Revised: February 2, 2007
By Richard G. Baldwin
File: Pf00150.htm
Practice Text
Another browser window
I recommend that you open another copy of this document in a separate browser window so that you can view the code and the discussion of that code at the same time.
C++, Java, C#, and some other modern programming languages make heavy use of a concept that we refer to as type, or data type. We refer to those languages as type-sensitive languages.
Not all languages are type-sensitive languages. In particular, some languages hide the concept of type from the programmer and automatically deal with type issues behind the scenes. The web programming language known as PHP is one such language.
So, what do we mean by type?
One analogy that comes to my mind is international currency. For example, many years ago, I spent a little time in Japan and quite a long time on an island named Okinawa (I believe that Okinawa is now part of Japan but that wasn't the case when I was there).
Types of currency
At that time, as now, the type of currency used in the United States was the dollar. The type of currency used in Japan was the yen, and the type of currency used on the island of Okinawa was also the yen. However, even though two of these currencies had the same name, they were different types of currency, as determined by the value relationships between them.
The exchange rate
As I recall, at that time, the exchange rate between the Japanese yen and the U.S. dollar was 360 yen to the dollar. The exchange rate between the Okinawan yen and the U.S. dollar was 120 yen to the dollar.
This suggests that the exchange rate between the Japanese yen and the Okinawan yen would have been three Japanese yen to one Okinawan yen.
Analogous to different types of data
So, why am I telling you this? I am telling you this to illustrate the concept that different types of currency are roughly analogous to different data types in programming.
Purchasing transactions were type sensitive
In particular, because these were three different types of currency, the differences in the types had to be taken into account in any purchasing transaction to determine the price in that particular currency. In other words, the purchasing process was sensitive to the type of currency being used for the purchase (type sensitive).
Different types of data
Type-sensitive programming languages deal with different types of data.
Whole-number types
Some data types involve whole numbers only (no fractional parts are allowed).
Fractional parts
Other data types involve numbers with fractional parts.
True and false
Some data types conceptually have nothing to do with numeric values, but conceptually deal with the concept of true or false.
Letters, numbers, and other characters
Some types deal with the concept of the individual letters of the alphabet, individual numeric characters, and individual punctuation characters (or groups of such characters known as strings).
Type specification
For every different type of data used with a particular programming language, there is a specification somewhere that defines two important characteristics of the type:
What do I mean by an instance of a type?
Think of the type specification as being analogous to the plan or blueprint for a model airplane. Assume that you build three model airplanes from the same set of plans. You will have created three instances of the plans.
We might say that an instance is the physical manifestation of a plan or a type.
Using mixed types
Somewhat secondary to this specification, but also extremely important, is a set of rules that define what happens when you perform an operation involving mixed types (such as making a purchase using some yen currency in combination with some dollar currency).
The short data type
There is a data type in C++ known as short.
Assuming that the short type is maintained as a 16-bit two's complement entity, if you have an instance of the short type, the set of all possible values that you can store in that instance is the set of all the whole numbers ranging from -32,768 to +32,767.
There is also a type known as unsigned short in which you can store all the whole numbers ranging from 0 to 65,535.
65,536 different combinations
This constitutes a set of 65,536 different values, including the value zero.
No other values allowed
No values other than those listed above can be stored in an instance of the type short or the type unsigned short.
For example, you cannot store the value 32,768 in an instance of the type short. If you need to store that value, you will need to use some type other than short.
Kind of like an odometer
This is somewhat analogous to the odometer in your car (the thing that records how many miles the car has been driven). For example, depending on the make and model of car, there is a specified set of values that can appear in the odometer. The value that appears in the odometer depends on how many miles your car has been driven.
00000 to 99999
It is fairly common for an odometer to be able to store and to display the set of all positive values ranging from 00000 to 99999.
If your odometer is designed to store that set of values and if you drive your car more than 99999 miles, it is likely that the odometer will roll over and start back at 00000 after you pass the 99999-mile mark.
Data becomes corrupt
In other words, that particular odometer does not have the ability to store a value of 100,000 miles. Once you pass the 99999-mile mark, the data stored in the odometer is corrupt. There is no way of knowing (based on the value in the odometer alone) how many hundreds of thousands of miles the car has been driven.
Operations on the type named short
Assume that you have two instances of the type short in a program. What are the operations that you can perform on those instances? For example:
There is a well defined set of operations that you are allowed to perform on those instances, and that set of operations is defined in the specification for the type short.
What if you want to do something different?
However, if you want to perform an operation that is not allowed by the type specification, then you will have to find another way to accomplish that purpose.
For example, some programming languages allow you to raise whole-number types to a power (example: four squared, six cubed, nine to the fourth power, etc.).
That operation is not allowed by the C++ specification for the type short. If you need to do that operation with a data value of the short type, you must find another way to do it.
Two major categories of type
C++ data types can be subdivided into two major categories:
What this means is that there is a core component to the language that is always available. Beyond that core component, individual programmers can extend the language to provide new capabilities.
The primitive types discussed in this section are the types that are part of the core language.
A later section will discuss user-defined types that become available when a programmer extends the language.
More subdivision
It seems that when teaching programming, I constantly find myself subdividing topics into sub-topics. I am going to subdivide the topic of Primitive Types into five categories:
Whole-number types
The whole-number types, often called integer types, are probably the easiest to understand. These are types that can be used to represent data without fractional parts.
Applesauce and hamburger
For example, consider purchasing applesauce and hamburger. At the grocery store where I shop, I am allowed to purchase cans of applesauce only in whole-number or integer quantities.
Can purchase integer quantities only
For example, the grocer is happy to sell me one can of applesauce and is even happier to sell me 36 cans of applesauce.
However, she would be very unhappy if I were to open a can of applesauce in the store and attempt to purchase 6.3 cans of applesauce.
Counting doesn't require fractional parts
A count of the number of cans of applesauce that I purchase is somewhat analogous to the concept of whole-number data types. Applesauce is not available in fractional parts of a can at my grocery store. (However, there is another grocery store nearby where you can probably scoop applesauce out of a large container and purchase as much or as little as you want based on weight.)
Fractional pounds of hamburger are available
On the other hand, the grocer is perfectly willing to sell me 6.3 pounds of hamburger. This is somewhat analogous to floating-point data types.
Accommodating applesauce and hamburger in a program
Therefore, if I were writing a program dealing with quantities of applesauce and hamburger, I might elect to:
Different whole-number types
In C++, there are at least four different basic whole-number types:
There are both signed and unsigned variations on each of the basic types.
The four types differ primarily in terms of the range of values that they can accommodate and the amount of computer memory required to store instances of the types.
Differences in operations?
Although there are some subtle differences among the different whole-number types in terms of the operations that you can perform on them, those differences are beyond the scope of this lesson.
Like a strange odometer (-128 to +127)
To form a crude analogy, the signed char type is sort of like a strange odometer in a new (and unusual) car that shows a mileage value of -128 when you first purchase the car.
As you drive the car, the negative values shown on the odometer increment toward zero and then pass zero.
Beyond that point they increment up toward the value of +127. When the value goes beyond 127, it starts over at -128.
Oops, numeric overflow!
When the value passes (or attempts to pass) +127 miles, something bad happens. From that point forward, the value shown on the odometer is not a reliable indicator of the number of miles that the car has been driven.
Floating-point types
Floating-point types are a little more complicated than whole-number types. I found the following definition of floating-point in the Free On-Line Dictionary of Computing at this URL:
"A number representation consisting of a mantissa, M, an exponent, E, and an (assumed) radix (or "base") . The number represented is M*R^E where R is the radix - usually ten but sometimes 2."So what does this really mean?
Assuming a base or radix of 10, I will attempt to explain it using an example.
Consider the following value:
623.57185
I can represent this value in any of the following ways (where * indicates multiplication):
.62357185*1000
6.2357185*100
62.357185*10
623.57185*1
6235.7185*0.1
62357.185*0.01
623571.85*0.001
6235718.5*0.0001
62357185.*0.00001
In other words, I can represent the value as a mantissa (62357185) multiplied by a factor where the purpose of the factor is to represent a left or right shift in the position of the decimal point.
Now consider the factor
Each of the factors shown above represents the value of ten raised to some specific power, such as ten squared, ten cubed, ten raised to the fourth power, etc.
Exponentiation
If we allow the following symbol (^) to represent exponentiation (raising to a power) and allow the following symbol (/) to represent division, then we can write the values for the above factors in the following ways. Note in particular the numbers in blue, which I will refer to later as the exponents.
1000 = 10^+3 = 1*10*10*10
100 = 10^+2 = 1*10*10
10 = 10^+1 = 1*10
1 = 10^+0 = 1
0.1 = 10^-1 = 1/10
0.01 = 10^-2 = 1/(10*10)
0.001 = 10^-3 = 1/(10*10*10)
0.0001 = 10^-4 = 1/(10*10*10*10)
0.00001 = 10^-5 = 1/(10*10*10*10*10)
In the above notation, the term 10^+3 means 10 raised to the third power.
The zeroth power
By definition, the value of any value raised to the zeroth power is 1. (Check this out in your old high-school algebra book.)
The exponent and the factor
Hopefully, at this point you will understand the relationship between the value shown in blue (the exponent) and the factor introduced earlier.
Different ways to represent the same value
Having reached this point, by using substitution, I can rewrite the original set of representations of the value 623.57185 in the following ways. It is very important to for you to understand that these are simply different ways to represent the same value.
.62357185*10^+3
6.2357185*10^+2
62.357185*10^+1
623.57185*10^+0
6235.7185*10^-1
62357.185*10^-2
623571.85*10^-3
6235718.5*10^-4
62357185.*10^-5
A simple change in notation
Finally, by making a simplifying change in notation where I replace *10^ by E I can rewrite the different representations of the value of 623.57185 in the following ways.
.62357185E+3
6.2357185E+2
62.357185E+1
623.57185E+0
6235.7185E-1
62357.185E-2
623571.85E-3
6235718.5E-4
62357185.E-5
Getting the true value
Floating point types represent values as a mantissa containing a decimal point along with an exponent value that tells how many places to shift the decimal point to the left or to the right in order to determine the true value.
Positive exponent values mean that the decimal point should be shifted to the right. Negative exponent values mean that the decimal point should be shifted to the left.
Maintaining fractional parts
One advantage of floating-point types is that they can be used to maintain fractional parts in data values.
Accommodating a very large range of values
Another advantage is that a very large range of values can be represented using a reasonably small amount of computer memory for storage of the values.
Another example
For example (assuming that I counted the number of digits correctly) the following very large value
62357185000000000000000000000000000000.
can be represented as
6.2357185E+37
Similarly, again assuming that I counted the digits correctly, the following very small value
.0000000000000000000000000000062357185
can be represented as
6.2357185E-30
When would you use floating-point?
As examples, you will need to use the floating-point types if you are working in an area where you:
Three floating-point types
C++ supports at least three different floating point types:
(Once again, the number of bits, the number of significant digits, and the range for each type was gleaned from http://www.cplusplus.com/doc/tutorial/tut1-2.html. Hopefully the character used above indicating plus or minus will survive the HTML publishing process.)
These three types differ primarily in terms of the range of values that they can support and the amount of memory required to store an instance of the type.
Values of any of the three types can be either positive or negative as indicated by the ± character.
The character types
There are two character types in C++:
Computers deal only in numeric values. They don't know how to deal directly with the letters of the alphabet and punctuation characters.
Therefore, the values actually stored as instances of these types are numeric values that represent characters.
Purpose of the char type
The char type has two purposes. One purpose is to represent whole numbers in the range from -128 to +127 or 0 to 255 as indicated above.
The second purpose is to represent characters as eight-bit numeric values in order to make it possible to represent the following internally in the computer:
This is accomplished by assigning a numeric value to each character, much as you may have done to create secret codes when you were a child.
The actual character codes used in many computers are shown at http://www.lookuptables.com/.
Purpose of the wchar_t type
The wchar_t type is a wide character type.
This type is designed as a type to store international characters of a two-byte (16-bit) character set.
(This type was recently added by the ANSI-C++ standard and some older compilers may not support it.)
What are the numeric values representing characters?
As long as the characters that you use in your program appear on your keyboard, you usually don't have a need to know the numeric value associated with the different characters.
Representing a character symbolically
Insofar as keyboard characters are concerned, you usually represent a character to the program by surrounding it with apostrophes as follows: 'A'.
Programming tools know how to cross reference that specific character symbol against a character table to obtain the corresponding numeric value.
(A discussion of the representation of characters that don't appear on your keyboard is beyond the scope of this lesson.)
Boolean types
The bool type is conceptually the simplest type supported by C++. It can have only two values:
As a result, it is possible to perform arithmetic operations on variables of type bool (but the resulting code may be very confusing).
The bool type is commonly used in some sort of a test to determine what to do next, such as:
if some test returns true, then
do this
otherwise
do that
The string type
The string type is also relatively new to C++. Although it probably isn't a true primitive type, it has become part of the core language and can be used much the same way primitive types are used provided that the string header file is included in the program using a compiler directive such as:
#include <string>
The purpose of the string type is to store strings of characters, such as in the following statements:
string stringVar = "Hello World";
cout << stringVar << endl;
As mentioned earlier, C++ is an extensible programming language. There is a core component to the language that is always available.
Beyond the core component, different programmers can extend the language in different ways to meet their individual needs.
Creating new types
One of the ways that individual programmers can extend the language is to create new types.
When creating a new type, the user must define:
No magic involved
While this might initially seem like magic, once you get to the heart of the matter, it is really pretty straight forward.
New types are created by combining instances of primitive types and instances of previously defined user-defined types.
An example
For example, the string type, which can be used to represent a person's last name is just a grouping of a bunch of instances of the primitive char or character type.
A user-defined Person type, which could be used to represent a person's first name and last name, might simply be a grouping of two instances of the string type.
The company telephone book
A programmer responsible for producing the company telephone book might create a new type that can be used to store the first and last names along with the telephone number of an individual.
Using this new type, the programmer could create an instance for each employee in the company.
(At this point, let me sneak a little jargon in and tell you that we will be referring to such instances as objects.)
The set of allowable values
The set of allowable values that could be stored in an instance of the type would be all possible combinations of allowable first and last names and allowable telephone numbers.
In order to limit the size of the set of allowable values, for example, restrictions may be established relative to:
A comparison operation
The programmer might define one of the allowable operations for the new type to be:
This operation could be used to sort the set of objects representing all of the employees into alphabetic order.
The set of sorted objects could then be used to print a new telephone book.
A name-change operation
Another allowable operation that the programmer might define would be:
For example when Suzie Smith marries Tom Jones, she might elect to thereafter be known as:
In this case, there would be a need to modify the object that represents her to reflect her newly-elected surname.
(Or perhaps Tom Jones might elect to thereafter be known as Tom Smith or Tom Jones-Smith, in which case it would be necessary to modify the object that represents him.)
A print operation
Still another useful operation might be an operation that formats the contents of each object in a way that is suitable for printing the telephone book.
For example, the printing format might be:
Last name, First name, telephone number
An updated telephone book
Using these operations, the programmer could:
Many user-defined types already exist
Unlike the primitive types which are predefined, I am unable to give you much in the way of specific information about user-defined types, simply because they don't exist until a user defines them.
I can tell you, however, that when you obtain modern object-oriented programming tools for different programming languages, you not only receive the core language containing the primitive types, you also usually receive a large library containing several thousand user-defined types that have already been defined.
(Note that in this case, the user who defines a new type may actually be an employee working for the company from which you obtain your programming tools. This person's job may be to create a large library of standard user-defined types for use by you and other customers.)
A large documentation package is usually also available to help you determine the individual characteristics of those user-defined types.
The most important thing
At this stage in your development as a programmer, the most important thing for you to know about user-defined types is that they are possible.
You can define new types in object-oriented programming.
Unlike earlier procedural programming languages such as C and early versions of Pascal, you are no longer forced to adapt your problem to the available tools.
Rather, you now have the opportunity to extend the tools to make them better suited to solve your problem.
The class definition
One specific C++ mechanism that makes it possible for you to define new types is a mechanism known as the class definition.
In C++, whenever you define a new class, you are at the same time defining a new type. Your new type can be as simple, or as complex as you want it to be.
An object of your new type can contain a very small amount of data, or it can contain a very large amount of data.
The operations that you allow to be performed on an object of your new type can be rudimentary, or they can be very powerful.
It is all up to you
Whenever you define a new class (type) you not only have the opportunity to define the data definition and the operations, you also have a responsibility to do so.
Much to learn and much to do
But, you still have much to learn and much to do before you will need to define new types.
There are a lot of fundamental programming concepts that you will need to learn before you seriously embark on a study involving the definition of new types.
For the present then, simply remember that such a capability exists, and if you work to expand your knowledge of modern object-oriented programming one small step at a time, you will be ready and eager to define new types when you reach that point in your education.
The C++ program shown in Listing 1 illustrates various data types. Variables of several different types are declared, initialized, and displayed.
The program displays the following output on the screen:
numCharVar= ≥ shortVar= 32767 intVar= 2147483647 longVar= 2147483647 overflow= -2147483648 charVar= F wideChVar= 66 floatVar= 6.66667 doubleVar= 66.6667 longDblVar= 666.667 boolVar= 1 boolVar*2= 2 stringVar= Hello World Press any key to continue
(Note that the character shown in the first line of output above is the character from the Extended ASCII character set that is represented by the numeric value 242 (see http://www.lookuptables.com/). This character is not correctly depicted in the comments at the beginning of the program in Listing 1, probably because the text editor in Dev C++ doesn't recognize it. Also note that the ± character used earlier is represented by the numeric value 241.)
The <string> header file
Several different things are worthy of note in Listing 1. For example, the <string> header file must be included in order to use the insertion operator along with cout to display a variable of type string when the program is compiled using Visual Studio 6.0. Otherwise, a compilation error results.
Not required for Dev C++
However, this is not the case when the program is compiled using Dev C++ and run from the command line. With Dev C++, the program will compile and execute successfully even if the <string> header file is not included. This is another example of the kinds of compatibility issues that you will encounter when programming in C++.
Type char displays as a character
The char type is an eight-bit signed or unsigned integer type that can be used to store the eight bits that represent characters in the Extended ASCII character set.
The char type can also be used as a general-purpose eight-bit type on which you can perform arithmetic.
When you use the insertion operator along with cout to display a variable of type char, the character represented by the numeric value is displayed instead of the numeric value itself.
Integer overflow in a twos-complement system
If you add one to a long variable containing the value 2147483647, it overflows and becomes the erroneous value -2147483648.
This is a characteristic of the twos-complement binary number system.
Integer overflow in the positive direction wraps around to the negative side.
Similarly integer overflow (or should I say underflow) in the negative direction wraps around to the positive side.
Representation of type int
Although type int may be represented differently on different platforms and operating systems, it appears to be represented the same as type long on my machine running Windows XP and Visual Studio 6.0. This also appears to be the case when running Dev C++ on my machine.
Wide character display
When the insertion operator is used with cout to display a variable of the wide character type wchar_t, it is the numeric value and not the character that is displayed. This is just the opposite of the result of displaying a variable of type char.
Display of floating-point types
When the insertion operator is used along with cout to display any of the three floating point types, the number of significant digits that is displayed is the same as type float even though types double and long double maintain more significant digits than type float in their internal representation.
Not strictly a boolean type
The boolean type named bool is not strictly a boolean type.
You can initialize it using the literal values true and false.
You can also initialize it with a numeric value where 0 represents false and any non-zero numeric value represents true.
Also, it is possible to perform arithmetic using a variable of type bool but you need to be very careful in this regard. If the variable was initialized using a non-zero numeric value with a fractional part, the arithmetic results are likely to something other than what you might expect.
Type bool displays as 1 or 0
When you use the insertion operator along with cout to display the value of a variable of type bool, the result is either 1 or 0, and is not true or false as you might expect.
Now for the code
Listing 1 shows the code for this program in its entirety.
/*File: Types01.cpp This C++ program illustrates various data types. Variables of several different types are declared, initialized, and displayed. The program displays the following output on the screen (note that the glyph for the first line of text is not correctly represented by this editor): numCharVar= = shortVar= 32767 intVar= 2147483647 longVar= 2147483647 overflow= -2147483648 charVar= F wideChVar= 66 floatVar= 6.66667 doubleVar= 66.6667 longDblVar= 666.667 boolVar= 1 boolVar*2= 2 stringVar= Hello World ************************************************/ #include <iostream> #include <string>//Required for type string using namespace std; class Types01{ public: static void classMain(){ //Instantiate an object in dynamic memory // and save its reference in a pointer // variable. Types01* ptrToObject = new Types01(); //Invoke an instance function on the object. ptrToObject -> doSomething(); }//End classMain function //-------------------------------------------// //An instance function of the Types01 class void doSomething(){ //Declare and initialize the values of // several different types of variables. //Integer types char numCharVar = 242; short shortVar = 32767; int intVar = 2147483647; long longVar = 2147483647; //Purposely overflow a long variable later. long overflow = 2147483647; //Character types char charVar = 'A';//also a numeric type charVar = charVar + 5;//convert to 'F' wchar_t wideChVar = 'B'; //Floating point types float floatVar = 20/3.0; double doubleVar = 200/3.0; long double longDblVar = 2000/3.0; //The string type string stringVar = "Hello World"; //Boolean types bool boolVar = true; //Displayed as a character, not as a numeric // value cout << "numCharVar= " << numCharVar << endl; //Displayed as numeric values cout << "shortVar= " << shortVar << endl; cout << "intVar= " << intVar << endl; cout << "longVar= " << longVar << endl; //Force an integer overflow to occur. cout << "overflow= " << (overflow + 1) << endl; //Displayed as a character, not as a numeric // value cout << "charVar= " << charVar << endl; //Displayed as a numeric value, not as a // character cout << "wideChVar= " << wideChVar << endl; //All floating types displayed with same // number of significant digits as float. cout << "floatVar= " << floatVar << endl; cout << "doubleVar= " << doubleVar << endl; cout << "longDblVar= " << longDblVar << endl; //Displayed as a numeric value, not as true // or false. cout << "boolVar= " << boolVar << endl; //Arithmetic can be performed on bool // variable. cout << "boolVar*2= " << boolVar*2 << endl; //Displayed as string of characters cout << "stringVar= " << stringVar << endl; }//end doSomething function };//End Types01 class //---------------------------------------------// int main(){ Types01::classMain(); return 0; }//end main Listing 1 |
Copyright 2005, Richard G. Baldwin. Reproduction in whole or in part in any form or medium without express written permission from Richard Baldwin is prohibited.
Richard has participated in numerous consulting projects and he frequently provides onsite training at the high-tech companies located in and around Austin, Texas. He is the author of Baldwin's Programming Tutorials, which have gained a worldwide following among experienced and aspiring programmers. He has also published articles in JavaPro magazine.
In addition to his programming expertise, Richard has many years of practical experience in Digital Signal Processing (DSP). His first job after he earned his Bachelor's degree was doing DSP in the Seismic Research Department of Texas Instruments. (TI is still a world leader in DSP.) In the following years, he applied his programming and DSP expertise to other interesting areas including sonar and underwater acoustics.
Richard holds an MSEE degree from Southern Methodist University and has many years of experience in the application of computer technology to real-world problems.
-end-