These C programs 'cases.c', 'table.c', 'funptr.c' implement the same
automata that was used in the lecture as an example for the product and
power set construction.  The regular language accepted is the intersection
of 'a b 1*' and '1* b a'.  The programs can be compile with 'make -f
<name>.make' and tested with './test <name>'.  By default 'gcc' is used as
compiler and coverage information is included.  Run for instance 'gcov
funptr.c' after testing it with './test funptr' to produce an annotated
listing of 'funptr.c'.

The 'cases.c' program is the obvious implementation using a 'switch'
statement.  An optimizing compiler may reduce the overhead of calculating
the next state to a constant lookup as well, but usually it is better to
use a transition matrix as in 'table.c'.  This is faster but also takes more
space.  It is the standard implementation technique in for instance parser
generators, such as 'yacc'.

The last version 'funptr.c' uses function pointers and also just needs
constant time for each state transition.  It is similar to the 'objects for
state design pattern' in object oriented programming.  It is very useful
for encapsulating the work done in each transition, for instance if
'extended' automata are used, which beside the state itself have additional
global data.

Armin Biere, JKU Linz, March 2005.