PyDAP handles its dependencies using setuptools. This means that
when you install it, a module called fpconst will also be installed,
even if you don't have setuptools on your machine. The module fpconst
is reponsible for handling NaNs and other constants/functions for
working with IEEE754 double-precision special values.

The client needs an implementation of N-arrays to handle matricial
data. You can use Numeric Python, Scipy Core or numarray. PyDAP
will use the first module it finds, in this order; if you want to
change this behaviour, edit the file dap/__init__.py and change the
import order.

The server requires additional modules, depending on what kind of
data you want to serve. For CSV (comma separated values) you won't
need anything else. The Matlab plugin requires Numeric Python, and
the netCDF plugin requires Numeric Python and Scientific Python.
If you want to use the SQL plugin with a RDBMS, you'll need to
install modules depending on your database: psycopg for PostgreSQL,
pysqlite for SQLite, etc.
