CodeGuru : Thinking in C++

Scoping

Scoping rules tell you where a variable is valid, where it is created and where it gets destroyed (i.e., goes out of scope). The scope of a variable extends from the point where it is defined to the first closing brace that matches the closest opening brace before the variable was defined. To illustrate:

//: C03:Scope.cpp
// How variables are scoped
int main() {
  int scp1;
  // scp1 visible here
  {
    // scp1 still visible here
    //.....
    int scp2;
    // scp2 visible here
    //.....
    {
      // scp1 & scp2 still visible here
      //..
      int scp3;
      // scp1, scp2 & scp3 visible here
      // ...
    } // <-- scp3 destroyed here
    // scp3 not available here
    // scp1 & scp2 still visible here
    // ...
  } // <-- scp2 destroyed here
  // scp3 & scp2 not available here
  // scp1 still visible here
  //..
} // <-- scp1 destroyed here

///:~

The above example shows when variables are visible, and when they are unavailable (that is, when they go out of scope ). A variable can only be used when inside its scope. Scopes can be nested, indicated by matched pairs of braces inside other matched pairs of braces. Nesting means that you can access a variable in a scope that encloses the scope you are in. In the above example, the variable scp1 is available inside all of the other scopes, while scp3 is only available in the innermost scope.

Defining variables on the fly

As noted earlier in this chapter, there is a significant difference between C and C++ when defining variables . Both languages require that variables be defined before they are used, but C (and many other traditional procedural languages) forces you to define all the variables at the beginning of a scope, so that when the compiler creates a block it can allocate space for those variables.

While reading C code, a block of variable definitions is usually the first thing you see when entering a scope. Declaring all variables at the beginning of the block requires the programmer to write in a particular way because of the implementation details of the language. Most people don’t know all the variables they are going to use before they write the code, so they must keep jumping back to the beginning of the block to insert new variables, which is awkward and causes errors. These variable definitions don’t usually mean much to the reader, and actually tend to be confusing because they appear apart from the context in which they are used.

C++ (not C) allows you to define variables anywhere in a scope, so you can define a variable right before you use it. In addition, you can initialize the variable at the point you define it, which prevents a certain class of errors. Defining variables this way makes the code much easier to write and reduces the errors you get from being forced to jump back and forth within a scope. It makes the code easier to understand because you see a variable defined in the context of its use. This is especially important when you are defining and initializing a variable at the same time – you can see the meaning of the initialization value by the way the variable is used.

You can also define variables inside the control expressions of for loops and while loops, inside the conditional of an if statement, and inside the selector statement of a switch. Here’s an example showing on-the-fly variable definitions:

//: C03:OnTheFly.cpp
// On-the-fly variable definitions
#include <iostream>
using namespace std;

int main() {
  //..
  { // Begin a new scope
    int q = 0; // C requires definitions here
    //..
    // Define at point of use:
    for(int i = 0; i < 100; i++) { 
      q++; // q comes from a larger scope
      // Definition at the end of the scope:
      int p = 12; 
    }
    int p = 1;  // A different p
  } // End scope containing q & outer p
  cout << "Type characters:" << endl;
  while(char c = cin.get() != 'q') {
    cout << c << " wasn't it" << endl;
    if(char x = c == 'a' || c == 'b')
      cout << "You typed a or b" << endl;
    else
      cout << "You typed " << x << endl;
  }
  cout << "Type A, B, or C" << endl;
  switch(int i = cin.get()) {
    case 'A': cout << "Snap" << endl; break;
    case 'B': cout << "Crackle" << endl; break;
    case 'C': cout << "Pop" << endl; break;
    default: cout << "Not A, B or C!" << endl;
  }

} ///:~

In the innermost scope, p is defined right before the scope ends, so it is really a useless gesture (but it shows you can define a variable anywhere). The p in the outer scope is in the same situation.

The definition of i in the control expression of the for loop is an example of being able to define a variable exactly at the point you need it (you can only do this in C++). The scope of i is the scope of the expression controlled by the for loop, so you can turn around and re-use i in the next for loop. This is a convenient and commonly-used idiom in C++; i is the classic name for a loop counter and you don’t have to keep inventing new names.

Although the example also shows variables defined within while, if and switch statements, this kind of definition is much less common than those in for expressions, possibly because the syntax is so constrained. For example, you cannot have any parentheses. That is, you cannot say:

while((char c = cin.get()) != 'q')

The addition of the extra parentheses would seem like a very innocent and useful thing to do, and because you cannot use them, the results are not what you might like. The problem occurs because ‘ !=’ has a higher precedence than ‘ =’, so the char c ends up containing a bool converted to char. When that’s printed, on many terminals you’ll see a smiley-face character.

In general, you can consider the ability to define variables within while, if and switch statements as being there for completeness, but the only place you’re likely to use this kind of variable definition is in a for loop (where you’ll use it quite often).

Contents | Prev | Next

Contact: webmaster@codeguru.com
CodeGuru - the website for developers. [an error occurred while processing this directive]