12:45 < Kirizaki> Hi

12:52 < zoq> Kirizaki: Hello there!

12:53 < Kirizaki> zoq, do You ever used DEBUG version of mlpack under WIN32?

12:54 < zoq> I don't think so.

12:54 < Kirizaki> ohh

12:55 < Kirizaki> I have this issue, if I cmake with -DDEBUG and change a little CMakeLists to add_definitions(-DDEBUG) for MSVC, it is not building later on with DEBUG flag

12:56 < Kirizaki> but if I, let's say FORCE this define DEBUG e.g. in core.hpp

12:56 < Kirizaki> and link any program against mlpack.lib, I'm getting some weird error, that Log::Debug symbols couldn't be found in lib

12:57 < Kirizaki> did ever had similar problem?

13:03 < zoq> Not that I'm aware of. Did you add "add_definitions(-DDEBUG)" to the else case of the DEBUG/MSVC case (line 135) or just above the DEBUG check in line 129?

13:05 < Kirizaki> I add it under IF (MSVC) line 283

13:05 < Kirizaki> and set DEBUG flag in CMakeLists ON as default

13:07 < zoq> Sounds good, I guess you tested that it's going into the MSVC case?

13:07 < Kirizaki> and as I know, official MSDN visual studio debug flag is _DEBUG

13:07 < Kirizaki> yes

13:07 < Kirizaki> tested.. I mean, I just add some MESSAGE(STATUS "it's IN!") :P

13:08 < zoq> yeah, okay

13:08 < Kirizaki> and later, when I was building I had MSVC and DEBUG flag :P

13:08 < Kirizaki> but, after I build mlpack.lib and link to my custom project

13:08 < Kirizaki> I'm getting runtime errors that linker cannot find symbols

13:09 < Kirizaki> later on I even add #define DEBUG 1; #define _DEBUG 1; #define MSVC into core.hpp and main.cpp of my custom project

13:09 < Kirizaki> but the same result

13:10 < zoq> Sounds really strange.

13:10 < Kirizaki> maybe is there something with MLPACK_EXPORT?

13:14 < Kirizaki> as I see, mlpack_export.hpp is generated by cmake

13:32 < zoq> MLPACK_EXPORT should be fine, did you also tested Log::Warn? Maybe there is a specific problem with Log::Debug not sure.

13:33 < Kirizaki> WARN, INFO work just fine

13:33 < Kirizaki> only debug and assert

13:33 < Kirizaki> which are dependent from -DDEBUG

13:35 < zoq> What's strange is that even if you build with DEBUG=OFF you can still use Log::Debug.

13:35 < zoq> Only that the output stream is null.

13:36 < Kirizaki> yes

13:36 < Kirizaki> it's due to NullOutStream I think

13:36 < Kirizaki> but I'm not sure enough

13:40 < zoq> What's the exact error you get if you use Log::Assert?

13:41 < Kirizaki> I cannot give exact message right now cause I'm rebuilding everything

13:42 < Kirizaki> but it's a LINKER error which cannot find lib symbols for mlpack::NullOutStream Debug or mlpack::PrefixedOutStream Debug

13:42 < Kirizaki> I don't remember right now

13:42 < Kirizaki> I'll copy message after I rebuild

13:43 < Kirizaki> if I'm building without DEBUG flag, Log::Debug and Log::Assert doesn't work at all

13:43 < Kirizaki> I mean, I don't have any error and any message

13:43 < Kirizaki> but program is running properly

13:44 < Kirizaki> if I could step into mlpack implementation while linking to custom project it could be helpful

13:44 < Kirizaki> but I don't know how to :P

13:47 < Kirizaki> ok, I think I know how to debug .lib, maybe thats how I'll find issue

13:59 < zoq> Kirizaki: Sounds great, also keep in mind that the output of Log::Info/Log::Debug is deactivated for the tests.

13:59 < Kirizaki> ok

14:00 < Kirizaki> thanks

16:44 < Guest5346> hi

16:45 < Guest5346> I really want to work on machine learning ..

16:46 < Guest5346> i have finished coursera course on machine learning and currently studying machine learning at my masters of engineering program at india

16:50 < zoq> Guest5346: Welcome! One place to start, if you have not already seen it, is: http://www.mlpack.org/involved.html

16:50 < zoq> There are a few issues on Github that might be a good place to jump in; we tag them by difficulty, here are the "easy" tickets:

16:50 < zoq> https://github.com/mlpack/mlpack/issues?q=is%3Aopen+is%3Aissue+label%3A%22D%3A+easy%22

16:51 < zoq> Maybe some of those are interesting?

16:51 < zoq> We are always open for interesting new methods and ideas. Let us know if we can clarify anything further.

16:56 < zoq> Guest5346: In case you missed the messages: http://www.mlpack.org/irc/

16:56 < Guest5346> yes i have seen this

16:56 < Guest5346> thanks for answer

16:57 < zoq> You are welcome!

16:59 < Guest5346> i know machine leaning a little bit, but i had no idea that people are implementing for C++

16:59 < Guest5346> I used matlab and python instead of C++

17:04 < zoq> Guest5346: Yeah, a lot of programmers are using "some language" (most likely Python) to express/evaluate the model. But in some cases the core isn't written in python or matlab e.g. the core of TensorFlow is written in C++/Cuda.

17:05 < Guest5346> Tension folw??

17:05 < Guest5346> i have no idea

17:07 < Guest5346> what do you mean by the core of TensionFlow ??

17:08 < rcurtin> Guest5346: https://www.tensorflow.org/ , it's Google's big new open-source library for deep learning

17:08 < rcurtin> I guess maybe it's not new anymore, it's been out for a year or so now I think

17:09 < Guest5346> ohh that's sound really interesting

18:05 < travis-ci> mlpack/mlpack#1725 (master - 26ac0d3 : Marcus Edel): The build passed.

18:05 < travis-ci> Change view : https://github.com/mlpack/mlpack/compare/48125493b947...26ac0d3f4c66

18:05 < travis-ci> Build details : https://travis-ci.org/mlpack/mlpack/builds/189569516

18:12 < travis-ci> mlpack/mlpack#1726 (master - d5148d9 : Marcus Edel): The build was broken.

18:12 < travis-ci> Change view : https://github.com/mlpack/mlpack/compare/26ac0d3f4c66...d5148d9c6a9b

18:12 < travis-ci> Build details : https://travis-ci.org/mlpack/mlpack/builds/189570097