02:00 < favre49> zoq I'm back, setting up took far longer than expected

02:01 < favre49> zoq I'm back, setting up took far longer than expected

02:01 < favre49> I'll start addressing comments

03:16 < jenkins-mlpack2> Yippee, build fixed!

03:16 < jenkins-mlpack2> Project mlpack - git commit test build #205: FIXED in 49 min: http://ci.mlpack.org/job/mlpack%20-%20git%20commit%20test/205/

04:09 < travis-ci> mlpack/mlpack#7648 (master - 0f33324 : Ryan Curtin): The build is still failing.

04:09 < travis-ci> Change view : https://github.com/mlpack/mlpack/compare/b88290902de4...0f333245052c

04:09 < travis-ci> Build details : https://travis-ci.org/mlpack/mlpack/builds/565772342

07:39 < jenkins-mlpack2> Project docker mlpack nightly build build #403: STILL UNSTABLE in 3 hr 25 min: http://ci.mlpack.org/job/docker%20mlpack%20nightly%20build/403/

14:47 < xiaohong> Hi, is anybody knows why when using the class in the ann/dists folder?

14:49 < xiaohong> I using ann::BernoulliDistribution, the compiler outputs that no type named 'BernoulliDistribution' in namespace 'mlpack::ann';

14:51 < xiaohong> did I use the wrong namespace?

14:59 < sreenik[m]> I checked and it worked for me (I didn't compile but it was there in suggestions). Did you forget to include it?

15:02 < xiaohong> Yes, I include the header file.

15:03 < xiaohong> I add new class in those folder, I want to use it. But in the same situation with BernoulliDistribution.

15:21 < sreenik[m]> Let's see if anyone else can debug this or come up with a solution other than rebuilding. For the time being you can include it separately (i.e., with the -I tag) while compiling if you haven't already tried that

15:22 < rcurtin> xiaohong: yeah, not all of the layers may be included by ann/layer.hpp; you may need to include the BernoulliDistribution file manually

15:27 < xiaohong> rcurtin: Do you mean include BernoulliDistribution.hpp in the file which I use the BernoulliDistribution?

15:28 < rcurtin> yeah

15:29 < rcurtin> just #include <mlpack/methods/ann/dists/bernoulli_distribution.hpp> I think should do it

15:35 < Yashwants19> Hi xiaohong: Are you using ann::BernoulliDistribution<>..??

15:36 < xiaohong> Actually, I am using the distribution I write by my own.

15:37 < xiaohong> But I using ann::NormalDistribution, the compiler can not find the NormalDistribution under ann namespace even though I include the head file.

15:37 < xiaohong> So I using ann::BernoulliDistribution as a test to see what's going on.

15:38 < xiaohong> The new distribution file in under the ann/dists/ folder.

15:38 < rcurtin> xiaohong: hmm, did you add the file to the SOURCES list in dists/CMakeLists.txt?

15:39 < xiaohong> Yes, I add it into the ann/dists/CmakeLists.txt.

15:41 < rcurtin> hm, ok, I see; the next guess I have is the header guard

15:42 < rcurtin> is it possible two files are defining, e.g., MLPACK_METHODS_ANN_DISTRIBUTIONS_BERNOULLI_DISTRIBUTION_HPP?

15:42 < rcurtin> (or some other name)

15:48 < xiaohong> rcurtin: Oh, I make clean the remake, it seems that I can use the BERNOULLI_DISTRIBUTION correctly.

15:49 < rcurtin> xiaohong: ah, ok. yeah, sometimes there are strange issues that get resolved with a 'make clean'... that probably means there is a bug in our CMake configuration somewhere but they are really hard and tedious to track down...

15:50 < xiaohong> Okay, it seems that I need to try more different ways. There has some bugs in my normal_distribution implementation, I need to fix it now.

15:50 < xiaohong> Thank you, all guys.

16:29 < rcurtin> sure, happy to help :)

21:33 < abernauer> rcurtin: Have you encountered issues with accessing memory addresses when handwriting bindings for other languages?

21:55 < abernauer> There is the appropriate C code to call the C++ code in R. Problem I' am encountering is with getting a reference to an overloaded function. Which is necessary to have an entry point into the C++ from R.