Re: Using FFTW in Blitz and SL++ (fwd)

From: Todd Veldhuizen (tveldhui@oonumerics.org)
Date: Thu Jun 11 1998 - 11:02:51 EST


Forwarded message:
>From athena@glauke.lcs.mit.edu Thu Jun 11 12:01:54 1998
Date: Thu, 11 Jun 1998 11:59:43 -0400
Message-Id: <199806111559.LAA10867@glauke.lcs.mit.edu>
From: Matteo Frigo <athena@glauke.lcs.mit.edu>
To: tveldhui@oonumerics.org
CC: fftw@theory.lcs.mit.edu, laurent.deniau@cern.ch, karmesin@lanl.gov,
        swhaney@lanl.gov
In-reply-to: <199806111544.LAA31500@oonumerics.org> (message from Todd
        Veldhuizen on Thu, 11 Jun 1998 11:44:31 -0400 (EDT))
Subject: Re: Using FFTW in Blitz and SL++
Return-Receipt-To: athena@glauke.lcs.mit.edu
Comments: Hyperbole mail buttons accepted, v04.023.
References: <199806111544.LAA31500@oonumerics.org>

> Deciding in advance that users may only do FFTs in double precision
> (or single precision) would be too restrictive.
>
> Did you have plans or thoughts about how to address this?

This seems to be a common request and we should definitely do
something about it. Our idea was to mangle the names by means of
macros, so that, say, FFTW_COMPLEX becomes FFTW_MANGLE(COMPLEX), and
FFTW_MANGLE(name) produces fftw_name by default, but can be changed to
fftws_name or fftwd_name or whatever. We never did it because this
solution is too ugly, but we haven't figured out anything better yet.
Any suggestion is welcome.

> - Did you have other ideas about how to handle this?
> - Would you be willing to incorporate such changes back into the
> main FFTW source tree, provided they weren't too ugly?

If it is not too ugly I don't have any problem with that (and
presumably Steven won't either). I would vote for

FFTW_DEFUN(foo)

instead of

FFTW_FUNCTION foo

but either way is fine.

Cheers,
Matteo Frigo



This archive was generated by hypermail 2b29 : Wed Feb 20 2002 - 04:30:04 EST