Re: GSoC 16: Sample implementation for ispolycw function

[John Swensen]

On Mar 23, 2016, at 10:28 AM, amr mohamed <address@hidden> wrote:

Dear all,

I have made a sample cc file for the ispolycw function using Boost/Geometry library.
I have tested it and it is working well.
Could you please review it so that i can finalize it? 
https://ideone.com/w6WFzB

Regards, 
Amr Mohamed

This looks like a good start of getting at least one function implemented using the Octave C++ API and including Boost::Geometry. 

Just two quick comments:

1) You might also want to work on making this handle self-intersections in the same way that the Matlab documentation indicates they handle self-intersections

2) To show that testing works (and correctly), you can include tests written in the Octave language as comments directly in the C++ code.

See http://hg.octave.org/octave/file/tip/libinterp/dldfcn/chol.cc where there is the following example that checks whether the inverse from the Cholesky decomposition gives the same result as the standard matrix inverse computation.

/*
%!shared A, Ainv
%! A = [2,0.2;0.2,1];
%! Ainv = inv (A);
%!test
%! Ainv1 = cholinv (A);
%! assert (norm (Ainv-Ainv1), 0, 1e-10);
%!testif HAVE_CHOLMOD
%! Ainv2 = inv (sparse (A));
%! assert (norm (Ainv-Ainv2), 0, 1e-10);
%!testif HAVE_CHOLMOD
%! Ainv3 = cholinv (sparse (A));
%! assert (norm (Ainv-Ainv3), 0, 1e-10);
*/

If you could include tests like this for at least one clockwise and one counterclockwise polygon, that would help show it works and allow there to be regression testing going forward in the case that something changes in Boost::Geometry or a bugfix causes a different error.

Let me know if you have any questions.

John S.