10:13 < kris> @zoq i have looking at the code and really is very diffcult to uderstand

10:14 < kris> i will go throught the meta template programming tutorials and then take a stab maybe that would be helpfule

16:46 < rcurtin> I think it's been a while since there has been a release, and that we should do one soon, but 3.x.x isn't ready yet I think (since the bindings aren't done and I think there is still work to do on the ANN code)

16:47 < rcurtin> so I wanted to suggest that maybe we do an incremental release, 2.1.2, that includes the various PRs and additions of the past few months rebased on the 2.1.x branch

16:49 < zoq> I agree, do you like to include DBSCAN and the Decision Tree?

16:57 < rcurtin> zoq: sure, I just have to fix the windows build for the decision tree... not sure what is wrong there, I have thought about it for a while

16:57 < rcurtin> I need to start experimenting and see what MSVC will build

16:59 < rcurtin> other PRs we could include could be #847 (perceptron tests), #840 (warning fix), #834 (GammaDistribution::Train()), #832 (Log compatibility with std::fixed and std::setprecision)

16:59 < rcurtin> #866 (refactor RSModel)

16:59 < rcurtin> I'll dig through more soon, maybe there are some I missed

17:04 < zoq> ah right the decision tree doesn't build on windows

17:06 < zoq> Some really nice fixes and additions.

17:16 < zoq> in best_binary_numeric_split_impl.hpp line 22 shouldn't it be BestBinaryNumericSplit<FitnessFunction>::AuxiliarySplitInfo<typename VecType::elem_type>& ..

17:25 < zoq> rcurtin: hm, if that's the case I wonder why it builds

18:19 < rcurtin> zoq: I'll try adding the FitnessFunction, but I saw the windows build failed on line 88 in that file also, and that line has the FitnessFunction in it too

18:19 < rcurtin> so maybe it's something else

18:19 < rcurtin> but who knows, worth a shot :)

18:21 < rcurtin> I think maybe this is a representation of what happens to the MSVC compiler when we put mlpack into it: https://www.youtube.com/watch?v=dq6T5BojXc8

18:27 < zoq> but I have to say the machine does handle that situation really well, until minute 2:22 it still works :)

18:29 < rcurtin> yeah, I am very impressed at how long it keeps going for

18:29 < rcurtin> apparently this guy spends his time destroying washing machines on youtube... now I'm watching one that's on fire

18:29 < rcurtin> and there's another video called "Washertron 2000 Killbot Battle Royale", which looks promising

18:31 < zoq> now I'm also watching the one with the fire ...

18:56 < govg> What is this conversation...

19:01 < zoq> govg: good question :)

19:03 < rcurtin> govg: it started with this message right before you logged in :)

19:03 < rcurtin> 18:21 < rcurtin> I think maybe this is a representation of what happens to the MSVC compiler when we put mlpack into it: https://www.youtube.com/watch?v=dq6T5BojXc8

19:03 < zoq> rcurtin: Perhaps if you also put typename before BestBinaryNumericSplit it fixes the problem

19:03 < rcurtin> I tried that, but then it broke on gcc

19:03 < rcurtin> but I might as well push that anyway to see what happens

19:03 < rcurtin> I have to step out for a few minutes, but I'll try it when I get back...

19:09 < govg> Hahah wtf is this.

19:10 < govg> The machine's pretty well built eh.

20:14 < zoq> rcurtin: That is also what I thought.

20:16 < rcurtin> yeah, I suspect this will also fail, because I think in this case ::template will need to be used, but I am not sure, let's see what MSVC says

22:49 < rcurtin> wow, it worked... so now I have to figure out how to get it to also compile on gcc... :)

23:16 < zoq> :) nice test

23:27 < rcurtin> yeah, I think this is definitely an MSVC bug, I just want to isolate it to the smallest possible test case so I can be sure of what's going on