Revised 13 May 2009

To build this library, follow the steps below.
Some system specific notes are at the end of this file.

Build Instructions
==================

  1. Run the configure script by typing

        ./configure

     The script will attempt to automatically detect various system-dependent
     variables used during compilation (such as the C++/fortran compiler, 
     compiler flags, and linker flags).

     If you want to specify a particular C++ / F90 compiler and their flags, 
     you can set them as environmental variables.  For example:

       FC=ifc FCFLAGS="-O2 -FR" ./configure

     Important variables are

       CXX       C++ compiler to use
       CXXFLAGS  C++ compiler flags to use
       CC        C compiler to use (for C demo program)
       CFLAGS    C compiler flags to use (for C demo program)
       FC        Fortran 90 compiler
       FCFLAGS   Fortran 90 compiler flags to use
       FCLIBS    Fortran 90 libraries needed to to link with C++ code.

     See ./configure --help to see other options. 

  3. The configure script should also have created the files 'config.h' and
     'include/arprec/arprec_config.h', which will contain the compile time 
     defines.  Examine these and edit them if necessary.  In most cases 
     no edits are necessary, since the options are detected when configure
     was run.

  4. Type "make".  This will build the library, and necessary Fortran
     wrappers.

  5. Optionally, one can build and run some simple test programs.
     To do this, type "make check".  Some programs run during this
     phase is a good demonstration of how to use the qd library in C++.

  6. You can now install the ARPREC library by issuing "make install".

  7. If you want to build some sample programs written in C++
     you can type "make cpp-demo".

  8. If you want to build some sample programs written in Fortran 90, 
     you can type "make fortran-demo".

  9. If you want to compile the Experimental Mathematician's Toolkit,
     type "make toolkit".  This will compile the Fortran-90 codes in
     the toolkit cirectory, including the "mathinit" and "mathtool"
     Read the "README" file in the toolkit directory for additional details.


System-Specific Notes
=====================

Linux x86 / Itanium
-------------------
You can use g++ to compile the C++ code.  The Fortran 90 codes
can be compiled using Intel Fortran 95 compiler

  http://www.intel.com/software/products/compilers/flin/

available freely for non-commercial uses.  There is also a C++
compiler available (for non-commercial use) at

  http://www.intel.com/software/products/compilers/clin/

which can be used to compile the C++ portion.  By default the 
configure script will use the Intel compiler if found.

Apple (OS X)
------------

For Apple OS X Intel-badsed systems, it is recommended that you use
the g++-4.0 (or higher) compiler and the gfortran compiler.  The
g++-4.0 (or higher) compiler is available as part of the xcode
package, which is iprovided by Apple.  gfortran can be downloaded
from:
http://www.macresearch.org/files/gfortran/gfortran-4.3-Nov.mpkg.zip

After installing these compilers, in the main arprec directory type

CXX=g++-4.0 FC=gfortran ./configure

to configure arprec to use these compi.lers.  This is with the bash
shell; with other shells you may need to type

./configure CXX=g++-4.0 FC=gfortran

On some systems additional flags must be included to ensure that
portions of the library are not built with 32 and 64 bit object
files. For example, on 64-Bit Mac OSX 10.6 (Snow Leopard) and OS 10.7
(Lion) the correct configure line using GNU compilers is: 

        ./configure CXX=g++ FC=gfortran FCFLAGS=-m64

IBM (Power)
-----------

With IBM's xlC/xlf90 compilers, you may want to experiment with
--enable-fma option which uses a faster code but relies on the
compiler to generate a fused multiply-accumulate instruction.
WARNING: since the compiler is not required to produce such
instructions, this is not guaranteed to work.  Please test before
using.

