Chapter 5 - Functions, variables, and prototypes OUR FIRST USER DEFINED FUNCTION Load and examine the file SUMSQRES.C for an example of a C program with functions. Actually this is not the first function we have encountered because the "main" program we have been using all along is technically a function, as is the "printf" function. The "printf" function is a library function that was supplied with your compiler. Notice the executable part of this program. It begins with a line that simply says "header()", which is the way to call any function. The parentheses are required because the C compiler uses them to determine that it is a function call and not simply a misplaced variable. When the program comes to this line of code, the function named "header" is called, its statements are executed, and control returns to the statement following this call. Continuing on we come to a "for" loop which will be executed 7 times and which calls another function named "square" each time through the loop, and finally a function named "ending" will be called and executed. For the moment ignore the "index" in the parentheses of the call to "square". We have seen that this program therefore calls a header, 7 square calls, and an ending. Now we need to define the functions. DEFINING THE FUNCTIONS Following the main program you will see another program that follows all of the rules set forth so far for a "main" program except that it is named "header()". This is the function which is called from within the main program. Each of these statements are executed, and when they are all complete, control returns to the main program. The first statement sets the variable "sum" equal to zero because we will use it to accumulate a sum of squares. Since the variable "sum" is defined as an integer type variable prior to the main program, it is available to be used in any of the following functions. It is called a "global" variable, and it's scope is the entire program and all functions. More will be said about the scope of variables near the end of this chapter. The next statement outputs a header message to the monitor. Program control then returns to the main program since there are no additional statements to execute in this function. It should be clear to you that the two executable lines from this function could be moved to the main program, replacing the header call, and the program would do exactly the same thing that it does as it is now written. This does not minimize the value of functions, it merely illustrates Page 31 Chapter 5 - Functions, variables, and prototypes the operation of this simple function in a simple way. You will find functions to be very valuable in C programming. PASSING A VALUE TO A FUNCTION Going back to the main program, and the "for" loop specifically, we find the new construct from the end of the last lesson used in the last part of the "for" loop, namely the "index++". You should get used to seeing this, as you will see it a lot in C programs. In the call to the function "square", we have an added feature, namely the variable "index" within the parentheses. This is an indication to the compiler that when you go to the function, you wish to take along the value of index to use in the execution of that function. Looking ahead at the function "square", we find that another variable name is enclosed in its parentheses, namely the variable "number". This is the name we prefer to call the variable passed to the function when we are in the function. We can call it anything we wish as long as it follows the rules of naming an identifier. Since the function must know what type the variable is, it is defined following the function name but before the opening brace of the function itself. Thus, the line containing "int number;" tells the function that the value passed to it will be an integer type variable. With all of that out of the way, we now have the value of index from the main program passed to the function "square", but renamed "number", and available for use within the function. This is the "classic" style of defining function variables and has been in use since C was originally defined. A newer method is gaining in popularity due to its many benefits and will be discussed later in this chapter. Following the opening brace of the function, we define another variable "numsq" for use only within the function itself, (more about that later) and proceed with the required calculations. We set "numsq" equal to the square of number, then add numsq to the current total stored in "sum". Remember that "sum += numsq" is the same as "sum = sum + numsq" from the last lesson. We print the number and its square, and return to the main program. MORE ABOUT PASSING A VALUE TO A FUNCTION When we passed the value of "index" to the function, a little more happened than meets the eye. We did not actually pass the value of index to the function, we actually passed a copy of the value. In this way the original value is protected from accidental corruption by a called function. We could have modified the variable Page 32 Chapter 5 - Functions, variables, and prototypes "number" in any way we wished in the function "square", and when we returned to the main program, "index" would not have been modified. We thus protect the value of a variable in the main program from being accidentally corrupted, but we cannot return a value to the main program from a function using this technique. We will find a well defined method of returning values to the main program or to any calling function when we get to arrays and another method when we get to pointers. Until then the only way you will be able to communicate back to the calling function will be with global variables. We have already hinted at global variables above, and will discuss them in detail later in this chapter. Continuing in the main program, we come to the last function call, the call to "ending". This call simply calls the last function which has no local variables defined. It prints out a message with the value of "sum" contained in it to end the program. The program ends by returning to the main program and finding nothing else to do. Compile and run this program and observe the output. NOW TO CONFESS A LITTLE LIE I told you a short time ago that the only way to get a value back to the main program was through use of a global variable, but there is another way which we will discuss after you load and display the file named SQUARES.C. In this file we will see that it is simple to return a single value from a called function to the calling function. But once again, it is true that to return more than one value, we will need to study either arrays or pointers. In the main program, we define two integers and begin a "for" loop which will be executed 8 times. The first statement of the "for" loop is "y = squ(x);", which is a new and rather strange looking construct. From past experience, we should have no trouble understanding that the "squ(x)" portion of the statement is a call to the "squ" function taking along the value of "x" as a variable. Looking ahead to the function itself we find that the function prefers to call the variable "in" and it proceeds to square the value of "in" and call the result "square". Finally, a new kind of a statement appears, the "return" statement. The value within the parentheses is assigned to the function itself and is returned as a usable value in the main program. Thus, the function call "squ(x)" is assigned the value of the square and returned to the main program such that "y" is then set equal to that value. If "x" were therefore assigned the value 4 prior to this call, "y" would then be set to 16 as a result of this line of code. Page 33 Chapter 5 - Functions, variables, and prototypes Another way to think of this is to consider the grouping of characters "squ(x)" as another variable with a value that is the square of "x", and this new variable can be used any place it is legal to use a variable of its type. The values of "x" and "y" are then printed out. To illustrate that the grouping of "squ(x)" can be thought of as just another variable, another "for" loop is introduced in which the function call is placed in the print statement rather than assigning it to a new variable. One last point must be made, the type of variable returned must be defined in order to make sense of the data, but the compiler will default the type to integer if none is specified. If any other type is desired, it must be explicitly defined. How to do this will be demonstrated in the next example program. Compile and run this program which also uses the "classic" method of defining function variables. FLOATING POINT FUNCTIONS Load the program FLOATSQ.C for an example of a function with a floating point type of return. It begins by defining a global floating point variable we will use later. Then in the "main" part of the program, an integer is defined, followed by two floating point variables, and then by two strange looking definitions. The expressions "sqr()" and "glsqr()" look like function calls and they are. This is the proper way in C to define that a function will return a value that is not of the type "int", but of some other type, in this case "float". This tells the compiler that when a value is returned from either of these two functions, it will be of type "float". This is, once again, the "classic" method of defining functions. Now refer to the function "sqr" near the center of the listing and you will see that the function name is preceded by the name "float". This is an indication to the compiler that this function will return a value of type "float" to any program that calls it. The function is now compatible with the call to it. The line following the function name contains "float inval;", which indicates to the compiler that the variable passed to this function from the calling program will be of type "float". The next function, namely "glsqr", will also return a "float" type variable, but it uses a global variable for input. It also does the squaring right within the return Page 34 Chapter 5 - Functions, variables, and prototypes statement and therefore has no need to define a separate variable to store the product. The overall structure of this program should pose no problem and will not be discussed in any further detail. As is customary with all example programs, compile and run this program. THERE IS A BUG IN THE FIRST VERSION OF TURBO C When you run this program, using version 1.0 of Turbo C, you will find that the function named "sqr()" will return a value of zero. This is because it receives a zero from the calling program. The program is written correctly but the compiler has a bug in it. A call to Borland resulted in a fix by simply using the "modern" method of function definition rather than the "classic" which has been used to this point. As you read articles on C, you will see programs written in the "classic" style, so you need to be capable of reading them. It would be highly recommended, however, that you learn and use the "modern" method which will be covered shortly in this tutorial. We will return to this program and show how to fix it so that it will work in spite of the bug. There will be times that you will have a need for a function to return a pointer as a result of some calculation. There is a way to define a function so that it does just that. We haven't studied pointers yet, but we will soon. This is just a short preview of things to come. SCOPE OF VARIABLES Load the next program, SCOPE.C, and display it for a discussion of the scope of variables in a program. Ignore the 4 statements in lines 2 through 5 of this program for a few moments, and we will discuss them later. The first variable defined is a global variable "count" which is available to any function in the program since it is defined before any of the functions. In addition, it is always available because it does not come and go as the program is executed. (That will make sense shortly.) Farther down in the program, another global variable named "counter" is defined which is also global but is not available to the main program since it is defined following the main program. A global variable is any variable that is defined outside of any function. Note that both of these variables are sometimes referred to as external variables because they are external to any functions. Page 35 Chapter 5 - Functions, variables, and prototypes Return to the main program and you will see the variable "index" defined as an integer. Ignore the word "register" for the moment. This variable is only available within the main program because that is where it is defined. In addition, it is an "automatic" variable, which means that it only comes into existence when the function in which it is contained is invoked, and ceases to exist when the function is finished. This really means nothing here because the main program is always in operation, even when it gives control to another function. Another integer is defined within the "for" braces, namely "stuff". Any pairing of braces can contain a variable definition which will be valid and available only while the program is executing statements within those braces. The variable will be an "automatic" variable and will cease to exist when execution leaves the braces. This is convenient to use for a loop counter or some other very localized variable. MORE ON "AUTOMATIC" VARIABLES Observe the function named "head1" in line 26 which looks a little funny because of "void" being used twice. The purpose of the uses of the word "void" will be explained shortly. The function contains a variable named "index", which has nothing to do with the "index" of the main program, except that both are automatic variables. When the program is not actually executing statements in this function, this variable named "index" does not even exist. When "head1" is called, the variable is generated, and when "head1" completes its task, the variable "index" is eliminated completely from existence. Keep in mind however that this does not affect the variable of the same name in the main program, since it is a completely separate entity. Automatic variables therefore, are automatically generated and disposed of when needed. The important thing to remember is that from one call to a function to the next call, the value of an automatic variable is not preserved and must therefore be reinitialized. WHAT ARE STATIC VARIABLES? An additional variable type must be mentioned at this point, the "static" variable. By putting the reserved word "static" in front of a variable declaration within a function, the variable or variables in that declaration are static variables and will stay in existence from call to call of the particular function. By putting the same reserved word in front of an external variable, one outside of any function, it makes the Page 36 Chapter 5 - Functions, variables, and prototypes variable private and not accessible to use in any other file. This implies that it is possible to refer to external variables in other separately compiled files, and that is true. Examples of this usage will be given in chapter 14 of this tutorial. USING THE SAME NAME AGAIN Refer to the function named "head2". It contains another definition of the variable named "count". Even though "count" has already been defined as a global variable, it is perfectly all right to reuse the name in this function. It is a completely new variable that has nothing to do with the global variable of the same name, and causes the global variable to be unavailable in this function. This allows you to write programs using existing functions without worrying about what names were used for variables in the functions because there can be no conflict. You only need to worry about the variables that interface with the functions. WHAT IS A REGISTER VARIABLE? Now to fulfill a promise made earlier about what a register variable is. A computer can keep data in a register or in memory. A register is much faster in operation than memory but there are very few registers available for the programmer to use. If there are certain variables that are used extensively in a program, you can designate that those variables are to be stored in a register if possible in order to speed up the execution of the program. Turbo C allows you to use 2 register variables and will ignore additional requests if you request more than 2. Turbo C allows you to use register variables for short int and int types along with their unsigned counterparts. It also allows you to use register storage for two byte pointers which we will study in about 3 chapters. WHERE DO I DEFINE VARIABLES? Now for a refinement on a general rule stated earlier. When you have variables brought to a function as arguments to the function, and you are using the "classic" style of programming, they are defined immediately after the function name and prior to the opening brace for the program. Other variables used in the function are defined at the beginning of the function, immediately following the opening brace of the function, and before any executable statements. Page 37 Chapter 5 - Functions, variables, and prototypes WHAT IS PROTOTYPING? A prototype is a "model" of the real thing and when programming in Turbo C, you have the ability to define a "model" of each function for the compiler. The compiler can then use the "model" to check each of your calls to the function and determine if you have used the correct number of arguments in the function call and if they are of the correct type. By using prototypes, you let the compiler do some additional error checking for you. The ANSI standard for C should be released late in 1987, and will contain prototyping as part of its recommended standard. Every good C compiler will have prototyping available, so you should learn to use it. Returning to lines 3, 4, and 5 in SCOPE.C, we have the prototypes for the three functions contained within the program. The first "void" in each line tells the compiler that these particular functions do not return a value, so that the compiler would flag the statement index = head1(); as an error because nothing is returned to assign to the variable "index". The word "void" within the parentheses tells the compiler that this function requires no parameters and if a variable were included, it would be an error and the compiler would issue a warning message. If you wrote the statement head1(index);, it would be a error. This allows you to use type checking when programming in C in much the same manner that it is used in Pascal, Modula 2, or Ada. You should enable prototype checking with Turbo C at this point to see how well it works. In the "Option" menu select "Compiler", then "Errors", followed by "Less common errors", then turn on "F: Call to function with no prototype" by hitting the "F" key. The program SCOPE.C will not produce any warnings, but RECURSON.C which is next will result in some warning messages. Line 2 of SCOPE.C tells the system to go to the include files and get the file named STDIO.H which contains the prototypes for the standard input and output functions so they can be checked for proper variable types. Don't worry about the "include" yet, it will be covered in detail later in this tutorial. Compile and run this program and, when you return to the Integrated Environment, observe that there are no warning messages. Page 38 Chapter 5 - Functions, variables, and prototypes STANDARD FUNCTION LIBRARIES Every compiler comes with some standard predefined functions which are available for your use. These are mostly input/output functions, character and string manipulation functions, and math functions. We will cover most of these in subsequent chapters. Prototypes are defined for you by the Turbo C compiler writer for all of the functions that are included with your compiler. A few minutes spent studying your Turbo C Reference Guide will give you an insight in where the prototypes are defined for each of the functions. In addition, most compilers have additional functions predefined that are not standard but allow the programmer to get the most out of his particular computer. In the case of the IBM-PC and compatibles, most of these functions allow the programmer to use the BIOS services available in the operating system, or to write directly to the video monitor or to any place in memory. These will not be covered in any detail as you will be able to study these unique aspects of the Turbo C compiler on your own. The entire Turbo C Reference Guide is devoted to defining functions. Many of these kinds of functions are used in the example programs in chapter 14. WHAT IS RECURSION? Recursion is another of those programming techniques that seem very intimidating the first time you come across it, but if you will load and display the example program named RECURSON.C, we will take all of the mystery out of it. This is probably the simplest recursive program that it is possible to write and it is therefore a stupid program in actual practice, but for purposes of illustration, it is excellent. Recursion is nothing more than a function that calls itself. It is therefore in a loop which must have a way of terminating. In the program on your monitor, the variable "index" is set to 8, and is used as the argument to the function "count_dn". The function simply decrements the variable, prints it out in a message, and if the variable is not zero, it calls itself, where it decrements it again, prints it, etc. etc. etc. Finally, the variable will reach zero, and the function will not call itself again. Instead, it will return to the prior time it called itself, and return again, until finally it will return to the main program and will return to DOS. Page 39 Chapter 5 - Functions, variables, and prototypes For purposes of understanding you can think of it as having 8 copies of the function "count_dn" available and it simply called all of them one at a time, keeping track of which copy it was in at any given time. That is not what actually happened, but it is a reasonable illustration for you to begin understanding what it was really doing. WHAT DID IT DO? A better explanation of what actually happened is in order. When you called the function from itself, it stored all of the variables and all of the internal flags it needs to complete the function in a block somewhere. The next time it called itself, it did the same thing, creating and storing another block of everything it needed to complete that function call. It continued making these blocks and storing them away until it reached the last function when it started retrieving the blocks of data, and using them to complete each function call. The blocks were stored on an internal part of the computer called the "stack". This is a part of memory carefully organized to store data just as described above. It is beyond the scope of this tutorial to describe the stack in detail, but it would be good for your programming experience to read some material describing the stack. A stack is used in nearly all modern computers for internal housekeeping chores. In using recursion, you may desire to write a program with indirect recursion as opposed to the direct recursion described above. Indirect recursion would be when a function "A" calls the function "B", which in turn calls "A", etc. This is entirely permissible, the system will take care of putting the necessary things on the stack and retrieving them when needed again. There is no reason why you could not have three functions calling each other in a circle, or four, or five, etc. The C compiler will take care of all of the details for you. The thing you must remember about recursion is that at some point, something must go to zero, or reach some predefined point to terminate the loop. If not, you will have an infinite loop, and the stack will fill up and overflow, giving you an error and stopping the program rather abruptly. Compile and run this program. If you left the prototype checking on from the last program you will find several "warnings" in the message window when you return to the Integrated Environment. Study these for a few minutes. One of the suggested exercises at the end of this Page 40 Chapter 5 - Functions, variables, and prototypes chapter is to modify this program to eliminate the prototype warnings. ANOTHER EXAMPLE OF RECURSION The program named BACKWARD.C is another example of recursion, so load it and display it on your screen. This program is similar to the last one except that it uses a character array. Each successive call to the function named "forward_and_backwards" causes one character of the message to be printed. Additionally, each time the function ends, one of the characters is printed again, this time backwards as the string of recursive function calls is retraced. This program uses the "modern" method of function definition and includes full prototype definitions. The "modern" method of function definition moves the types of the variables into the parentheses along with the variable names themselves. The final result is that the line containing the function name looks more like the corresponding line in Pascal, Modula 2, or Ada. Don't worry about the character array defined in line 9 or the other new material presented here. After you complete chapter 7 of this tutorial, this program will make sense. It was felt that introducing a second example of recursion was important so this file is included here. Compile and run this program with prototype checking on, and observe the results. HOW TO WORK AROUND THE TURBO C (v1.0) BUG Load and display the program named FLOATSQ2.C which is an exact copy of the program FLOATSQ.C which we considered earlier with prototyping added. The return of the erroneous zeros is now repaired and the correct values are returned. Apparently, when the coders of Turbo C at Borland were testing this compiler, they used prototyping and never found this bug. The use of prototyping is a good practice for all C programmers to get into. Several things should be mentioned about this program. First, the word "float" at the beginning of lines 27 and 35 indicate to the compiler that these functions are functions that return "float" type values. Also, since the prototypes are given before "main", the functions are not required to be identified in line 12 as they were in line 7 of FLOATSQ.C earlier in this chapter. They can be included in line 12, but they are not required to be. Page 41 Chapter 5 - Functions, variables, and prototypes Notice also that the type of the variable "inval" is included within the parentheses in line 27. It would be very educational for you to modify this program so that you included a call to "sqr" with a variable of type "int" within the parentheses to see what kind of a warning you would get. Do the same thing in the program without prototype checking, FLOATSQ.C. PROGRAMMING EXERCISES 1. Rewrite TEMPCONV.C, from an earlier chapter, and move the temperature calculation to a function. 2. Write a program that writes your name on the monitor 10 times by calling a function to do the writing. Move the called function ahead of the "main" function to see if the Turbo C compiler will allow it. 3. Add prototyping to the program named RECURSON.C to eliminate the warnings. Page 42