 16:35 |
@wiking | HeikoS, i mean what u need to make sure | 16:35 |
@wiking | that the detected | 16:35 |
@wiking | python interpreter | 16:35 |
@wiking | and the detected python lib | 16:35 |
@wiking | coming from the same src | 16:35 |
@wiking | this error is usually trying to circumvent the situaton | 16:35 |
@wiking | when you detect the runtime/interpreter under /usr/bin/python | 16:35 |
@wiking | but it finds the libs of some other python installed somewhere else | 16:36 |
@wiking | with PREFIX stuff | 16:36 |
@wiking | u can actually make sure | 16:36 |
@wiking | that the right python is being found | 16:36 |
@HeikoS | /nfs/nhome/live/ucabhst/miniconda2/bin/python | 16:36 |
@HeikoS | which python | 16:36 |
@HeikoS | mmh | 16:36 |
@HeikoS | ok yeah | 16:36 |
@HeikoS | aha | 16:37 |
@HeikoS | anaconda doesnt set python path | 16:37 |
@HeikoS | wtf? | 16:37 |
@wiking | dunno | 16:38 |
@wiking | dont ask | 16:38 |
@wiking | i dont wanna see python | 16:38 |
@wiking | HeikoS, use javascript man | 16:38 |
@wiking | faster better stronger | 16:38 |
@wiking | :) | 16:38 |
lisitsyn | PUT ON PYTHON | 16:39 |
@sukey | Pull Request #3666 "Remove FeaturesUtil class, use CFeatures methods instead" merged by lambday - https://github.com/shogun-toolbox/shogun/pull/3666 | 16:49 |
@sukey | New Commit "Merge pull request #3666 from lambday/develop | 16:49 |
@sukey | Remove FeaturesUtil class, use CFeatures methods instead" to shogun-toolbox/shogun by lambday: https://github.com/shogun-toolbox/shogun/commit/17907f45de1659752cd8cdc50510962843729a5e | 16:49 |
@HeikoS | ah finally | 16:50 |
@HeikoS | it compiles | 16:50 |
@HeikoS | swig even | 16:50 |
@HeikoS | man we need docker here | 16:51 |
@HeikoS | I am pestering the it people for a while already | 16:51 |
@HeikoS | all this time spent on custom configs | 16:51 |
@HeikoS | nightmare | 16:51 |
@sukey | New branch feature/bsd_license created on shogun-toolbox/shogun | 16:51 |
@sukey | New Commit "changed the license to Shogun style BSD" to shogun-toolbox/shogun by lambday: https://github.com/shogun-toolbox/shogun/commit/fa521cff60a2323c7119e3f32c574ddda1fc7126 | 16:51 |
@lambday | HeikoS: so, shall I keep those #ifdefs on? | 16:51 |
@HeikoS | lambday: ? | 16:52 |
@HeikoS | well the code should compile if GPL is turned off | 16:52 |
@HeikoS | so you cannot include that thing | 16:52 |
@HeikoS | which means you need to guard your code as well | 16:52 |
@lambday | HeikoS: those #ifdefs were disabled for GPL... turning them on again should be fine | 16:52 |
@lambday | yeah code was guarded but for some reason I commented those guards out | 16:52 |
@HeikoS | lambday: no it wont build | 16:53 |
@lambday | HeikoS: yeah.. fixing | 16:53 |
@HeikoS | libqp is not available if GPL is off | 16:53 |
-!- nagdev [dce39546@gateway/web/freenode/ip.220.227.149.70] has quit [Ping timeout: 260 seconds] | 17:06 | |
shogun-buildbot | build #153 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/153 blamelist: Soumyajit De <heavensdevil6909@gmail.com> | 17:08 |
@HeikoS | lisitsyn: you have a minute? | 17:09 |
lisitsyn | HeikoS: forsure | 17:09 |
@HeikoS | ok so I want to ask | 17:09 |
@HeikoS | why get rid of the member variables again? | 17:09 |
@HeikoS | why not just register them? | 17:09 |
@HeikoS | so one can directly access | 17:09 |
@wiking | mmm lambday ^ what happened there | 17:09 |
lisitsyn | HeikoS: uhmm for what purpose | 17:10 |
@HeikoS | having a proper c++ lib | 17:10 |
lisitsyn | HeikoS: we want to get rid of set_something/get_something | 17:10 |
lisitsyn | because we want plugins | 17:10 |
@HeikoS | sure I get that | 17:10 |
@HeikoS | all SWIG getters/setters should be that | 17:10 |
@HeikoS | but inside | 17:11 |
lisitsyn | ah so you ask | 17:11 |
lisitsyn | ok | 17:11 |
@HeikoS | the lib itself | 17:11 |
@lambday | wiking: erm.. buildbot failure? | 17:11 |
lisitsyn | well because if we get rid of them | 17:11 |
@wiking | lambday, yes | 17:11 |
lisitsyn | the only thing we will do is register them | 17:11 |
@HeikoS | why not get the speed of having the things as members? | 17:11 |
@lambday | GradientModelSelection.select_model_ep_inference .. mmhh.. on it | 17:11 |
@wiking | lambday, The difference between nlZ and 3.334009 is nan, which exceeds 1E-3, where | 17:11 |
@wiking | nlZ evaluates to nan, | 17:11 |
@wiking | 3.334009 evaluates to 3.334009, and | 17:11 |
@wiking | 1E-3 evaluates to 0.001. | 17:11 |
@lambday | yeah | 17:11 |
lisitsyn | HeikoS: I don't think it is an issue | 17:11 |
@lambday | weird!... | 17:11 |
@lambday | let me open a bug and assign it to myself.. | 17:11 |
@wiking | kk | 17:11 |
@wiking | thnx | 17:12 |
@lambday | nw | 17:12 |
lisitsyn | HeikoS: if you access it multiple times just get it once | 17:12 |
lisitsyn | and store into variable | 17:12 |
@HeikoS | lisitsyn: because I was thinking | 17:12 |
@lambday | wiking: HeikoS BTW the license change patch - can you guys merge it? I sent to a feature branch | 17:12 |
@sukey | Pull Request #3705 "Add SGIO::progress unit tests and enhance SGIO::progress" opened by geektoni - https://github.com/shogun-toolbox/shogun/pull/3705 | 17:12 |
@HeikoS | we should aim to have a proper library where the internals are not dictated by the interfaces | 17:12 |
@HeikoS | lambday: sure | 17:13 |
@HeikoS | lambday: so the license thing wasnt the issue | 17:13 |
lisitsyn | HeikoS: it is good when class uses only its public interface | 17:13 |
@HeikoS | more the build failure | 17:13 |
lisitsyn | it makes things better | 17:13 |
lisitsyn | I believe | 17:13 |
@sukey | Issue #3706 "GradientModelSelection.select_model_ep_inference test fails on buildbot" opened by lambday - https://github.com/shogun-toolbox/shogun/issues/3706 | 17:13 |
@sukey | Issue #3706 "GradientModelSelection.select_model_ep_inference test fails on buildbot" lambday added label: "BUG" - https://github.com/shogun-toolbox/shogun/issues/3706 | 17:13 |
@sukey | Issue #3706 "GradientModelSelection.select_model_ep_inference test fails on buildbot" lambday added label: "bugfixing" - https://github.com/shogun-toolbox/shogun/issues/3706 | 17:13 |
@HeikoS | I agree in some ways | 17:13 |
@sukey | Issue #3706 "GradientModelSelection.select_model_ep_inference test fails on buildbot" assigned to: lambday by lambday - https://github.com/shogun-toolbox/shogun/issues/3706 | 17:13 |
@HeikoS | lisitsyn: I am just quite nerveous about all these get calls | 17:13 |
lisitsyn | HeikoS: I mean if you have setter and getter you never should access directly | 17:13 |
@HeikoS | at places where there wasnt something like this before | 17:13 |
@HeikoS | and the fact that things like basic types are not contiguous in memory anymore | 17:14 |
@HeikoS | but if you just register | 17:14 |
@HeikoS | then you get the getter/setter | 17:14 |
@HeikoS | but still keep the layout | 17:14 |
@lambday | HeikoS: umm what to do you mean? I am talking about this one here : https://github.com/shogun-toolbox/shogun/commit/fa521cff60a2323c7119e3f32c574ddda1fc7126 | 17:14 |
@lambday | HeikoS: for the GPL stuff, sending a new patch | 17:14 |
lisitsyn | HeikoS: I don't think it is an issue again :P I mean *some* libs are in python | 17:15 |
lisitsyn | and accessing self.something is nothing but dictionary lookup | 17:15 |
@wiking | lisitsyn, lets not implement python's way of implementing a class | 17:15 |
@wiking | in c++ | 17:15 |
@wiking | i.e. a fucking hashmap | 17:15 |
@wiking | it's insane :) | 17:15 |
@HeikoS | lisitsyn: what about memory layout of objects? | 17:15 |
lisitsyn | HeikoS: I can't even come up with some example of really bottleneck of field access | 17:16 |
@wiking | ok | 17:16 |
@wiking | how do you do | 17:16 |
@wiking | alignas(64) myStupidField x; | 17:16 |
lisitsyn | HeikoS: one may get_w() set_w() often | 17:17 |
lisitsyn | in linear model | 17:17 |
lisitsyn | but that's easy to spot on code review | 17:17 |
lisitsyn | I mean if someone accesses field too often it would become quite ugly | 17:17 |
@HeikoS | mmmh | 17:18 |
@HeikoS | I mean lambday had this thing where he put a bunch of kernel instances next to each other in memory, which made things faster when calling methods | 17:18 |
lisitsyn | HeikoS: there are other advantages | 17:19 |
@lambday | HeikoS: I would.. haven't tried that yet.. an array of kernels | 17:20 |
@HeikoS | lisitsyn: I just want to discuss a bit | 17:20 |
lisitsyn | HeikoS: if there is no direct access we can do quite interesting things | 17:20 |
@wiking | :D | 17:20 |
@HeikoS | because it is quite a significant step | 17:20 |
lisitsyn | one may come up with some elegant model selection | 17:20 |
@HeikoS | yeah I know | 17:20 |
@HeikoS | there *are* cool things about this | 17:20 |
@HeikoS | just wondering about downsides | 17:20 |
@HeikoS | and also I mean: c++ doesnt do it this way, and there are probably reasons for that?? | 17:21 |
lisitsyn | e.g. one object being employed by multiple threads so these parameters are thread-local | 17:21 |
@wiking | lisitsyn, but if you dont have | 17:21 |
@wiking | control over the layout | 17:21 |
@wiking | you actually can fuckup your cacheline | 17:21 |
@wiking | eight? | 17:21 |
@wiking | *right? | 17:21 |
@lambday | umm getter/setters are hashmap lookups? | 17:22 |
@wiking | lambday, yes | 17:22 |
@HeikoS | only if the member is deleted | 17:22 |
@HeikoS | it can also stay and the tag points to its location as a class member | 17:22 |
@wiking | HeikoS, still the getter/setter is a hashmap lookup even then | 17:23 |
@wiking | :D | 17:23 |
@lambday | lisitsyn is right though.. frequent access of struct members through getters/setters should raise a red flag.. but memory layout should be programmable | 17:23 |
@wiking | lambday, how do u set the right padding? | 17:23 |
@lambday | wiking: yeah that's the concern | 17:23 |
@wiking | or you set an object? :) | 17:24 |
lisitsyn | HeikoS: well it is obvious because additional layer of indirection is always helpful for flexibility :) | 17:24 |
@wiking | and that object inside has it's own layout? :) | 17:24 |
@lambday | wiking: that shouldn't be an issue.. in Shogun objects are on heap.. so it's always sizeof ptr | 17:24 |
@wiking | lambday, no what i mena | 17:24 |
@wiking | that the getter setter | 17:25 |
@wiking | is actually for primitive types | 17:25 |
@wiking | as well | 17:25 |
@lambday | hmm | 17:25 |
@wiking | anyhow | 17:25 |
@wiking | would be cool not to reimplement python | 17:26 |
@wiking | :) | 17:26 |
@lambday | btw, with lisitsyn's stuff, we can pretty much do reflections on Shogun it seems :D | 17:26 |
@wiking | lambday, yeah | 17:26 |
@lambday | although, we should **not** be using reflections | 17:26 |
@HeikoS | mmh | 17:26 |
@wiking | ok guys | 17:26 |
@wiking | have fun | 17:26 |
@wiking | i'm out | 17:26 |
@HeikoS | so we all agree to go ahead with replacing all member access with get/set? | 17:26 |
@HeikoS | wiking: see you! | 17:27 |
@wiking | no | 17:27 |
@HeikoS | wiking: ok good :D | 17:27 |
@wiking | me not at all | 17:27 |
@HeikoS | need to discuss this for hackathon | 17:27 |
@wiking | especially not because of the ridicolous patches | 17:27 |
@lambday | HeikoS: memory layout is a bigger concern rather than getter setters | 17:27 |
@wiking | that will clear off member vars | 17:27 |
@HeikoS | pro/con would be good | 17:27 |
@lambday | they are just one call per method.. they copy it to local inside methods, uses the local instead | 17:27 |
@wiking | i mean if it's just an indirection | 17:27 |
@wiking | idk | 17:27 |
@wiking | but if we really wanna remove the member vars | 17:27 |
@wiking | plz bitch | 17:27 |
@wiking | sukey flip | 17:28 |
@sukey | (ノ ゜Д゜)ノ ︵ ┻━┻ | 17:28 |
@HeikoS | lisitsyn: is there a reason to remove member vars? | 17:29 |
-!- micmn [4e0d90fe@gateway/web/freenode/ip.78.13.144.254] has joined #shogun | 17:33 | |
-!- HeikoS [~heiko@eduroam-int-pat-8-248.ucl.ac.uk] has quit [Remote host closed the connection] | 17:37 | |
-!- travis-ci [~travis-ci@ec2-54-198-186-122.compute-1.amazonaws.com] has joined #shogun | 17:55 | |
travis-ci | it's Soumyajit De'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/211402652 | 17:56 |
-!- travis-ci [~travis-ci@ec2-54-198-186-122.compute-1.amazonaws.com] has left #shogun [] | 17:56 | |
-!- HeikoS [~heiko@eduroam-int-pat-8-248.ucl.ac.uk] has joined #shogun | 18:01 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 18:01 | |
@HeikoS | lambday: jo still there? | 18:02 |
@HeikoS | I started doing this thing which is quite handy: ssh tunnelling the ipython notebook on sperry to my laptop | 18:02 |
@HeikoS | so I get the fast machine but can use my laptop to edit | 18:02 |
@sukey | Pull Request #3707 "Added GPL guard in optimization solver that uses libqp" opened by lambday - https://github.com/shogun-toolbox/shogun/pull/3707 | 18:03 |
@lambday | HeikoS: you're doing it from a different network? | 18:04 |
@HeikoS | yep | 18:04 |
@lambday | sweet! | 18:04 |
@lambday | :D | 18:04 |
@HeikoS | from laptop to gateway to sperry | 18:04 |
@HeikoS | double hop tunneling | 18:04 |
@lambday | HeikoS: when is arthur's course gonna start? | 18:06 |
@HeikoS | which one? | 18:07 |
@lambday | HeikoS: they are the ones using shogun cloud, is it | 18:07 |
@HeikoS | yesterday was a session | 18:07 |
@HeikoS | another one in may | 18:07 |
@lambday | okay | 18:07 |
@HeikoS | for that we could write a testing notebook | 18:07 |
@HeikoS | did a cool session for this one | 18:07 |
@lambday | HeikoS: testing related? | 18:07 |
@HeikoS | https://github.com/karlnapf/machine_learning_course | 18:07 |
@HeikoS | nope | 18:07 |
@HeikoS | classification | 18:07 |
@HeikoS | check the notebook | 18:08 |
@lambday | HeikoS: looks awesome man! | 18:08 |
@HeikoS | lambday: yeah it was nice | 18:13 |
@HeikoS | but can do some testing stuff as well | 18:13 |
@lambday | nice explanation about the overfitting :) | 18:13 |
@lambday | just sent it to one of my friend | 18:14 |
@HeikoS | there are some bugs | 18:14 |
@HeikoS | but generally it is opk | 18:14 |
@HeikoS | lisitsyn: you still there? | 18:14 |
@HeikoS | I have another point to discuss | 18:15 |
@HeikoS | lambday: also with you | 18:15 |
lisitsyn | yes | 18:15 |
@HeikoS | I am annoyed that the libshogun API is so shit | 18:15 |
lisitsyn | rat shit bat shit | 18:15 |
@HeikoS | and that it is so heavily shitted on by the fact that we have swig | 18:15 |
@HeikoS | so I think we should start separating that more | 18:15 |
lisitsyn | ex-pla-in | 18:15 |
lisitsyn | :) | 18:15 |
@lambday | HeikoS: what's the idea? | 18:15 |
@HeikoS | I would like to use interfaces (like java) | 18:16 |
@HeikoS | for example | 18:16 |
lisitsyn | what api | 18:16 |
@HeikoS | c++ API | 18:16 |
OXPHOS | wiking: send u a new patch to fix viennacl | 18:16 |
lisitsyn | well we have no api | 18:16 |
lisitsyn | we have a bunch of classes | 18:16 |
OXPHOS | wiking: thanks | 18:16 |
@HeikoS | lisitsyn: true | 18:17 |
@HeikoS | but for example | 18:17 |
@HeikoS | subsets | 18:17 |
lisitsyn | well as a general rant | 18:17 |
@HeikoS | there are some methods in Features | 18:17 |
lisitsyn | accepted | 18:17 |
lisitsyn | :D | 18:17 |
@HeikoS | and also in Labels | 18:17 |
lisitsyn | everything is shit :P | 18:17 |
@HeikoS | they do the same thing | 18:17 |
@HeikoS | they have the same name | 18:17 |
@HeikoS | so I should be able to pass them around as "subsettable" | 18:17 |
@HeikoS | for e.g. for x-validation | 18:18 |
lisitsyn | so you suggest to use more of interfaces? | 18:18 |
@HeikoS | it is just one example | 18:18 |
@HeikoS | use interfaces to get rid of such things | 18:18 |
@HeikoS | Also in features that happens a lot | 18:18 |
@HeikoS | and kernel | 18:18 |
@HeikoS | GaussianARDKernel | 18:18 |
@HeikoS | SparseGaussianARDKernel | 18:18 |
lisitsyn | just avoid MI | 18:18 |
lisitsyn | :D | 18:18 |
@lambday | what about the case when one class displays multiple of such properties | 18:19 |
@lambday | yeah | 18:19 |
@HeikoS | implements multiple interfaces | 18:19 |
@lambday | that's where I was getting at :( | 18:19 |
@lambday | HeikoS: I'd like to have this as a "have property" thing.. like features do | 18:19 |
@lambday | instead of multiple inherience | 18:20 |
lisitsyn | C++ guys are still working on 'concepts' | 18:20 |
@HeikoS | well this is just one thing | 18:20 |
lisitsyn | but this is not the thing | 18:20 |
@lambday | lisitsyn: concepts won't work here.. those are for static type-constraints if I am not mistaken | 18:20 |
lisitsyn | ya | 18:20 |
@lambday | we're dynamic as hell :D | 18:21 |
lisitsyn | properties are good | 18:21 |
lisitsyn | but they don't add any methods | 18:21 |
@lambday | lisitsyn: yeah... | 18:21 |
@HeikoS | my general point is: it would be cool to do proper software engineering in the libshogun library | 18:21 |
@HeikoS | and then "glue" the SWIG interface on top | 18:22 |
lisitsyn | who knows what's proper SE | 18:22 |
lisitsyn | :P | 18:22 |
@HeikoS | well one example is the multiple inh. | 18:22 |
@lambday | HeikoS: I'd again vote for separating SWIG wrappers and the internals.. in the internals we can do static things.. compile time type check contraints | 18:22 |
lisitsyn | no way calling it good SE | 18:22 |
lisitsyn | :) | 18:22 |
lisitsyn | never good | 18:22 |
@HeikoS | lisitsyn: sure but at least be able to clean up things a bit more | 18:23 |
@HeikoS | SWIG should only see this "base ML interface" | 18:23 |
@HeikoS | and not the lib internals | 18:23 |
@HeikoS | for plugins anyawys | 18:24 |
lisitsyn | yes | 18:24 |
lisitsyn | we have to come up with clean interface | 18:24 |
@HeikoS | and then we can clean up the sh** inside the libshogun | 18:24 |
lisitsyn | I am for cleaning interface | 18:24 |
lisitsyn | and then rat shit bat shit | 18:24 |
@lambday | question is : where do we start | 18:24 |
lisitsyn | start with interface I am pretty sure that's the way | 18:24 |
@HeikoS | yeah | 18:25 |
lisitsyn | fake it till you make it and all these words | 18:25 |
@HeikoS | and the iterate over the internals | 18:25 |
@lambday | you mean, swig? | 18:25 |
@HeikoS | lambday: the things you did for testing, there are a lot of good design patterns that we could apply shogun wirde | 18:25 |
lisitsyn | HeikoS: I have an exercise | 18:25 |
lisitsyn | solve some ml problem just like you have the library you need :) | 18:26 |
lisitsyn | pretending it is here | 18:26 |
@HeikoS | yeah good point | 18:26 |
@HeikoS | should write that stuff down on hackathon | 18:26 |
@lambday | one such design pattern that I have used is in datamanager... no matter what kind of mix-and-match feature object you put in, it always works.. so classes that uses datamgr interface will never have to bother about handling different feature types again | 18:26 |
@lambday | I'd like our kernel interface do the same thing | 18:27 |
@HeikoS | lambday: features, I mean, what is happening there is ridiculus | 18:27 |
@HeikoS | either we use features base class and make everything flexible, or we use matrix directly | 18:27 |
lisitsyn | dot features is the best idea around | 18:27 |
@HeikoS | but using feature base pointer and then assert it to be matirx in one class and sparse matrix in other class | 18:27 |
@HeikoS | yea | 18:27 |
@lambday | yeah | 18:27 |
@HeikoS | http://www.shogun-toolbox.org/api/latest/classshogun_1_1CGaussianARDSparseKernel.html | 18:28 |
@HeikoS | thats my fav | 18:28 |
@HeikoS | should use interfaces inside | 18:29 |
@lambday | why a different class for sparse? | 18:29 |
@HeikoS | because it is called from SparseInference.h | 18:29 |
@lambday | umm I don't understand | 18:30 |
@lambday | so it does the same thing - just with sparse features | 18:30 |
-!- nagdev [dce39546@gateway/web/freenode/ip.220.227.149.70] has joined #shogun | 18:30 | |
@HeikoS | no, just with sparse inference | 18:31 |
@HeikoS | so what happens if we allow java like interfaces? | 18:32 |
lisitsyn | I think about complications from swig | 18:33 |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has joined #shogun | 18:35 | |
travis-ci | it's lambday'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/211403069 | 18:35 |
-!- travis-ci [~travis-ci@ec2-54-226-247-39.compute-1.amazonaws.com] has left #shogun [] | 18:35 | |
@lambday | weird - it blames even when build is green :D | 18:36 |
@lambday | HeikoS: just going a bit off topic here, can you merge this branch feature/bsd_license to develop? | 18:36 |
@HeikoS | lambday: checking | 18:38 |
@lambday | brb | 18:39 |
@HeikoS | lambday: link=? | 18:42 |
@HeikoS | lambday: so this is just one commit in feature branch | 18:44 |
@HeikoS | why not a pr for that? | 18:45 |
@HeikoS | why a feature branch? | 18:45 |
@sukey | Pull Request #3708 "changed the license to Shogun style BSD" opened by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3708 | 18:45 |
@sukey | Pull Request #3708 "changed the license to Shogun style BSD" merged by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3708 | 18:46 |
@sukey | New Commit "Merge pull request #3708 from shogun-toolbox/feature/bsd_license | 18:46 |
@sukey | changed the license to Shogun style BSD" to shogun-toolbox/shogun by karlnapf: https://github.com/shogun-toolbox/shogun/commit/1aa10b7342fcf1f35f4fd0b5a3e52f4241132c6d | 18:46 |
@lambday | HeikoS: umm other commits related to similar changes can go there | 18:51 |
shogun-buildbot | build #154 of deb1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/154 | 18:51 |
@HeikoS | lambday: mmh | 18:51 |
@HeikoS | not sure ;) | 18:52 |
@HeikoS | but yeah merged | 18:52 |
@sukey | Pull Request #3707 "Added GPL guard in optimization solver that uses libqp" synchronized by lambday - https://github.com/shogun-toolbox/shogun/pull/3707 | 18:53 |
@sukey | Pull Request #3709 "Sg memcpy" opened by beautifulsoup4 - https://github.com/shogun-toolbox/shogun/pull/3709 | 18:53 |
@sukey | Pull Request #3709 "Sg memcpy" closed by beautifulsoup4 - https://github.com/shogun-toolbox/shogun/pull/3709 | 18:59 |
@HeikoS | lambday: you still there? | 19:02 |
@lambday | HeikoS: yessir | 19:02 |
@HeikoS | I need your help :) | 19:02 |
@lambday | HeikoS: tell me | 19:03 |
@HeikoS | I want to compile shogun against anaconda | 19:03 |
@HeikoS | but I always get egfaults when importing | 19:03 |
@HeikoS | usually these happen when python libs dont match | 19:03 |
@HeikoS | but I have set | 19:03 |
@HeikoS | PYTHON_EXECUTABLE | 19:03 |
@HeikoS | PYTHON_INCLUDE_DIR | 19:03 |
@HeikoS | PYTHON_LIBRARY | 19:03 |
@HeikoS | all to the anaconda thing | 19:04 |
@HeikoS | yet when I fire it up, shogun segfaults | 19:04 |
@lambday | HeikoS: what segfaults? | 19:04 |
@lambday | anything in particular? | 19:05 |
@HeikoS | import modshogun as sg; sg.GaussianKernel() | 19:05 |
@HeikoS | you know how I can see which library modshogun.so refers to? | 19:05 |
@lambday | ls -l should show you the symlinks | 19:06 |
@lambday | no sorry | 19:08 |
@lambday | simply ldd | 19:08 |
@lambday | HeikoS: ^ | 19:08 |
@lambday | ldd _modshogun.so | 19:08 |
-!- HeikoS [~heiko@eduroam-int-pat-8-248.ucl.ac.uk] has quit [Ping timeout: 256 seconds] | 19:10 | |
-!- mikeling [uid89706@gateway/web/irccloud.com/x-uzcsyziwrtraokjz] has quit [Quit: Connection closed for inactivity] | 19:20 | |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has joined #shogun | 19:24 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 19:24 | |
@lambday | bye guys.. see you tomorrow | 19:50 |
@lambday | HeikoS: did you solve the python issue BTW? | 19:50 |
@HeikoS | ldd _modshogun.so gives me | 19:51 |
@HeikoS | libpython2.7.so.1.0 => /usr/lib/x86_64-linux-gnu/libpython2.7.so.1.0 (0x00007f8f30ec4000) | 19:51 |
@HeikoS | even though I set the python lib to be a different one using cmake | 19:51 |
@HeikoS | ha! | 19:52 |
@HeikoS | lambday: so there is something weird | 19:52 |
@HeikoS | libpython2.7.so.1.0 => /nfs/nhome/live/ucabhst/miniconda2/lib/libpython2.7.so.1.0 | 19:53 |
@HeikoS | libpython2.7.so.1.0 => /nfs/nhome/live/ucabhst/miniconda2/bin/../lib/libpython2.7.so.1.0 | 19:53 |
@HeikoS | thats the location of _modshogun and of my python execiutaböe | 19:53 |
@HeikoS | so they actually do match | 19:53 |
@HeikoS | gna | 19:53 |
@HeikoS | must be another lib | 19:53 |
@HeikoS | lambday: you got an idea how to debug such things? | 19:55 |
@HeikoS | python -c 'import modshogun as sg; sg.LibSVM()' | 19:55 |
@HeikoS | Segmentation fault (core dumped) | 19:55 |
@lambday | HeikoS: did you 'ldd' your modshogun.so? | 19:55 |
@lambday | oh | 19:56 |
@lambday | sorry | 19:56 |
@HeikoS | yes thats the thingg above | 19:56 |
@lambday | cmake screws up I thin | 19:57 |
@lambday | wait | 19:57 |
@lambday | what does ldd _modshogun.so | grep libshogun give you? | 19:58 |
@lambday | HeikoS: ^ | 19:58 |
@HeikoS | libshogun.so.18 => /nfs/nhome/live/ucabhst/git/shogun/shogun_kef_installed/lib/libshogun.so.18 | 19:58 |
@HeikoS | install location | 19:58 |
@lambday | sorry man.. that's the extent of my knowledge on this | 19:59 |
@HeikoS | I have the feeling it is libxml | 19:59 |
@HeikoS | by anaconda vs my system one | 20:00 |
@HeikoS | activating the conda env sets this lib to something on its own | 20:00 |
@lambday | that would cause segfault in new sgobject? | 20:00 |
@HeikoS | segfault is caused if the ld libs dont match | 20:01 |
@HeikoS | like you compile against some python lib and then run with another one | 20:01 |
@HeikoS | disabling xml | 20:02 |
@HeikoS | as it is the only other shared lib | 20:02 |
@HeikoS | if that doesnt work I dont know | 20:02 |
@HeikoS | the thing is | 20:02 |
@HeikoS | I want to use a recent jupyter | 20:02 |
@HeikoS | because trustys cannot open my notebook | 20:03 |
@HeikoS | shogun works if I compile against system python | 20:03 |
@HeikoS | but not for the conda one | 20:03 |
@HeikoS | but on systems ypthon i only have shitty old notebook version | 20:03 |
@HeikoS | and pip install notebook fails compile error | 20:03 |
@HeikoS | I want docker :( | 20:03 |
@HeikoS | this python packaging is such a desaster | 20:04 |
@HeikoS | really I mean | 20:05 |
@HeikoS | I just wasted half my day on this | 20:05 |
@HeikoS | I just want to use shogun | 20:05 |
@HeikoS | but NOTHING is backwards compatible | 20:05 |
@HeikoS | so if you are stuck with ubuntu 14.04 you are fucked | 20:05 |
@HeikoS | and need to mess around with ld path and all that | 20:05 |
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has quit [Ping timeout: 260 seconds] | 20:19 | |
CaBa | wiking: https://scontent-lhr3-1.xx.fbcdn.net/v/t1.0-9/16640897_1703399106353449_7288643772238740893_n.jpg?oh=ec08e5e8abf34affd7ec507cf07972ce&oe=592AC09D | 20:46 |
-!- nagdev [dce39546@gateway/web/freenode/ip.220.227.149.70] has quit [Ping timeout: 260 seconds] | 20:50 | |
-!- DerpGuy9001_ [7560e4c7@gateway/web/freenode/ip.117.96.228.199] has joined #shogun | 21:34 | |
DerpGuy9001_ | Hey! i was looking in the GSoC projects and found a few to be something that i possibly contribute towards. can anyone tell me what will be the requirements? | 21:36 |
DerpGuy9001_ | about me: I'm a 2nd year undergrad student doing computer science. I've had about a years experience using ml libraries and a little more with AI/ML. | 21:38 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 22:34 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 22:34 | |
-!- lambday [6a3314a2@gateway/web/freenode/ip.106.51.20.162] has quit [Ping timeout: 260 seconds] | 22:34 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 240 seconds] | 22:47 | |
@wiking | DerpGuy9001_, please read the wiki of shogun in order to find out what we require from the students in order to have a successful application for GSoC | 23:17 |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has quit [Remote host closed the connection] | 23:39 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has joined #shogun | 23:41 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has quit [Ping timeout: 268 seconds] | 23:47 | |
-!- eldorz [~eldorz@202.44.190.97.static.nexnet.net.au] has joined #shogun | 23:54 | |
-!- goksinen [~goksinen@rrcs-50-75-193-138.nyc.biz.rr.com] has joined #shogun | 23:56 | |
--- Log closed Thu Mar 16 00:00:06 2017 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!