math/fftw with --enable-float ?

[Ray Phillips <ray.phillips%gmail.com@localhost>]

I wonder if the fftw binary package could be compiled using the
--enable-float option  ?  At the moment it hasn't been, as far as I
can tell (fftw-3.3.6pl1 is the package I've installed).

Would there be any drawbacks in doing that?

I ask because I've tried to install LALSuite [1] on NetBSD/amd64 7.1
and not only is one of its core dependencies FFTW, but its configure
script demands to have the floating-point version of the library.  The
last couple of lines of its output when attempting to use the binary
package were:

checking for FFTW3... no
checking for fftwf_execute_dft in -lfftw3f... no
configure: error: could not find the fftw3f library

Searching /usr/pkg for a file name containing fftw3f drew a blank.

However, LALSuite's configure was happy when I installed FFTW from
source using the  --enable-float  configuration option:

checking for FFTW3... yes
checking for fftwf_execute_dft in -lfftw3f... yes
checking for fftw_execute_dft in -lfftw3... yes
checking fftw3.h usability... yes
checking fftw3.h presence... yes
checking for fftw3.h... yes


Ray

[1]  https://wiki.ligo.org/DASWG/LALSuite