README_build_ferret

As of August 2011 we now have a single make procedure for compiling and linking Ferret. 


NetCDF4.1 or higher is required for building Ferret. The library netcdf4.2 or higher
is recommended.  For building the libraries and Ferret, we have used 
gfortran. The netcdf downloads can be found here: http://www.unidata.ucar.edu/software/netcdf/.
Build netcdf4.+ with the appropriate vesion of hdf5 given in the netCDF instructions; and with 
--enable-fortran in the configuration. Specify the hdf5 libraries, --enable-netcdf-4
--enable-dap for   for netcdf-4 features and OPeNDAP dataset access.

The current official release of Ferret is linked with netcdf4.2 and has syntax for 
writing netcdf4 datasets (hdf5 chunking and compression.)  See the release notes 
for Ferret v6.6 at http://porter.pmel.noaa.gov/Ferret/home/documentation/v6-6-release-notes


To build Ferret:

  * Under the FERRET directory you will find a file called site_specifid.mk.  Edit this file
  and insert the locations of the HDF5, NetCDF, and readline directories, and java.  Java is
  used only for the ThreddsBrowser facility of Ferret and is not required.

  * Still in the top-level FERRET directory, run "make". This will compile and link Ferret, 
  gksm2ps, external functions, and the fonts files.  The Ferret executable is the file
  FERRET/fer/ferret_c
  

  * as it finishes, you will see a line similar to 

        make[1]: Leaving directory '/home/users/ansley/linux/FERRET/bin/build_fonts/unix'
	
    Take note of this directory.


  * Get the environment tar file for one of our linux versions. You will need the go scripts, 
  and palette files. See the bin/Finstall script and the Ferret installation instructions for 
  details about setting up the environment to run Ferret.  You could run Finstall as described
  in the installation instructions, but replace the binary file "ferret_vxx" with the executable
  you have built, and bin/ppl/fonts/ with the fonts files in .../FERRET/bin/build_fonts/unix

  * Once you have an executable, go into the FERRET/bench directory to run tests. There is a 
  simple README file which will get you started with running the Ferret benchmark scripts.  
  You will run a shell script called "run_all" which starts the benchmark job.