CS 541 Lecture -*- Outline -*-

* Introduction to Operational Semantics

** definition
------------------------------------------
    OPERATIONAL SEMANTICS

def: an operational semantics of
 a language L is:


------------------------------------------
        ... a map from L into some language M,
            and a definition of M

	Q: Have we made any progress in defining L?
	Q: what's an example?
		e.g., compiler translates program into machine language

** problems
*** well-definedness
	worry that the M may not be well-defined if it's complex
		(a real machine)
	solution is to use a simple, formally defined machine
		e.g. Turing machine or Landin's SECD machine
*** using the semantics to reason about a program in the original language
	hard if have to follow lots of steps in the computation of the machine
	harder if the machine is complex

	one solution is to make a "1 step" machine,
		where reasoning is directly reflected in map into the machine.
		but there is still a problem if you have any kind of looping
			or recursion.

** advantages
*** applications to compilation
	- study of compiler optimizations
	- correctness proofs for compilers
	- design of new compilers by design of new abs. machine
				e.g., Warren abstract machine (Prolog)
					graph reduction engines (functional)

*** easiest kind of semantics to write down
	not a lot of complex math to master,
		basically Prolog

*** best for semantics of concurrency (it seems)
	denotational approaches have trouble in their foundations
                     for concurrency,
                     although there are stream-based techniques
                     in denotational semantics that help


** remarks
	it's because of the structural framework I'll present (due to Plotkin)
		that operational semantics has become more and more important

** Things to think about specifying
	- recursion
	- errors
	- can you leave some things unspecified