[SciPy-dev] fftpack: what should the API for fftn be ?

[David Cournapeau]

Hi,

    While taking a look at the ticket 244, there is something I don't
quite understand about the current behaviour for scipy fftn. The problem
in the ticket is that when both shape and axes arguments are given, fftn
wrongly ignore axes when padding the input according to shape. OTOH,
scipy fftn accepts shape and axes of different lengths (contrary to
numpy fftn), and I don't understand the expected behavior in that case.
Should we forbid the case len(shape) != len(axes); if not, what's the
expected behavior ?

cheers,

David