C++ Primer Plus полностью

explicit Student(const std::string & s)

    : name(s), scores() {}

explicit Student(int n) : name("Nully"), scores(n) {}

Recall that a constructor that can be called with one argument serves as an implicit conversion function from the argument type to the class type. This often is not a good idea. In the second constructor, for instance, the first argument represents the number of elements in an array rather than a value for the array, so having the constructor serve as an int-to-Student conversion function does not make sense. Using explicit turns off implicit conversions. If this keyword were omitted, code like the following would be possible:

Student doh("Homer", 10);  // store "Homer", create array of 10 elements

doh = 5;   // reset name to "Nully", reset to empty array of 5 elements

Here, the inattentive programmer typed doh instead of doh[0]. If the constructor omitted explicit, 5 would be converted to a temporary Student object, using the constructor call Student(5), with the value of "Nully" being used to set the name member. Then assignment would replace the original doh with the temporary object. With explicit in place, the compiler will catch the assignment operator as an error.

C++ and Constraints

C++ is full of features that allow programmers to constrain programmatic constructs to certain limits—explicit to remove the implicit conversion of single-argument constructors, const to constrain the use of methods to modify data, and more. The underlying motive is simply this: Compile-time errors are better than runtime errors.

Initializing Contained Objects

Note that constructors all use the by-now-familiar member initializer list syntax to initialize the name and scores member objects. In some cases earlier in this book, such as the following, the constructors use it to initialize members that are built-in types:

Queue::Queue(int qs) : qsize(qs)  {...} // initialize qsize to qs

This code uses the name of the data member (qsize) in a member initializer list. Also constructors from previous examples, such as the following, use a member initializer list to initialize the base-class portion of a derived object:

hasDMA::hasDMA(const hasDMA & hs) : baseDMA(hs) {...}

For inherited objects, constructors use the class name in the member initializer list to invoke a specific base-class constructor. For member objects, constructors use the member name. For example, look at the last constructor in Listing 14.1:

Student(const char * str, const double * pd, int n)

       : name(str), scores(pd, n) {}

Because it initializes member objects, not inherited objects, this constructor uses the member names, not the class names, in the initialization list. Each item in this initialization list invokes the matching constructor. That is, name(str) invokes the string(const char *) constructor, and scores(pd, n) invokes the ArrayDb(const double *, int) constructor, which, because of the typedef, really is the valarray(const double *, int) constructor.

What happens if you don’t use the initialization list syntax? As with inherited components, C++ requires that all member objects be constructed before the rest of an object is constructed. So if you omit the initialization list, C++ uses the default constructors defined for the member objects’ classes.

Initialization Order

When you have a member initializer list that initializes more than one item, the items are initialized in the order in which they were declared, not in the order in which they appear in the initializer list. For example, suppose you write a Student constructor this way:

Student(const char * str, const double * pd, int n)

       : scores(pd, n),  name(str) {}

The name member would still be initialized first because it is declared first in the class definition. The exact initialization order is not important for this example, but it would be important if the code used the value of one member as part of the initialization expression for a second member.

Using an Interface for a Contained Object

The interface for a contained object isn’t public, but it can be used within the class methods. For example, here is how you can define a function that returns the average of a student’s scores:

double Student::Average() const

{

    if (scores.size() > 0)

        return scores.sum()/scores.size();

    else

        return 0;

}

This defines a method that can be invoked by a Student object. Internally, it uses the valarray size() and sum() methods. That’s because scores is a valarray object, so it can invoke the member functions of the valarray class. In short, the Student object invokes a Student method, and the Student method uses the contained valarray object to invoke valarray methods.

Similarly, you can define a friend function that uses the string version of the << operator:

// use string version of operator<<()

ostream & operator<<(ostream & os, const Student & stu)

{