Installation of the FFTW package in the VLT context
---------------------------------------------------

Basically all you have to do is execute the following command in the present
directory:

make update

This unpacks the source files, include files and documentation, add VLT
headers to the code, and put it into the relevant directories under ws/ and
lcu/. Actually, most of the code is copied into the ws module, and the
Makefile in the lcu module resolves all necessary files through symbolic
links.

All you have to do then is go to the ws/src and lcu/src subdirectories,
and compile as usual like standard VLT modules.


Notes about the original FFTW source code
-----------------------------------------

- The code used here is only a subset of the original library. Unwanted
  functionalities have been left aside. Therefore the provided documentation
  may refer to routines which are not available here.

- Under Unix, the library is generally compiled with the help of the
  "configure" program. The source files included here have been created in
  such a way for HP-UX, using the command

   ./configure --enable-float

  Indeed, the library is used only in simple precision in the present context.
  
  The .c and .h files in the fftw/ and rfftw/ of the original package have
  then been packed into the fftw.tar.gz archive of the present directory.
  Unwanted files have been left aside, like the ones dealing with the Fortran77
  interface.

- The "config.h" file for the VxWorks version has been edited by hand to
  reflect the configuration available on this platform. It is included in the
  fftw.tar.gz archive under the name "config-vxworks.h". The installation
  shell will copy it under the right name in the right directory.

- Useful explanations about configuration and installation of the FFTW package
  are included in the documentation, in the chapter "Installation and
  customization".


