Lab – Parsing & Abstract Syntax

	Progress with Jlex?
	Using CUP to construct Parser
	Syntactic and Semantic Analysis
	Abstract Syntax Trees
	Adding semantic actions to Parser
		The actions are to construct an abstract syntax tree

Building parser

	Need to write a grammar specification
		Grm.cup  - on your handout
		Look over PLDI3 notes for grammar specifications
	Structure of this file
		explained in book and on my web page in documentation

Syntactic vs Semantic analysis

	Syntax
		concerned with the structure of a program
		consider at both Lexical level
			stream of characters into tokens
		and Phrase level
			stream of tokens into phrases
	Semantic
		concerned with the meaning of a program
	Sets of rules to cover both areas

Engineering Syntax and Semantic checks

	Recursive descent, single-pass
		muddle all of these operations together
			lexical, syntax, semantic analysis
			code generation
	Multiple passes here
		how do we transfer information between passes?  What do we need?

ASTs – Abstract Syntax Trees

	Consider
		( a := 5 ; a + 1 )

Using semantic actions in Parser

	Series of classes to represent ASTs
		Absyn directory
		Use classes in action part of productions

For the lab

	Thoroughly understand the handout
		in conjunction with book and web
	Download materials in advance
		this time I WILL mail you when it's done

Examine Parse/Grm.cup

	Notice
		Structure of a production
		Naming of items in the production
		The action – constructing abstract syntax
			Referring to the items
			What is kleft?
		Definitions of AST classes held in directory Absyn

To do:

	Add a new rule for function declarations
		Read Tiger spec to be sure what a function should look like
		Follow style used for TypeDec since they’re similar
		Call the rule FunDec and add other rules as necessary
		Write new AST class – Absyn.FunctionDec

"Add rule/production for function..."

	Add rule/production for function invocation
		Where should it go?
		Write a new AST class Absyn.CallExp
		Reuse other ASTs and productions where possible
	Questions for the next session
		What is sym all about?
		Why is –expect 3 required when building the parser?
			Can we avoid it somehow?
			Do we need to avoid it?
			[Look at CUP pages and at chapter 3]