Repetition: unconditional jumps

Repetition occurs prominently in many programming applications; C provides a number of particularly attractive and elegant constructions. In Figure 2.1, we require the main body of the program to be repeated indefinitely. One possible construction which avoids using the infamous ‘goto’ is:

This is a for( ) loop which has no initial condition, no terminating condition and no action which is applied on each iteration of the loop. Since there is no terminating condition, it does not ever terminate; it is a forever loop.

The program in Listing 2.1 is designed to read a number from the keyboard, to output the binary value to leds attached to port A and to read/display the contents of port A. Repetition is illustrated by the flowchart Figure 2.1. Alongside this is a construction called Structured English. It is an alternative way of describing a problem to be committed to software. More is explained about it on page 262. It is a descriptive structure which does not require graphic symbols and yet has a very good correlation to the code which it describes. The indentation is used to help visually align code under control of the LOOP construction.

Connecting the data display circuit Figure 1.7 demonstrates the effect of the program. Alternatively this program construction successfully controls the speed of a small d.c. motor. A suitable circuit is shown in Figure 2.2.

Integers 0 to 15, input from the keyboard, are processed through the 4-bit D-to-A and power amplifier to provide 16 unique speeds. This simple uni-directional open-loop control system provides the basis for more advanced closed-loop control.

Named constants #define

Listing 2.2 generates a square wave of approximately 5 V p–p at a frequency of 500 kHz, by causing PA0 to go repeatedly high, then low. As a programming exercise it demonstrates how to replace names for constants, improving the readability of the program. Most C/C++ programmers use upper-case letters when naming a #define constant: this helps to distinguish variables from defined constants. Notice that the constants are defined outside the main program and not terminated with a semicolon.

Infamous goto

The primitive, repetitive nature of this program provides an excuse to demonstrate an unconditional jump; using the goto statement. Many of the erudite books written about C/C++ insist that it should be avoided, or used with extreme caution. This is good advice, since its use encourages unstructured programming and makes programs difficult to read. Not wishing to be contentious, we show how the structured infinite for-loop can be replaced by the unstructured goto, without loss of readability (in this example). When the goto statement is executed, control is transferred to the C/C++ code following the label (in this case start). The label is always terminated by a colon.

The goto is generally considered to be a no-no for serious programmers, because it can lead to ‘spaghetti code’ which is very difficult to follow and debug. Many professional software houses forbid its use in any high level language programs. Having said that, it is the main method of transferring program control in assembly language programs. These are only included in small doses within a C/C++ program, usually to speed up a critical part of its operation. There is never an application in C/C++ which cannot be programmed without a goto. Sometimes, its use does make the resultant code more compact, but if its use is allowed, then use it cautiously and sparingly!

For-loops in greater detail – in a binary counter

Educationally, Listing 2.3 is particularly rewarding, because it demonstrates a number of different loop constructions in a single program. Connecting the data indication circuit (Figure 1.7) to port A configured as an output provides a binary representation of the current status of the count. The denary status is displayed on the monitor using the printf and cout functions. Counting from 0 to 255 is achieved using the post increment operator (i++), a versatile construction meaning i = i+1 (this could also be written using the C/C++ convention i + = 1).

Inspection of the structure shows the program consists of a number of nested loops; the outer loop

ensures that the procedure is continued indefinitely. The actual count is controlled by the construction

The terse nature of the expression: i++ < = 255 deserves further explanation. It means test if ‘i’ is less than or equal to 255 and regardless of the result, increase i by one. Two useful features of C/C++ are the increment and decrement operators, meaning i = i + 1 and i = i − 1. The versatility of C allows you to test the value of ‘i’ and then add 1 to it – post increment operator i + +. Or add 1 and then test, the pre-increment operator + +i. A similar construction is provided for decrementing. To produce an observable display it was necessary to introduce a significant time delay into the loop. This is exploited to demonstrate the ‘for’ construction in greater detail.

The for-loop construction is particularly elegant in C, succinctly combining the three loop parameters (initialize, test and increment) in a single bracketed term. Expression 1 initializes the count. Expression 2 indicates the condition for the count to continue and expression 3 denotes the action which must be taken before the loop repeats. Referring to my example program, notice that k is incremented after the main body of the for-loop has been executed. Incidentally, the use of i, j and k as loop counters was introduced as early as the 1960s when the FORTRAN programming language reserved variables as integers which started with these letters.

Port monitoring with a do-while construction

In this example (Listing 2.4) the program monitors the switches on port A and displays the decimal value on the screen, before latching the binary value to leds on port B. Including the conditional do-while statement ensures the body of instructions are always executed at least once, before the test is made. Thereupon, looping continues until the result of the test is false. Observe how the test monitors the input port and compares the current contents (called new_contents in the program) with the old_contents. When this expression is false (the double equal signs = = mean ‘of the same value’ in a comparison) control transfers to the instruction following the while-statement.

The versatility of C provides an alternative while-do construction, which tests an expression and if true, permits looping to continue until the result of the test is false, at which point control passes to the next instruction in the program. These structured flowchart constructions are shown in Figure 2.5.

Be careful with the use of = and = =. The C/C++ expression

as shown in Listing 2.4 compares the values of variables new_contents and old_contents. If these are the same, then the statement(s) which follow it will be executed. Compare this with

This is a perfectly valid C/C++ expression. However, it allocates to the variable new_contents the value held in a variable called old_contents. Since new_contents and old_contents are now the same, the while statement then evaluates the contents of the bracketed term as zero. As was stated in Chapter 1, a zero is FALSE, so the statements which follow the while expression will never be executed.

Light chaser effect

This program Listing 2.5 is designed to produce the effect of a light running repeatedly across the leds connected to port A, which is conditioned as an output. It works by taking powers of 2 and writing the byte to the port. Don’t imagine this is frivolous; it does have a serious side, demonstrating how to write your own functions using C. Observe that the structure is composed of the main function that controls the flow of the program, together with an external function called power(x,n) that can be invoked and executed by a single statement in the main program. The new line int power(int x, int n); is called a function prototype. All functions in the program must be declared prior to the main( ) function. This is so that the compiler knows what functions to expect and what types of data will be passed in and out. A function declaration has the...