Symbol Tables

Symbol tables map identifiers to their attributes

1
 Symbol Table
A compiler or interpreter maintains a symbol table containing
information about known identifiers (and maybe other symbols).
Each new scope is given a number. Scope 1 (this file)
Name Scope Cat. Type Other
float f(float x) { f 2 func. float (param)
float y = x*x + 1; main 3 func. int (void)
return y; Scope 2
}
Name Scope Cat. Type Other
x 2 param float -
y 2 local float -
int main( ) {
float sum; Scope 3
sum = f(1.0) + f(2.5); Name Scope Cat. Type Other
} sum 3 local float -

2
 Predefined Symbols Table
The outer most symbol table (0) contains predefined identifiers, such
as "int", "float". Each new scope needs its own symbol table.

Unresolved Symbols: Scope 0 (predefined)

Name Scope Cat. Type Other Name Scope Cat. Type Other
pow - func ? (float,int) float 0 type float reserved
int 0 type int reserved

float f(float x, int n) { Name Scope Cat. Type Other

float sum = 0; x 2 param float -
for(int k=1;k<=n;k++) { n 2 param int -
sum += pow(x,k); sum 2 local float -
}
return sum; Name Scope Cat. Type Other
} k 3 local int -

3
 Implementing symbol table structure
 One strategy is to use a stack of symbol tables.
 When a scope is entered, its symbol table is put on top of stack.

Name Scope Cat. Type Other

n 3 param int -
import static
count 3 local int -
java.Math.PI;
class A {
Name Scope Cat. Type Other
private int count;
count A attrib int private
private int x;
fun A func. int (int)
/* new scope */ x A attrib int private
int fun( int n ) {
int count = 0;
count += n*x*PI; Name Scope Cat. Type Other
PI
}
... 4
 Implementing symbol table structure
 Another strategy is to use a linked list of symbol definitions.
 This requires a second data structure: a scope stack to indicate
the order of active scopes.
Name Scope Cat. Type Next
class A { x A param int -
private int count; count A attrib int *
fun A func. int (int)
private int x;
n 3 param int -
/* new scope */
count 3 local int -
int fun( int n ) { ...
int count = 0;
count += n*x;
}
... Scope stack
int main( ) { A
A a = new A(20); 1 (main)
a.fun(10); 3 (fun) 5
 Symbol Table and Scope

/* Java */ /* C++ */
class who { public int who( ) {
private String who; int n = 10;
for(int k=1;;k<10) {
public int who( ) { int n = 10*k;
... sum = sum + n;
} }

public void main(...){ cout << "n=" << n ;

int who; }
who =
(new who()).who();
}
}
6
Symbol table structure
 Symbol table is constructed as declarations are
encountered (insert operation).
 Lookups occur as names are encountered in dynamic

scope (in symbol table to that point).

 In lexical scope, lookups occur either as names are

encountered in symbol table to that point (declaration

before use—C), or all lookups are delayed until after
the symbol table is fully constructed and then
performed (Java class—scope applies backwards to
beginning of class).

7
Symbol table structure evaluated
 Which organization is better?
 Table of linked lists is simpler (C, Pascal).

 Stack of symbol tables is more versatile, and helpful when

you need to recover outer scopes from within inner ones or

from elsewhere in the code (Ada, Java, C++). Examples:
this.x = x; // assign local value to attrib.
super.toString( );
Math.PI;
 Normally, no specific table structure is part of a language
specification: any structure that provides the appropriate
properties will do.

8
Ada example
(global symbol table:)
name bindings

ex procedure
symtab

name bindings

x integer

y character

procedure
p
symtab

name bindings

x fl oat

block
A
symtab

name bindings

y array (1..10) of
integer

9
Overloading
 Overloading is a property of symbol tables that allows
them to successfully handle declarations that use the
same name within the same scope.
int max( int a, int b ) { return ( a > b ) ? a : b; }
float max ( int a, int b ) { return (a > b ) ? (float) a : (float)b; }
int max( int a, int b, int c ) { return ( max(a,b) > max(b,c) ) ? max(a,b) : ...
}
float max ( float a, float b ) { return (a > b ) ? a : b; }

int main( ) {
float y = max( 3, 4);
max( 2.5, 7 );
10
Overloading
 It is the job of the symbol table to pick the correct
choice from among the declarations for the same name
in the same scope.
 This is called overload resolution.
 It must do so by using extra information, typically the
data type of each declaration, which it compares to the
probable type at the use site, picking the best match.
 If it cannot successfully do this, a static semantic error
occurs.

11
Overloading (2)
 Overloading typically applies only to functions or methods.
 Overloading must be distinguished from dynamic binding in an
OO language.
 Overloading is made difficult by weak typing, particularly
automatic conversions.
 In the presence of partially specified types, such as in ML,
overload resolution becomes even more difficult, which is why
ML disallows it.
 Scheme disallows it for a different reason: there are no types on
which to base overload resolution, even during execution.

12
Overloading (3)
 An example in Java:
public class Overload {
public static int max(int x, int y)
{ return x > y ? x : y;}
public static double max(double x, double y)
{ return x > y ? x : y;}
public static int max(int x, int y, int z)
{ return max(max(x,y),z);}
public static void main(String[] args)
{ System.out.println(max(1,2));
System.out.println(max(1,2,3));
System.out.println(max(4,1.3));
}
}
 Adding more max functions that mix double and int parameters is
ok.
 Adding functions that mix double and int return values is not!
13
Overloading (4)
 C++ and Ada are even more challenging for overload resolution:
 C++ allows many more automatic conversions,
 In Ada the return type is also used to resolve overloading (Ada
can do this only because it allows no automatic conversions).
 It is possible for languages to also keep different symbol tables for
different kinds of declarations.
 In Java these are called "name spaces," and they also
represent a kind of overloading.
 Java uses different name spaces for classes, methods,
variables, labels, and even packages.

14