> Please comment on the following proposals: > > 1. x[...] > 2. x[...] = value > 3. x[..., newaxis] +1 for all of them as they provide a more generic way to write algorithms (n-dimensional FFT/interpolation/etc...) without worrying about corner cases.