Lab 3b: Controlling Function Behavior

Introduction

Today's lab explores some C++ statements that we can use to write more sophisticated functions. The exercise consists of two parts: In the first part, we examine a problem whose solution requires more complicated behaviors than we have seen thus far, and introduce the new statements to elicit that behavior. In the second part, we introduce a new program-development tool called the debugger and use it to study the behavior of the new statements.

Part I: The Payroll Problem

Our problem today is to write an interactive payroll program, that a small-business owner might use to simplify computing the payroll at his or her business. For the sake of simplicity, we will assume that all employees at the business are hourly employees.

Recall that object-centered design involves several stages:

1. Identify the behavior we wish the program to exhibit in order to solve the problem.
2. Identify the objects in that behavior.
3. Identify the operations in that behavior.
4. Organize the operations and objects into an algorithm.

1. Identify the behavior we wish the program to exhibit in order to solve the problem.
2. Identify the objects in that behavior.
3. Identify the operations in that behavior.
   If there is no predefined way to perform an operation,
     build a function to perform it.
   If a function is reuseable, store it in a library.
4. Organize the operations and objects into an algorithm.

Today's exercise is to use these stages to develop a program that solves our payroll problem.

Preparing Your Workspace

Design

As always, spending a bit of time planning how to attack our problem will result in a better solution. To do so, we follow the steps of object-centered design.

Behavior. We can begin by visualizing and writing down how we want our program to behave. One approach is to have our program behave something like the following:

   This program computes the payroll interactively.

   To begin, enter the number of employees: 3

   Enter the name, hours and rate for employee 1: Joe 25 5.25
   Joe     131.25

   Enter the name, hours and rate for employee 2: Mary 35 6.15
   Mary    215.25

   Enter the name, hours and rate for employee 3: Sue 45 6.15
   Sue     292.125

   display on the screen a greeting, followed by a prompt for the 
   number of employees, which it should then read from the keyboard.  
   For each employee, our program should then display a prompt for 
   their name, hours and rate of pay.  The program should then
   read these values from the keyboard, and compute and display
   the employee's pay, along with their name.

Objects. If we identify the nouns in this behavioral description, we get the following list:

Description	Type	Kind	Name
The screen	ostream	varying	cout
A greeting	string	constant	--
A prompt for input	string	constant	--
The number of employees	int	varying	numberOfEmployees
The keyboard	istream	varying	cin
An employee's name	string	varying	name
An employee's hours of work	double	varying	hours
An employee's rate of pay	double	varying	rate
An employee's pay	double	varying	pay

From this list, we can build a precise specification of how our program is to behave:

   Input:	The number of employees;
		Each employee's name, hours of work, and rate of pay.
   Output: 	Each employee's name and pay.

Operations. If we identify the operations in our behavioral description, we get this list:

Description	Predefined?	Name	Library?
display a string (greeting, prompts, labels)	yes	<<	iostream
Read an int (numberOfEmployees) from the keyboard	yes	>>	iostream
read a string (name) from the keyboard	yes	>>	iostream
read a double (hours, rate) from the keyboard	yes	>>	iostream
compute an employee's pay, given their hours and rate from the keyboard	no	??	??
display a double (pay) on the screen	yes	<<	iostream
repeat operations 3-6 once for each employee	yes	for	--

As indicated, most of these operations are provided for us by C++. Operations 1-4 and 6 should be familiar by now. There is no predefined C++ capability to perform operation 5, and since accounting for overtime pay makes it non-trivial, we will write a function to perform this operation. Since it seems like an operation that might be useful again some day, we will store it in a library. Finally, operation 7 involves a new C++ statement that we haven't seen before called the for statement, which provides a convenient way to repeat a group of statements a predetermined number of times.

Algorithm. Even without knowing the details of how we will compute the pay, we can organize our operations into an algorithm for our problem:

1. Via cout, display a greeting on the screen, plus a prompt for the number of employees.
2. From cin, read an integer, storing it in numberOfEmployees.
3. For each value empNum in the range 1 to numberOfEmployees:
   1. Via cout, display a prompt for the name, hours and rate of employee empNum.
   2. From cin, read a string, a double and a double, storing them in name, hours and rate.
   3. Compute pay, using hours and rate.
   4. Via cout, display name and pay.
   End loop.

Coding, Testing and Debugging

   int main()
   {
   }

Given the minimal C++ program, we are ready to encode our algorithm using stepwise translation. We therefore begin with the first step:

1. Via cout, display a greeting on the screen,
  plus a prompt for the number of employees.

Coding: This step can be performed using a C++ output statement, which we have seen before:

   cout << Value1 << Value2 << ... << ValueN;

   This program computes the payroll interactively.

   To begin, enter the number of employees:

Before proceeding to the next step, check the correctness of what you just wrote by recompiling your program. The compiler will alert you to any syntax errors in your statement. If an error is listed, you can infer that the error(s) lies in the text ' you just added, since the program was error-free before that. Find your error(s) within those lines and correct them.

When your source program compiles correctly, execute payroll to test that it displays the intended message. If not, the statements you have added contain logic errors. (i.e., the statements you have added are syntactically correct, but they don't accomplish their task correctly.) Compare your program's statements against the output produced by payroll and modify them as needed. When your program is error-free, proceed to the next step of our algorithm.

2. From cin, read an integer, storing it in numberOfEmployees.

Coding: We can encode this step in C++ using an input statement:

   cin >> Var1 >> Var2 >> ... >> VarN; 

Note that we could use an assert() at this point to check that numberOfEmployees is non-negative. However, doing so is unnecessary, thanks to our next step.

3. For each value empNum in the range 1 to numberOfEmployees:
  a. Via cout, display a prompt for the name, hours and rate
      of employee empNum.
  b. From cin, read a string, a double and a double,
      storing them in name, hours and rate.
  c. Compute pay, using hours and rate.
  d. Via cout, display name and pay.
End loop.

Coding: Let's take this step a piece at a time, starting with the outer part (3) and then doing the inner parts (a-d).

3. For each value empNum in the range 1 to numberOfEmployees:
  ...
End loop.

The purpose of step 3 is to count from 1 to the number of employees, and repeat steps a­d that many times. That is, if there are three employees, then steps a­d should be repeated three times.

For situations like this that require repetitive behavior, C++ supplies the for statement. The for statement can use its own local variable, called a loop-control variable, to do the counting. A simplified general form of a for statement that counts from firstValue to lastValue is:

   for (Type loopVar = firstValue; loopVar <= lastValue; loopVar++)
   {
      Statements
   }

1. loopVar is declared and initialized to firstValue.
2. loopVar is compared against lastValue.
3. If the comparison evaluates to true:
   1. Statements get executed.
   2. loopVar++ is executed.
   3. Go to 2.
   Otherwise, control proceeds to the next statement.

In our problem, we must count from 1 to the value stored in numberOfEmployees, using empNum as the name of the loop-control variable. To do so, we can write:

   for (int empNum = 1; empNum <= numberOfEmployees; empNum++)
   {
   }

Note that if the user enters a negative value for numberOfEmployees, the body of the loop will not be executed, because the loop's body is only executed if the condition controlling the loop evaluates to true, and a negative value for numberOfEmployees will make this condition false.

Add this to payroll.cpp and then check its syntax. When the compiler generates no errors, proceed to steps a-d.

3a. Via cout, display a prompt for the name, hours and rate
  of employee empNum. This is a normal output statement, like those we have seen before, except that in addition to displaying a string, it displays the value of our loop-control variable to generate an "employee number." That is, if numberOfEmployees is 3, then our loop will execute three times, so add an output statement to the body of the loop that will generate the following:

   Enter the name, hours and rate for employee 1:
   Enter the name, hours and rate for employee 2:
   Enter the name, hours and rate for employee 3:

Then check the syntax of what you have written using the compiler. When your program translates correctly, run it and enter 3 for the number of employees, to verify that what you have written is free of logic errors.

3b. From cin, read a string, a double and a double,
  storing them in name, hours and rate.

This step can be encoded by adding a normal input statement to the body of the loop, after the output statement, so take a moment to do so.

However, before such a statement will compile correctly, the objects name, hours and rate must be declared. This raises an interesting question: Where should these objects be declared?

If we declare name, hours and rate within the body of the loop, then they will be redeclared anew each execution of the body, wasting time. For the sake of efficiency, they should instead be declared immediately before the for statement. That way, they will still be declared near their first use, but time will not be wasted reprocessinging their declarations every time the loop body executes.

Add the necessary statements to your program to (i) declare name, hours and rate(outside the loop), and (ii) fill these objects with values entered from the keyboard (inside the loop). To declare name as a string, you will need to include the string system file:

   #include <string>

Anytime we program an input operation, we should consider the possibility of human error: what could the user do to foul up our program? There are several possibilities here:

1. The user could enter a negative value for hours.
2. The user could enter a value for hours greater than 168 (the number of hours in a week).
3. The user could enter a negative value for rate.

Since any of these three errors will cause our program to generate incorrect pay amounts, our code should safeguard against them. To do so, add an assert() to the program that only allows the program to continue if hours is non-negative, and hours is 168 or less, and rate is non-negative. Use a single assert() call to do so. (Don't forget its #include directive!)

Then check the syntax of what you have written using the compiler, then run your program and test that your assert() halts the program if any of these three errors occur. When it correctly "catches" these errors, continue.

3c. Compute pay, using hours and rate.

This operation is not predefined, and so we will design and build a function to perform it.

Function Analysis. To compute pay for an arbitrary employee, our function needs their hours and rate. Since these values will differ from employee to employee, our function should receive these values from its caller.

Function Behavior. Our function should receive the hours and rate from its caller. If hours is less than or equal to the number of hours in a work week (40 in the U.S.), then our function should compute the total pay as hours times rate. Otherwise, it should compute the normal pay as the work week times the rate; compute the overtime pay as the overtime hours times the rate; and then compute the total pay as the sum of the overtime pay and the normal pay.

Function Objects. Ignoring nouns like "our function" and "caller", we can identify the following objects in this description:

Description	Type	Kind	Movement	Name
hours of work	double	varying	received	hours
pay rate ($$/hour)	double	varying	received	pay
hours in a work week	double	constant	local	NORMAL_WEEK
total pay	double	varying	returned	totalPay
normal pay	double	varying	local	normalPay
overtime pay	double	varying	local	overtimePay
overtime hours	double	varying	local	hours - NORMAL_WEEK
overtime pay factor	double	constant	local	OVERTIME_FACTOR

For each received object, we must provide a parameter to store that value. We can thus specify the behavior of our function as follows:

   Receive: hours and rate, both doubles.
   Precondition: 0 <= hours and hours <= 168 and 0 <= rate.
   Return: the total pay, a double.

   double ComputePay(double hours, double rate)
   {
   }

Since this seems like a function that might be reuseable some day, we will store its definition in a library named pay. Open pay.cpp, and write this stub there. Since other programs besides payroll.cpp may be calling this function (and not checking the precondition ahead of time), make the first line in the function an assert() call that checks the function's precondition. (Copy-and-paste your assert() from payroll.cpp.)

Before this stub, add an include directive:

   #include "pay.h"

Function Operations. From our behavioral description, we have these operations:

	Description	Defined?	Name	Library?
1	receive hours and rate from caller	yes	function call mechanism	built-in
2	compare hours and NORMAL_WEEK in less-than-or-equal relationship	yes	<=	built-in
3	compute totalPay normally (hours times rate)	yes	*	built-in
4	compute totalPay for overtime
4a	compute normalPay (NORMAL_WEEK times rate)	yes	*	built-in
4b	compute overtimePay (hours minus NORMAL_WEEK times rate times OVERTIME_FACTOR)	yes	-, *, *	built-in
4c	compute totalPay (normalPay plus overtimePay)	yes	+	built-in
5	if (2) is true, perform (3), otherwise perform (4)	yes	if	built-in
6	return totalPay	yes	return	built-in

We have seen each of these operations before, except for operation (5), which introduces a new behavior in which we must select one group of statements or another, but not both. As we shall see shortly, the C++ if statement provides this behavior.

Function Algorithm. We can organize our objects and operations as follows:

1. Receive hours and rate from the caller.
2. If hours <= NORMAL_WEEK
     Compute totalPay = hours * rate.
   Otherwise
     a. Compute normalPay = NORMAL_WEEK * rate.
     b. Compute overtimePay = (hours - NORMAL_WEEK) * rate * OVERTIME_FACTOR.
     c. Compute totalPay = normalPay + overtimePay.
3. Return totalPay.

Function Coding. To encode this algorithm in our ComputePay() stub, we must know the syntax of the C++ if statement. The general pattern is as follows:

   if (BooleanExpression)
   {
      Statements1
   }
   else
   {
      Statements2
   }

For our problem, we want BooleanExpression to perform the comparison described in operation (2). In the stub for ComputePay(), we can add the following if statement that compares hours and NORMAL_WEEK using the less-than-or-equal relationship:

   if (hours <= NORMAL_WEEK)
   {
   }
   else
   {
   }

To check this much, compile pay.cpp, and you should just get the one error for not providing any return value. Continue if this is the case, otherwise fix the other errors.

Coding operations (3) and (4) is straightforward, so add statements to the if's true section to perform (3), and add statements to its false section to perform (4). Note that each section of the if computes totalPay, however the true section uses the formula appropriate for no overtime, while the false section uses the formula appropriate for overtime. Since each section uses totalPay, it must be declared prior to both sections (i.e., before the if statement).

Finish the function by adding a return statement that returns the value of totalPay. Then recheck the correctness of ComputePay(), continuing when it is free of syntax errors.

Now that we have a syntactically correct ComputePay() function to perform step 2c of our program's algorithm, we are ready to code that step by calling our function. In the for loop within payroll.cpp, add an assignment statement with a call to ComputePay(). Don't forget to pass it hours and rate as arguments!

To check the correctness of our call, build payroll.exe which will compile payroll.cpp and link its object file to that of pay.cpp. When what you have written is free of syntax errors, continue to the final step of our algorithm.

3d. Via cout, display name and pay.

This step is easily implemented using an output statement. Take a few moments and add an output statement so that if employee Chris earned $150.75, the following is displayed:

   Pay Chris $150.75

Testing

When you are finished, use Build to retranslate our program. Then test it a few times using sample data, to check that it computes correct results.

Maintenance

Recall that program maintenance is the final stage of program development, in which modifications and improvements are added during its lifetime. Some studies have shown that the cost to write a program is only 20% of its total cost -- the rest is consumed by maintenace! One of the goals of object-oriented orogramming is to reduce this maintenance cost, by writing code that is reusable.

To simulate program maintenance, we can improve payroll by modifying it so that it displays pay in monetary format (i.e., with 2 decimal digits instead of the default number). The number of decimal digits in a real number is called the precision of that number. To show only two decimal digits, we must alter the default precision. As we have seen before, this can be done with an I/O manipulator called setprecision(), whose pattern is:

   setprecision( IntegerExpression )

• fixed, which ensures that the number is displayed in fixed point notation, rather than floating-point (scientific) notation.
• showpoint, which causes the decimal point and trailing zeros to be displayed when the decimal portion is zero.

If is included and these manipulators are placed in an output statement:

   cout << Val1
        << setprecision(2) 
        << fixed << showpoint
        << Val2 <<  ... << ValN;

Modify your source program so that pay is displayed with precision 2, and using the fixed and showpoint manipulators. Don't forget to #include the iomanip library! Then test the correctness of your modification.

Part II: Using the Debugger

While the compiler will inform us of any syntax errors our program contains, it cannot detect logic errors. The simplest way to check for logic errors is to examine the values of the variables in our program as the program executes, and verify that they are what they are supposed to be. Visual C++ provides a simple way to check variable values, using a special tool called a debugger. A debugger allows you to execute a program statement-by-statement, and examine the values of variables (or expressions) as the program runs.

Starting the Debugger

To use the debugger, choose

   Build -> Debug -> Step Into

Use F11 again to step into the program's main function. Then take a moment to reposition your windows so that you can see both the Developers Studio window and the payroll.exe window.

The yellow arrow should be pointing at the first statement in payroll.cpp, which is an output statement. Rather than stepping into the call to the insertion function (<<), we should instead step over the call using

  Debug -> Step Over

Using F10, step through the remainder of the program, one line at a time. If necessary, you can always use Debug -> Restart to begin again at the program's beginning.

Note that when you reach the input statement, execution will not proceed further until you switch to the payroll.exe window and enter the required data. Enter the data and continue stepping. Note in particular how the program behaves when execution reaches the for loop.

Examining Object Values

A nice feature of this debugger is that when you reach a declaration statement (e.g., numberOfEmployees), an entry for each object in that statement automatically appears in the auto subwindow at the bottom of the Developer's Studio window, and persists there until execution reaches a line of code where that object is no longer accessible. This allows you to examine the values of your program's variables and see how the execution of a statement affects them.

Restarting

As was mentioned earlier, you can always restart the debugger at the beginning of the program by choosing Debug -> Restart.

However, if you continue stepping until you reach the end of the program, one of the anomolies of the Visual C++ debugger (or is it a feature?) is that you can keep stepping beyond the end of the program and Visual C++ will wrap back around and restart the program!

Using either of these methods, restart the debugger at the program's beginning.

Setting and Clearing Breakpoints

Sometimes stepping one line at a time can get tedious. One way to skip past a block of code is to position the cursor at the end of the block of code (i.e., where you want execution to stop) and then choose

   Debug -> Run to Cursor

If you need to stop at the same place more than once (e.g., you want to stop at the same line inside a loop every repetition), an alternative is to set a breakpoint -- a place such that execution will stop whenever it reaches that line.

To set a breakpoint, position the cursor inside your for loop in payroll.cpp and then click the Insert/Remove Breakpoint button (this button is at the right end of the button bar and has a white upraised hand icon signifying stop) or use the F9 keyboard shortcut. You should see a red "stop sign" appear in the left margin of that line within the loop. If you then choose Debug -> Go or use the F5 shortcut, the program will run until it encounters the breakpoint (or until the program terminates).

You can remove a breakpoint by placing the cursor on the same line as the breakpoint and clicking the Breakpoint button again.

Set a breakpoint at the line containing the call to ComputePay() and then continue in the exercise.

Stepping Into and Out Of a Function

While there are situations where we want to step over a function (such as a system function like << or >>), there are other situations where we want to step into the function. For example, to examine the behavior of our ComputePay() function, we might want to step into it, rather than over it.

To illustrate, run your program using Debug -> Go and enter some non-overtime value for hours and a value for wage. The program should then continue to the breakpoint you just set. If you now use F11 to step into ComputePay(), the yellow arrow should "jump" from the call to the first line of the function. Note that the auto window changes, showing the function's parameters and their values. The objects from the main function disappear since they cannot be accessed within ComputePay().

You can then step through the function using F11 or F10. As you step past declaration statements (e.g., totalPay), the objects being defined should appear in the auto window. Note how the program behaves when execution reaches the if statement: the code that computes "normal" pay is performed, while the code that computes overtime pay is skipped.

Use Debug -> Go to continue execution of the program, and when it pauses for input again, this time enter an overtime value for hours (and a value for wage). When the program stops at the breakpoint again, step into ComputePay() and note that this time control skips the 'normal' part of the if but moves through its overtime part.

If at any time you want to leave a function (this is especially useful if you mistakenly step into a system function), choose Debug -> Step Out (or using its Shift+F11 shortcut) which will complete the execution of the function currently executing and stop back in its caller.

Quitting the Debugger

Whenever you want, you can quit using the debugger and return to the normal Visual C++ environment by choosing Debug -> Stop Debugging or its Shift+F5 shortcut.

Summary

The key things to remember in using the debugger are these:

• Use Build -> Debug -> Step Into (or F11) to begin use of the debugger. The Build menu will be replaced by a Debug menu.
• Use F10 (step-over) for 'normal' stepping through statements.
• Use F11 (step-into) to jump into a function call in order to examine the internal behavior of that function.

• Within a function, use Debug -> Step Out (or Shift+F11) to have the function run to completion.
• To have the program run to an arbitrary line, position the cursor there and use Debug -> Run to Cursor (or Ctrl+F10).
• Within a loop, use the "white hand" button (or F9) to set or remove a breakpoint.
• To run the program to the next breakpoint, use Debug -> Go (or F5).
• Use Debug -> Stop Debugging (Shift+F5) to quit the debugger.

From this exercise, you should now know the basics of using the debugger, as well as have a better understanding of how if statements and for loops work.

Phrases you should now understand:

Submit:

Back to This Lab's Home Page

Back to the Prelab Questions

Forward to the Homework Projects