How a COBOL Program Is Built and Run

If you are coming to COBOL from Python, JavaScript, or PHP, you are used to interpreted languages. In an interpreted language, a program called an interpreter reads your code line by line and executes it immediately. If there is a typo on line 100, the program will run perfectly for 99 lines before suddenly crashing.

COBOL is fundamentally different. It is a compiled language.

The Two-Step Process

A COBOL program cannot be run directly. It must first be processed by a compiler—a separate piece of software that translates your human-readable source code into binary machine code that the processor can understand natively.

1. Compilation: The compiler reads the entire program, checks the syntax, and generates a binary executable (often called a load module or object module). If there is even one syntax error, the compiler stops and generates no executable.
2. Execution: The operating system runs the generated executable file.

This means a COBOL program will never crash halfway through because of a simple typo—the compiler catches those upfront.

How We Run Code Here

Because COBOL requires a compiler, this course uses the open-source GnuCOBOL compiler behind the scenes.

When you see a program block, imagine that we are running the cobc (COBOL compiler) command to turn the text into an executable, and then immediately running that executable to produce the output.

Interpreted language developers often get tripped up by the compilation step. For example, in a script like Python, you cannot call a function before you define it—the interpreter will crash. But because a COBOL compiler reads the entire file before creating the executable, it already knows about every paragraph before the program even begins running!

       IDENTIFICATION DIVISION.
       PROGRAM-ID. RESOLVE-TEST.
       PROCEDURE DIVISION.
           PERFORM LATER-STEP.
           DISPLAY "First step complete.".
           STOP RUN.
           
       LATER-STEP.
           DISPLAY "Running the later step!".

Running the later step!
First step complete.

If you try to remove the PROGRAM-ID or make a typo, the system will show you a compilation error rather than execution output. This strictness is exactly what you want when dealing with enterprise financial systems!

Under the Hood of GnuCOBOL

When you compile with GnuCOBOL, it actually does something quite clever: it translates your COBOL source code into C code first, and then uses a standard C compiler (like GCC) to turn that into the final binary.

       IDENTIFICATION DIVISION.
       PROGRAM-ID. COMP-STEP.
       PROCEDURE DIVISION.
           DISPLAY "Step 1: cobc translates this to C.".
           DISPLAY "Step 2: A C compiler creates an executable.".
           DISPLAY "Step 3: The executable is run.".
           STOP RUN.

Step 1: cobc translates this to C.
Step 2: A C compiler creates an executable.
Step 3: The executable is run.

You can compile directly to an executable using the -x flag (cobc -x program.cob), or stop at the object file stage using -c (cobc -c program.cob). In large enterprise systems, you’ll often compile dozens of COBOL files into object modules, and then link them all together into one giant application.

But for this course, all you need to know is that behind every piece of output, a compiler successfully verified every single line of your code!