Core Java: Programming Error and Debugging

There are generally three classes of programming errors:

 

·        Compilation Error (or Syntax Error): can be fixed easily.

 

·        Runtime Error: program halts pre-maturely without producing the results - can also be fixed easily.

 

·        Logical Error: program completes but produces incorrect results. It is easy to detect if the program always produces wrong result. It is extremely hard to fix if the program produces the correct result most of the times, but incorrect result sometimes. For example, 

 

// Can compile and execute, but give wrong result – sometimes!

if (mark > 50) {

   System.out.println("PASS");

} else {

   System.out.println("FAIL");

}

 

This kind of errors is very serious if it is not caught before production. Writing good programs helps in minimizing and detecting these errors. A good testing strategy is needed to ascertain the correctness of the program. Software testing is an advanced topics which is beyond our current scope.

  

Debugging Programs

 

Here are the common debugging techniques:

 

·        Stare at the screen! Unfortunately, errors usually won't pop-up even if you stare at it extremely hard.

 

·        Study the error messages! Do not close the console when error occurs and pretending that everything is fine. This helps most of the times.

 

·        Insert print statements at appropriate locations to display the intermediate results. It works for simple toy program, but it is neither effective nor efficient for complex program.

 

·        Use a graphic debugger. This is the most effective means. Trace program execution step-by-step and watch the value of variables and outputs.

 

·        Advanced tools such as profiler (needed for checking memory leak and method usage).

 

·        Proper program testing to wipe out the logical errors.

Testing Your Program for Correctness

 

How to ensure that your program always produces correct result, 100% of the times? It is impossible to try out all the possible outcomes, even for a simple program. Program testing usually involves a set of representative test cases, which are designed to catch the major classes of errors. Program testing is beyond the scope of this writing.