Skip to main content

Good Style of Programming

Here are few tips to program your C / C++ code in good style :


Using break is ok in loops. Ideally, the loop should be structured to iterate in the most straightforward way. The break in the body can detect the exceptional case that comes up during the iteration.

for (i = 0; i < length; i++ )
    if (<found>) break;


while (current != NULL)
    if (<found>) break;


while (TRUE) types loops are fine if they are really necessary. Often you need this for the loop-and-a-half-type situation where you need to first do some processing before you are able to test whether you need to exit the loop. If the first statement of a while (TRUE) loop is the test, then you should just put the test directly into the while statement. If the bounds of iteration are known, then a for loop is preferable.


A return in a place other than the end of a function body is potentially vulgar. The early return can be used nicely if it detects and immediately exits on an exceptional case-for example, a recursive base case or an error case right at the beginning of a function. Sometimes return can be used like a break inside a loop when some condition becomes true. Be careful with return in the bodies of your functions- experience shows they are responsible for a disproportionate number of bugs. The programmer forgets about the early-return case and assumes the function runs all the way to its end.

Increment Operators ++, –

Nice obvious uses of these are fine, but nesting it inside something complicated is just asking for trouble. Find a more useful outlet for your cleverness.

for (i = 0; i < length; i++) … // Ok
while (–length) … // Ok
while (*t++ = *s++) … // Sick


If you ever exploit the fall-through property of cases within a switch, your documentation should definitely say so. It’s pretty unusual. !=, ==

This is just a minor readability issue, but it’s nice to put in what you are really testing for, rather than rely on the anything-non-zero-is-true property. Even though the code may compile to exactly the same thing, it reads a little nicer.

if (*current) more clearly is if (*current != ”)
if (current) more clearly is if (current != NULL)
if (!count) more clearly is if (count == 0)

Boolean Values

Boolean expressions and variables seem to be prone to redundancy and awkwardness. Replace repetitive constructions with the more concise and direct alternatives. If you have a boolean value, don’t use additional != or == operators on it to test its value. Just use its value directly or inverted. Also, watch for if-else statements that assign a variable to true or false. The result from evaluating the test can go right into the variable. A few examples :

if (flag == TRUE) more succinctly is if (flag)
if (matches > 0) more succinctly is found = (matches > 0);
found = TRUE;
found = FALSE;
if (hadError == FALSE) more succinctly is return !hadError;
return TRUE;
return FALSE;


#define-d constants should be independent; that is, you should only need to change one #define to change something about a program. For example :

#define RectWidth 3
#define RectHeight 2 /* WARNING: problem */
#define RectPerimeter 10

is not so hot, because if you wanted to change RectWidth or RectHeight, you would also have to remember to change RectPerimeter. A better way is :

#define RectWidth 3
#define RectHeight 2
#define RectPerimeter (2*RectWidth + 2*RectHeight)


Popular posts from this blog

JUnit – Run unit test in an Sequence / Order

In JUnit, we can use @FixMethodOrder( to run the test methods in a sequence or order.

import org.junit.FixMethodOrder;import org.junit.Test;import org.junit.runners.MethodSorters;importstatic;importstatic org.junit.Assert.assertThat;//Sorts by method name@FixMethodOrder(MethodSorters.NAME_ASCENDING)publicclassExecutionOrderTest{@TestpublicvoidtestB(){assertThat(1+1,is(2));}@Testpublicvoidtest1(){assertThat(1+1,is(2));}@TestpublicvoidtestA

Create Runnable Jar - Eclipse Options

When exporting to a Runnable Jar, there are three options in eclipse Helios. Extract required libraries into JARPackage required libraries into JARCopy required libraries into sub folder next to JAR. What are differences : Extract required libraries into JAR - Extracts the actual .class files from the libraries your app uses and puts those .class files inside the runnable JAR. So, the runnable JAR will not only contain the .class files of your application, but also the .class files of all the libraries your application uses. Package required libraries into JAR - Puts the actual JAR files of the libraries into your runnable JAR. Normally, a JAR file within a JAR file cannot be loaded by the JVM. But Eclipse adds special classes to the runnable JAR to make this possible. Copy required libraries into sub folder next to JAR - Keeps the library JARs completely separate from the runnable JAR, so the runnable JAR will only contain the .class files of your application. Option #2 is convenient be…