FDL extends the functional data model to computational completeness while also 
supporting the persistence of any function, whether extensionally or 
intentionally defined.
FDL improves on previous implementations of a functional data model
as follows :
<UL>
<LI>
The language is formally based on the lambda calculus and so any recursive 
function can be defined.  Furthermore, all functions are stored in the database.
<LI>
All functions, whether used for data modelling purposes or for computation,
are treated uniformly with respect to their definition, evaluation, update 
and persistence.
<LI>
The data types of the language include arbitrarily nested lists, sums and 
products and all of them are persistent.
<LI>
Functions can be partly extensionally defined, partly intentionally 
defined and so default knowledge can be represented.
<LI>
All procedural data is stored in a pre-interpreted form ready for 
subsequent evaluation by a lambda calculus evaluator.
</UL>

We discuss these and other features of FDL in Sections 3 and 4 of the paper.
In Section 2 we give a brief introduction to <I>functional programming</I>,
which forms the theoretical basis of FDL.