On 29-lug-10, at 15:31, dsimcha wrote:
== Quote from Fawzi Mohamed ([email protected])'s article
otherwise, as I already said http://www.netlib.org/fftpack/fft.c is
public domain, has reasonable performance, and supports all sizes.
It might not be the always beautiful, but it is stable and works
well...
Fawzi
Yeah, I just looked at this code. The problem is that I wouldn't do
a straight,
mechanical port to D, I'd D-ify the code a little while I was at it
(such as
making it work on generic random access ranges if it's an out of
place transform,
making real-only inputs vs. complex inputs DRYer with templates,
etc.). Since
this code seems to have been straight-up translated from FORTRAN, it
scores about
a -2 the readability scale, from 1 to 10. In other words, I would
probably be
able to translate it and make it run, but even the slightest
modification would be
near impossible.
eheh this must meant that too much fortran inured me, I did not find
the code too bad, I classified it as a typical numerical intensive
code (even rather clean where one can easily find the boundaries of
the arrays, and without too many goto), and templating float, using
slices as input seemed easy.
But yes probably it is not the best basis to develop your own thing.
Ahrg fortran ruined me ;).
ciao
Fawzi
