Chapter 6
MORE ENCAPSULATION
The purpose of this chapter is to illustrate how to use some of the
traditional aspects of C or C++ with classes and objects. Pointers
to an object as well as pointers within an object will be
illustrated. Arrays embedded within an object, and an array of
objects will be illustrated. Since objects are simply another C
data construct, all of these things are possible and can be used
if needed.
In order to have a systematic study, we will use the program named
BOXES1.CPP from the last chapter as a starting point and we will
add a few new constructs to it for each example program. You will
recall that it was a very simple program with the class definition,
the class implementation, and the main calling program all in one
file. This was selected as a starting point because we will
eventually make changes to all parts of the program and it will be
convenient to have it all in a single file for illustrative
purposes. It must be kept in mind however that the proper way to
use these constructs is to separate them into the three files as
was illustrated in BOX.HPP, BOX.CPP, and BOXES2.CPP in the last
chapter. This allows the implementor of box to supply the user
with only the interface, namely BOX.HPP. Not giving him the
implementation file named BOX.CPP, is practicing the technique of
information hiding.
As we have said many times, it seems silly to break up such a small
program into three separate files, and it is sort of silly. The
last chapter of this tutorial will illustrate a program large
enough to require dividing the program up into separate files.
AN ARRAY OF OBJECTS
_________________________________________________________________
Examine the file named OBJARRAY.CPP for our ================
first example of an array of objects. This file OBJARRAY.CPP
is identical to the file named BOX1.CPP until we ================
come to line 41 where an array of 4 boxes are
declared.
Recalling the operation of the constructor you will remember that
each of the four box objects will be initialized to the values
defined within the constructor since each box will go through the
constructor as they are declared. In order to declare an array of
objects, a constructor for that object must not require any
parameters. (We have not yet illustrated a constructor with
initializing parameters, but we will in the next program.) This
is an efficiency consideration since it would probably be an error
to initialize all elements of an array of objects to the same
value. We will see the results of executing the constructor when
we compile and execute the file later.
Page 6-1
Chapter 6 - More Encapsulation
Line 46 defines a for loop that begins with 1 instead of the normal
starting index for an array leaving the first object, named
group[0], to use the default values stored when the constructor was
called. You will observe that sending a message to one of the
objects uses the same construct as is used for any object. The
name of the array followed by its index in square brackets is used
to send a message to one of the objects in the array. This is
illustrated in line 47 and the operation of that code should be
clear to you. The other method is called in the output statement
in lines 54 and 55 where the area of the four boxes in the group
array are listed on the monitor.
Another fine point should be pointed out. The integer variable
named index is declared in line 46 and is still available for use
in line 53 since we have not yet left the enclosing block which
begins in line 40 and extends to line 62.
An extra variable was included for illustration, the one named
extra_data in line seven. Since the keyword static is used to
modify this variable in line 7, it is an external variable and only
one copy of this variable will ever exist. All seven objects of
this class share a single copy of this variable which is global to
the objects defined in line 41. Line 20 of the constructor sets
the single global variable to 1 each time an object is declared.
Only one assignment is necessary so the other six are actually
wasted code. To illustrate that there is only one variable shared
by all objects of this class, the method to read its value also
increments it. Each time it is read in lines 57 through 61, it is
incremented and the result of the execution proves that there is
only a single variable shared by all objects of this class. You
will also note that this method is declared within the class
declaration so it will be assembled into the final program as
inline code.
Be sure you understand this program and especially the static
variable, then compile and execute it to see if you get the same
result as listed at the end of the program.
A STRING WITHIN AN OBJECT
_________________________________________________________________
Examine the program named OBJSTRNG.CPP for our ================
first example of an object with an embedded OBJSTRNG.CPP
string. Actually, the object does not have an ================
embedded string, it has an embedded pointer, but
the two work so closely together that we can
study one and understand both.
You will notice that line 7 contains a pointer to char named
line_of_text. The constructor contains an input parameter which
is a pointer to a string and will be assigned to the stored pointer
named line_of_text within the constructor. We could have defined
Page 6-2
Chapter 6 - More Encapsulation
the variable line_of_text as an actual array in the class, then
used strcpy() to copy the string into the object and everything
would have worked the same, but we will leave that as an exercise
for you at the end of this chapter. It should be clear to you that
if a single parameter can be passed to a constructor, any number
can be passed.
You will notice that when the three boxes are declared this time,
we supply a string constant as an actual parameter with each
declaration which is used by the constructor to assign the string
pointer some data to point to. When we call get_area() in lines
48 through 53, we get the message displayed and the area returned.
It would be prudent to put these operations in separate methods
since there is no apparent connection between printing the message
and calculating the area, but it was written this way to illustrate
that it can be done. What this really says is that it is possible
to have a method that has a side effect, the message output to the
monitor, and a return value, the area of the box. However, as we
discussed in chapter 4 when we studied DEFAULT.CPP, the order of
evaluation is sort of funny, so we broke each line into two lines.
After you understand this program, compile and execute it.
AN OBJECT WITH AN INTERNAL POINTER
_________________________________________________________________
The program named OBJINTPT.CPP is our first ================
example program with an embedded pointer which OBJINTPT.CPP
will be used for dynamic allocation of data. ================
In line 7 we declare a pointer to an integer
variable, but it is only a pointer, there is no storage associated
with it. The constructor therefore allocates an integer type
variable on the heap for use with this pointer in line 21. It
should be clear to you that the three objects created in line 45
each contain a pointer which points into the heap to three
different locations. Each object has its own dynamically allocated
variable for its own private use. Moreover each has a value of 112
stored in its dynamically allocated data because line 22 stores
that value in each of the three locations, once for each call to
the constructor.
In such a small program, there is no chance that we will exhaust
the heap, so no test is made for unavailable memory. In a real
production program, it would be expedient to test that the value
of the returned pointer is not NULL to assure that the data
actually did get allocated.
The method named set() has three parameters associated with it and
the third parameter is used to set the value of the new dynamically
allocated variable. There are two messages passed, one to the
small box and one to the large box. As before, the medium box is
left with its default values.
Page 6-3
Chapter 6 - More Encapsulation
The three areas are displayed followed by the three stored values
in the dynamically allocated variables, and we finally have a
program that requires a destructor in order to be completely
proper. If we simply leave the scope of the objects as we do when
we leave the main program, we will leave the three dynamically
allocated variables on the heap with nothing pointing to them.
They will be inaccessible and will therefore represent wasted
storage on the heap. For that reason, the destructor is used to
delete the three variables as their respective objects go out of
existence. In this case, lines 37 and 38 assign values to
variables that will be automatically deleted. Even though these
lines of code really do no good, they are legal statements.
Actually, in this particular case, the three variables will be
automatically reclaimed when we return to the operating system
because all program cleanup is done for us at that time. If this
were a function that was called by another function however, the
heap space would be wasted. This is an illustration of good
programming practice, that of cleaning up after yourself when you
no longer need some dynamically allocated variables.
One other construct should be mentioned once again, that of the
inline method implementations in line 11 and 12. As we mentioned
in chapter 5 and repeated earlier in this chapter, inline functions
can be used where speed is of the utmost in importance since the
code is assembled inline rather than by actually making a method
call. Since the code is defined as part of the declaration, the
system will assemble it inline, and a separate implementation for
these methods is not needed. Remember that if the inline code is
too involved, the compiler is allowed to ignore the inline request
and actually assemble it as a separate method, but it will do it
invisibly to you and will probably not even tell you about it.
Remember that we are interested in using information hiding and
inline code prevents hiding of the implementation, putting it out
in full view. Many times you will be more interested in speeding
up a program than you are in hiding a trivial implementation.
Since most inline methods are trivial, feel free to use the inline
code construct.
Be sure to compile and execute this program.
A DYNAMICALLY ALLOCATED OBJECT
_________________________________________________________________
Examine the file named OBJDYNAM.CPP for our ================
first look at a dynamically allocated object. OBJDYNAM.CPP
This is not any different than any other ================
dynamically allocated object, but an example is
always helpful.
Page 6-4
Chapter 6 - More Encapsulation
In line 39 we declare a pointer to an object of type box and since
it is only a pointer with nothing to point to, we dynamically
allocate an object for it in line 44, with the object being created
on the heap just like any other dynamically allocated variable.
When the object is created in line 44, the constructor is called
automatically to assign values to the two internal storage
variables. Note that the constructor is not called when the
pointer is declared since there is nothing to initialize. It is
called when the object is allocated.
Reference to the components of the object are handled in much the
same way that structure references are made, through use of the
pointer operator as illustrated in lines 50 through 52. Of course
you can use the pointer dereferencing method without the arrow such
as (*point).set(12, 12); as a replacement for line 51 but the arrow
notation is much more universal and should be used. Finally, the
object is deleted in line 54 and the program terminates. If there
were a destructor for this class, it would be called as part of the
delete statement to clean up the object prior to deletion.
You have probably noticed by this time that the use of objects is
not much different from the use of structures. Be sure to compile
and execute this program after you have studied it thoroughly.
AN OBJECT WITH A POINTER TO ANOTHER OBJECT
_________________________________________________________________
The program named OBJLIST.CPP contains an object ===============
with an internal reference to another object of OBJLIST.CPP
its own class. This is the standard structure ===============
used for a singly linked list and we will keep
the use of it very simple in this program.
The constructor contains the statement in line 21 which assigns the
pointer the value of NULL to initialize the pointer. This is a
good idea for all of your programming, don't allow any pointer to
point off into space, but initialize all pointers to something.
By assigning the pointer within the constructor, you guarantee that
every object of this class will automatically have its pointer
initialized. It will be impossible to overlook the assignment of
one of these pointers.
Two additional methods are declared in lines 12 and 13 with the one
in line 13 having a construct we have not yet mentioned in this
tutorial. This method returns a pointer to an object of the box
class. As you are aware, you can return a pointer to a struct in
standard C, and this is a parallel construct in C++. The
implementation in lines 48 through 51 returns the pointer stored
within the object. We will see how this is used when we get to the
actual program.
An extra pointer named box_pointer is declared in the main program
for use later and in line 66 we make the embedded pointer within
Page 6-5
Chapter 6 - More Encapsulation
the small box point to the medium box, and in line 67 we make the
embedded pointer within the medium box point to the large box. We
have effectively generated a linked list with three elements. In
line 69 we make the extra pointer point to the small box.
Continuing in line 70 we use it to refer to the small box and
update it to the value contained in the small box which is the
address of the medium box. We have therefore traversed from one
element of the list to another by sending a message to one of the
objects. If line 70 were repeated exactly as shown, it would cause
the extra pointer to refer to the large box, and we would have
traversed the entire linked list which is only composed of three
elements.
ANOTHER NEW KEYWORD this
_________________________________________________________________
Another new keyword is available in C++, the keyword this. The
word this is defined within any object as being a pointer to the
object in which it is contained. It is implicitly declared as;
class_name *this;
and is initialized to point to the object for which the member
function is invoked. This pointer is most useful when working with
pointers and especially with a linked list when you need to
reference a pointer to the object you are inserting into the list.
The keyword this is available for this purpose and can be used in
any object. Actually the proper way to refer to any variable
within a list is through use of the predefined pointer this, by
writing this->variable_name, but the compiler assumes the pointer,
and we can simplify every reference by omitting the pointer. Use
of the keyword this is not illustrated in a program at this point,
but will be used in one of the larger example programs later in
this tutorial.
You should study this program until you understand it completely
then compile and execute it in preparation for our next example
program.
A LINKED LIST OF OBJECTS
_________________________________________________________________
The next example program in this chapter is ===============
named OBJLINK.CPP and is a complete example of OBJLINK.CPP
a linked list written in object oriented ===============
notation.
This program is very similar to the last one. In fact it is
identical until we get to the main program. You will recall that
in the last program the only way we had to set or use the embedded
pointer was through use of the two methods named point_at_next()
and get_next() which are listed in lines 40 through 51 of the
Page 6-6
Chapter 6 - More Encapsulation
present program. We will use these to build up our linked list
then traverse and print the list. Finally, we will delete the
entire list to free the space on the heap.
In lines 56 to 58 we declare three pointers for use in the program.
The pointer named start will always point to the beginning of the
list, but temp will move down through the list as we create it.
The pointer named box_pointer will be used for the creation of each
box. We execute the loop in lines 61 through 69 to generate the
list where line 62 dynamically allocates a new object of the box
class and line 63 fills it with nonsense data for illustration.
If this is the first element in the list, the start pointer is set
to point to this element, but if elements already exist, the last
element in the list is assigned to point to the new element. In
either case, the temp pointer is assigned to point to the last
element of the list, in preparation for adding another element if
there is another element to be added.
In line 72, the pointer named temp is pointed to the first element
and it is used to increment its way through the list by updating
itself in line 75 during each pass through the loop. When temp has
the value of NULL, which it gets from the last element of the list,
we are finished traversing the list.
Finally, we delete the entire list by starting at the beginning and
deleting one element each time we pass through the loop in lines
79 through 84.
A careful study of the program will reveal that it does indeed
generate a linked list of ten elements, each element being an
object of class box. The length of this list is limited by the
practicality of how large a list we desire to print out, but it
could be lengthened to many thousands of these simple elements
provided you have enough memory available to store them all.
Once again, the success of the dynamic allocation is not checked
as it should be in a correctly written program. Be sure to compile
and execute this example program.
NESTING OBJECTS
_________________________________________________________________
Examine the program named NESTING.CPP for an ===============
example of nesting classes which results in NESTING.CPP
nested objects. A nested object could be ===============
illustrated with your computer in a rather
simple manner. The computer itself is composed
of many items which work together but work entirely differently,
such as a keyboard, a disk drive, and a power supply. The computer
is composed of these very dissimilar items and it is desireable to
discuss the keyboard separately from the disk drive because they
are so different. A computer class could be composed of several
Page 6-7
Chapter 6 - More Encapsulation
objects that are dissimilar by nesting the dissimilar classes
within the computer class.
If however, we wished to discuss disk drives, we may wish to
examine the characteristics of disk drives in general, then examine
the details of a hard disk, then the differences of floppy disks,
and finally bernolli drives. This would involve inheritance
because much of the data about all three drives could be
characterized and applied to the generic disk drive then used to
aid in the discussion of the other three. We will study
inheritance in the next three chapters, but for now we will look
at the embedded or nested class.
This example program contains a class named box which contains an
object of another class embedded within it in line 16, the
mail_info class. This object is available for use only within the
class implementation of box because that is where it is defined.
The main program has objects of class box defined but no objects
of class mail_info, so the mail_info class cannot be referred to
in the main program. In this case, the mail_info class object is
meant to be used internally to the box class and one example is
given in line 21 where a message is sent to the label.set() method
to initialize the variables. Additional methods could be used as
needed, but these are given as an illustration of how they can be
called.
Of prime importance is the fact that there are never any objects
of the mail_info class declared directly in the main program, they
are inherently declared when the enclosing objects of class box are
declared. Of course objects of the mail_info class could be
declared and used in the main program if needed, but they are not
in this example program. In order to be complete, the box class
should have one or more methods to use the information stored in
the object of the mail_info class. Study this program until you
understand the new construct, then compile and execute it.
If the class and the nested classes require parameter lists for
their respective constructors an initialization list can be given.
This is a very advanced technique, and it is rarely needed, so it
is beyond the scope of this tutorial. See Bjarne Stroustrup's book
for further details if you need more information on the use of this
technique.
OPERATOR OVERLOADING
_________________________________________________________________
The example file named OPOVERLD.CPP contains ================
examples of overloading operators. This allows OPOVERLD.CPP
you to define a class of objects and redefine ================
the use of the normal operators. The end result
is that objects of the new class can be used in
as natural a manner as the predefined types. In fact, they seem
to be a part of the language rather than your own add-on.
Page 6-8
Chapter 6 - More Encapsulation
In this case we overload the + operator and the * operator, with
the declarations in lines 10 through 12, and the definitions in
lines 16 through 40. The methods are declared as friend functions
so we can use the double parameter functions as listed. If we did
not use the friend construct, the function would be a part of one
of the objects and that object would be the object to which the
message was sent. Including the friend construct allows us to
separate this method from the object and call the method with infix
notation. Using this technique, it can be written as object1 +
object2 rather than object1.operator+(object2). Also, without the
friend construct we could not use an overloading with an int type
variable for the first parameter because we can not send a message
to an integer type variable such as int.operator+(object). Two of
the three operator overloadings use an int for the first parameter
so it is necessary to declare them as friend functions.
The header in line 16 illustrates the first overloading where the
+ operator is overloaded by giving the return type followed by the
keyword operator and the operator we wish to overload. The two
formal parameters and their types are then listed in the
parentheses and the normal function operations are given in the
implementation of the function in lines 18 through 21. There is
nothing unusual about this implementation, it should be easily
understood by you at this point. For purposes of illustration,
some silly mathematics are performed in the method implementation,
but any desired operations can be done.
The biggest difference occurs in line 56 where this method is
called by using the infix notation instead of the usual message
sending format. Since the variables small and medium are objects
of the box class, the system will search for a way to use the +
operator on two objects of class box and will find it in the
overloaded operator+ method we have just discussed. The operations
within the method implementation can be anything we need them to
be, and they are usually much more meaningful than the silly math
included here.
In line 58 we ask the system to add an int type constant to an
object of class box, so the system finds the other overloading of
the + operator beginning in line 25 to perform this operation.
Also in line 60 we ask the system to use the * operator to do
something to an int constant and an object of class box, which it
satisfies by finding the method in lines 34 through 40. Note that
it would be illegal to attempt to use the * operator the other way
around, namely large * 4 since we did not define a method to use
the two types in that order. Another overloading could be given
with reversed types, and we could use the reverse order in a
program.
You will notice that when using operator overloading, we are also
using function name overloading since some of the function names
are the same.
Page 6-9
Chapter 6 - More Encapsulation
When we use operator overloading in this manner, we actually make
our programs look like the class is a natural part of the language
since it is integrated into the language so well. C++ is therefore
an extendible language and can be molded to fit the mechanics of
the problem at hand.
OPERATOR OVERLOADING CAVEATS
_________________________________________________________________
Each new topic we study has its pitfalls which must be warned
against and the topic of operator overloading seems to have the
record for pitfalls since it is so prone to misuse and has several
problems. The overloading of operators is only available for
classes, you cannot redefine the operators for the predefined
simple types. This would probably be very silly anyway since the
code could be very difficult to read if you changed some of them
around.
The logical and (&&) and the logical or (||) operators can be
overloaded for the classes you define, but they will not operate
as short circuit operators. All members of the logical
construction will be evaluated with no regard concerning the
outcome. Of course the normal predefined logical operators will
continue to operate as short circuit operators as expected, but not
the overloaded ones.
If the increment (++) or decrement (--) operators are overloaded,
the system has no way of telling whether the operators are used as
preincrement or postincrement operators. Which method is used is
implementation dependent, so you should use them in such a way that
it doesn't matter which is used.
Be sure to compile and execute OPOVERLD.CPP before continuing on
to the next example program.
FUNCTION OVERLOADING IN A CLASS
_________________________________________________________________
Examine the program named FUNCOVER.CPP for an ================
example of function name overloading within a FUNCOVER.CPP
class. In this program the constructor is ================
overloaded as well as one of the methods to
illustrate what can be done.
This file illustrates some of the uses of overloaded names and a
few of the rules for their use. You will recall that the function
selected is based on the number and types of the formal parameters,
and the type of the return value.
In this case there are three constructors. The constructor which
is actually called is selected by the number and types of the
parameters in the definition. In line 77 of the main program the
Page 6-10
Chapter 6 - More Encapsulation
three objects are declared, each with a different number of
parameters and inspection of the results will indicate that the
correct constructor was called based on the number of parameters.
Since a constructor does not have a return value, according to the
definition of the C++ language, it cannot be a part of the
selection criteria of a constructor.
In the case of the other overloaded methods, the number and type
of parameters is clearly used to select the proper method. You
will notice that one method uses a single integer and another uses
a single float type variable, but the system is able to select the
correct one. As many overloadings as desired can be used provided
that all of the parameter patterns are unique.
You may be thinking that this is a silly thing to do but it is, in
fact, a very important topic. Throughout this tutorial we have
been using an overloaded operator and you haven't been the least
confused over it. It is the cout operator which operates as an
overloaded function since the way it outputs data is a function of
the type of its input variable or the field we ask it to display.
Many programming languages have overloaded output functions so you
can output any data with the same function name.
Be sure to compile and execute FUNCOVER.CPP.
SEPARATE COMPILATION
_________________________________________________________________
Separate compilation is available with C++ and it follows the
identical rules as given for ANSI-C separate compilation. As
expected, separately compiled files can be linked together.
However, since classes are used to define objects, the nature of
C++ separate compilation is considerably different from that used
for ANSI-C. This is because the classes used to create the objects
are not considered as external variables, but as included classes.
This makes the overall program look different from a pure ANSI-C
program.
As we have been declaring and using classes in these last two
chapters, we have been including the classes with the #include
preprocessor directive rather than by using the ANSI-C extern
keyword. Your programs will take on a different appearance as you
gain experience in C++.
WHAT SHOULD BE THE NEXT STEP?
_________________________________________________________________
At this point you have learned enough C++ to write meaningful
programs and it would be to your advantage to stop studying and
begin using the knowledge you have gained. Because C++ is an
extension to ANSI-C, it can be learned in smaller pieces than would
be required if you are learning a completely new language. You
Page 6-11
Chapter 6 - More Encapsulation
have learned enough to study and completely understand the example
program given in chapter 12, the Flyaway adventure game. You
should begin studying this program now.
One of your biggest problems is learning to think in terms of
object oriented programming. It is not a trivial problem if you
have been programming in procedural languages for any significant
length of time. However, it can be learned by experience, so you
should begin trying to think in terms of classes and objects
immediately. Your first project should use only a small number of
objects and the remainder of code can be completed in standard C
using procedural methods. As you gain experience, you will write
more of the code for any given project using classes and objects
but every project will eventually be completed in procedural code.
After you have programmed for a while using the techniques covered
up to this point in the tutorial, you can continue on to the next
few chapters which will discuss inheritance and virtual functions.
PROGRAMMING EXERCISES
_________________________________________________________________
1. Rewrite OBJSTRNG.CPP to use an array instead of only a pointer
to the array and copy the string in the constructor. Add
another method to allow you to change the value stored in the
string by sending a message from the main program and use the
new method to change the value stored in each of the three
objects.
2. Modify OBJDYNAM.CPP to make the objects named small and medium
pointers, then dynamically allocate them prior to using them.
3. Modify the loop in line 61 of OBJLINK.CPP so that the loop
will store 1000 elements in the list before stopping. You
will probably wish to remove the printout from line 74 so the
program will stop in a reasonable time. You may also get an
integer overflow indication if you send a message to
get_area() with such large numbers. That will depend upon
your compiler. You should also add a test to assure that the
memory did not become exhausted after each dynamic allocation.
Page 6-12