James Bergstra writes: > Admittedly, what I had in mind was in fact a third interface, that adds > considerable code, and no new functionality. I don't believe that it should > replace the gsl fft api. I like the extension system, and I think this is > the > sort of thing that would make a good extension. > > My goal was an API that is as easy to call as gsl's, and that admits an FFTW > implementation, but doesn't require it ( gsl has a priority of being > self-contained).
I am pretty sure it should be possible to create a wrapper that allows FFTW to be substituted for GSL calls with no changes to the user code (by using suitable #defines or dynamic linking to substitute the alternative functions). For single-threaded code at least the FFTW plan can be created in the GSL wavetable setup. -- Brian Gough
