Do we want to use Cython for wrapping?

[kyleabeauchamp]

We could convert rmsdcalc.c to a cython wrapper, as in current MDTraj.

@rmcgibbo

[ihaque]

👍 also consider cffi

[kyleabeauchamp]

One argument for cython is that it has really nice built-in support for Numpy arrays--I think cffi is still lacking some of these features. My favorite part about the new Cython wrappers is that I can use array indexing to grab pointers. I know this sounds really dumb, but this is my favorite line of code ever:

&xyz2[i,0,0],

[ihaque]

I can use array indexing to grab pointers

What, you don't like computing pointers manually from stride arrays?

[rmcgibbo]

That, @ihaque, is why when I do numerical C, I now use the gsl and functions like gsl_matrix_const_view_array_with_tda :)

But in all seriousness, re-wrapping the code in cython should be a pretty low priority, given that the C wrapper already works just fine. As we move towards py3k (the c wrappers are slightly different format for py3k, but cython will handle that for you) or other python interpreters (pypy), then it becomes an issue again. But in the mean time, this is nbd.