Previously I had implemented some "fast" routines using Cython which
were poorly written loops that were marginally faster than looping using
NumPy arrays. These have now been superceded by comparably faster
routines using JIT compilation in Numba, which now reside in the
lineshapes module.
For this reason, I am removing Cython routines from the library, which
in turn means we no longer need to run cythonize during setup time,
which had always plagued Windows installations. This references issue
[#15].
Previously I had implemented some "fast" routines using Cython which were poorly written loops that were marginally faster than looping using NumPy arrays. These have now been superceded by comparably faster routines using JIT compilation in Numba, which now reside in the
lineshapes
module.For this reason, I am removing Cython routines from the library, which in turn means we no longer need to run cythonize during setup time, which had always plagued Windows installations. This references issue [#15].