* Created a distinct parser for each expression type and for identifiers
* Removed all expression rules from PrismParser, using new parsers instead
* Reduced excessive usage of boost::lambda, using semantic actions only for single calls
* Moved actual state to new class (-> VariableState, whole two-run-logic can probably implemented there)
* Much cleanup
Work should be finished on expression parser, but not yet on prism parser...
This includes clone() routines (that accept the renaming information) for all ir::expressions classes, additional constructors for all ir classes and additional rules in PrismParser.
This is the first version that has the chance to work and actually compiles. (Insert some swearing here...)
Testing is to be done, as this will most probably not work yet.
gereon
dehnert