CSCE 330 Programming Language Structures Ch.2: Syntax and Semantics

1
CSCE 330Programming Language StructuresCh.2
Syntax and Semantics

• Fall 2005
• Marco Valtorta
• mgv_at_cse.sc.edu

2
Syntax and Semantics

• Syntax is the set of rules that specify the
  composition of programs from letters, digits and
  other characters.
• Semantics is the set of rules that specify what
  the result/outcome of a program is.
• Problems with English language description of
  Syntax and Semantics
• verbosity
• ambiguity

3
Syntax

• What is syntax?
• syntax vs. lexical rules
• Regular languages and context-free languages
• Backus Normal Form (a.k.a. Backus-Naur Form)BNF
• A syntax metalanguage
• Derivation vs. recognition
• Syntax Diagram
• Extended BNF (EBNF)

4
BNF History
In Java ltIfThenStatgt if (ltExprgt)
ltStatgt ltIfThenElseStatgt if (ltExprgt)
ltStatNoShortIfgt else ltStatgt
5
Recursive Descent Parsing

• Parsing is the process of constructing a parse
  tree
• A recursive descent parser is a kind of leftmost
  parser with very limited lookahead
• Recursive descent parsers are built directly from
  (E)BNF rules
• Recursive descent parsers do not work with
  left-recursive grammars
• We provide a simple example for parsing terms
  made of factors

6
The Concept of Binding

• entities (e.g., variables, statements,
  subprograms, declarations...) have attributes
  (e.g., for variable name, type, storage area)
• Binding is the specification of the exact nature
  of an attribute.
• When does binding occur? Binding time.
• language definition time
• language implementation time
• compile time
• run time
• Example the Fortran type INTEGER is bound partly
  at language definition time, partly at language
  implementation time.
• static (established before run-time, cannot be
  changed) and dynamic binding

7
Variables

• Name
• Scope
• Lifetime
• Value
• l-value (memory location) and r-value (contents
  of a memory location)
• Type

8
Variable Scope

• Variables have scope the range of program
  instructions over which the variable is known,
  and therefore manipulable
• scope binding can be static or dynamic
• dynamic scoping is easy to implement, but more
  confusing for most programmers
• most modern languages use static scoping

9
Variable Type

• type is the range of values a variable can take,
  together with operations to create, access, and
  modify values
• variable type declarations
• implicit in FORTRAN, explicit almost in any other
  language
• dynamic binding between variables and types is
  unusual
• APL and SNOBOL4 have it

10
Variable Value

• binding of variable and value is dynamic, except
  for symbolic constants
• Algol v. Pascal manifest constants
• reference (pointer)
• access path (chain of pointers)
• primary means of accessing anonymous variables
• shared objects

11
Semantics

• (So-called) Static Semantics
• Context-Sensitive Grammars
• Scope and Type
• Attribute Grammars
• Operational Semantics
• Denotational Semantics
• Axiomatic Semantics

12
Previewing Postscript

• In this course, most notes from the instructor
  are in Postscript format
• Postscript previewers are installed on the
  departmental Unix machines
• gv is available on all solaris machines (P.
  OKeefe)
• and on the College Windows machines
• GSView is located in all of the prototype
  machines. I copied the shortcuts in cse apps
  folder (H. Naik)
• a link to obtaining PostScript previewers is
  provided on the course web site. It points
  (indirectly) to http//www.cs.wisc.edu/ghost/

13
Static Semantics

• Static semantics are used to check type and scope
  rules
• Semantics is a misnomer in this context
• We provide only an example
• for checking the type rules of the variable and
  expression in the assignment statement of a
  simple language
• using attribute grammars (which are a
  formalization of contextual grammars)

14
Declarations, Expressions, Commands

• A command is executed to update variables and
  perform I/O
• An expression is evaluated to yield a value
• A declaration is elaborated (at compile time) to
  produce bindings. It may also have the side
  effect of allocating and initializing variables

15
Axiomatic, Denotational, and Operational Semantics

• Axiomatic semantics formalizes language commands
  by describing how their execution causes a state
  change. The state is formalized by a first-order
  logic sentence. The change is formalized by an
  inference rule
• Denotational semantics associates each language
  command with a function from the state of the
  program before execution to the state after
  execution
• Operational semantics associates each language
  command to a sequence of commands in a simple
  abstract processor