Components :
------------

datagen		: write data files
schemagen	: write schema files 
loadgen		: write load files 
querygen	: write query files


Design Comments :
-----------------

The key design elements are :

	- an array of relations
	- each relation has an array of attributes

	- an array of queries
	- each query has an array of relations
	- each query has an array of joins
	- each query has an array of projections
	- each query has an array of restrictions

I did wonder about using linked lists or hashes instead of this
primitive approach. However it seems to work reasonably well as
it is. 

I have attempted to design this to be reasonably easily configurable,
however no real effort was made to provide a non programmable interface
to the parameters which control the config.

I decided to generate random data in the fillers, instead of repeating text.
This seemed like a good idea - as rel life data is seldom full of repeating
text :-). This uses a bit more cpu when running the datagen program, but
the effect is mininal - even on old (300Mhz) cpus.
