Thread: Copy constructor

There are several places in C++ programming where a copy of an object is made. If, for example, you pass an object to a function, the normal "pass by value" mechanism is used, and a copy of the object is made.

If you have not specified a copy constructor, the compiler uses the default copy constructor to make the object. The default is a bitwise copy, i.e. it is exactly the same as the object in the calling routine. When the called function exits, the copy goes out of scope and is destroyed.

In many cases, this is not a problem, but supposing the object contains a pointer to some dynamically allocated memory? The bitwise copy will contain exactly the same address, and if, as it should, the destructor for the object frees that memory, the bitwise copy will free memory that the original object in the calling routine still "thinks" is allocated - a disaster waiting to happen.

With a copy constructor, you can create a new dynamic memory element, and copy the contents of the original objects dynamic structures to the new object. When it goes out of scope and is destroyed, it is the copies that are deallocated and the original object is not damaged.

There are other reasons, but in general it is wise to build a copy constructor for all but the simplest of classes.