                     Numerical Python

    Versions after 20.0 require Python 2.0 or later. To take advantage of 
    the "rich comparisons" (i.e., to be able to compare arrays and get
    back a boolean result) you need Python 2.1.

Web site: http://numpy.sourceforge.net
Project page: http://sourceforge.net/projects/numpy
Discussion group: numpy-discussion@lists.sourceforge.net

===> Silicon Graphics bug:
    Due to a bug in the SGI compiler's handling of complex
    variables, you must build your *python* without optimization
    on a certain file. Consult the Python installation instructions
    for the details.

INSTALLING BINARY DISTRIBUTIONS ON WINDOWS
    Get the .exe installers and simply run them. Python must already be installed.

BUILDING AND INSTALLING NUMERIC PTYHON FROM SOURCE
    For Windows get the .zip sources. For other platforms use the .tar.gz sources.
    After unpacking change to the top-level directory.

    Using the python into which you wish to install Numeric Python, execute:

    python setup.py install  # Core only
            or
    python setup_all.py install  # Core and all optional packages (see below)

    To clean up completely after an install, on Unix you can do:
    /bin/rm -fr build Packages/*/build

    See the Distutils documentation for many command-line options as to 
    installation locations, etc.

    By default Numeric builds itself with a small version of the LAPACK and
    BLAS libraries. See setup.py. To use your own BLAS and LAPACK you will 
    need to edit the settings in setup.py.

OPTIONAL PACKAGES 
    Subdirectory Packages contains optional packages you may wish to install.
    Each package has its own setup.py file or you can use the 
    setup_all.py file in the main directory.  You can edit setup_all.py to 
    change the list of packages to be installed.

    Package Properties is required for package kinds and package MA.

CVS Repository, Bug Tracking, Documentation, Patches, Mailing Lists

   Access to the source via CVS, a bug tracking facility, a patch submission
   facility, and the numpy-discussion mailing list are all on the 
   Numerical Python Project page:  http://sourceforge.net/projects/numpy

   The Numerical Python Web Site is http://numpy.sourceforge.net. It contains
   links to the project page and the documentation. It also has a few links 
   to related sites.

History

This software was originally written by Jim Hugunin when he was at MIT, later
at CNRI.

Then Paul Dubois of LLNL became the "keeper". LLNL has released any 
contributions to it by LLNL personnel for free redistribution. 
However, many people have contributed to Numpy.  It is a classic open source effort 
that nobody "owns".  So, Numerical Python is placed in the SourceForge under 
the Python license. The LLNL license is in Legal.htm.

 -- Paul F. Dubois, Jan. 13, 2000.
