Why C Language Has the Main Function

By Xah Lee. Date: 2023-11-07. Last updated: 2025-10-04.

2025-10-04

Why Do Some Programing Language Require a Function Named Main? is it Necessary?

Why do most compiled programing language have “main” function? e.g. C Cpp Java. Is it necessary?

The “main”, is man-made complexity.
The concept does not exist in computer science, nor in math. (similar to closure.)
It is not necessary.

here's a mathematician's view of what does programing means:

We write a program, in a programing language, which is a language to specify algorithms, which is a sequence of step-by-step instructions, and this, is run on a specific machine.
(think of an abacus as machine, and a language that says column n move bead 3 up/down as the programing language. A list of such instructions as the source code.)

In the source code, the code runs from top to bottom.
Such as defining functions, and calling the function.
This top to bottom is always necessary, unless the programing language is designed in a special way to have other order (such as JavaScript function name-hoisting)
Theoretically you never need to specify a “main” about where to “begin”.
Again, unless the programing language for some reason is designed that way, and does something extra to make that work.

The reason many programing languages such C Cpp Java have main, is due to historical reasons, specifically on how electronic computer happened to be designed and evolved, and the way we do things.
C began it, and Cpp Java etc followed it by force of habit.

Basically, code is compiled to several object files (for human management reason).
The object files, each is a binary number.
In the object files, parts are function definitions, parts are data, parts are instructions.
We developed ways to distinguish them by various means (by binary marker, file name, or convention, etc.)
And a so-called linker assembles them into so-called executable.
All of the above, relates to how the physical machine is actually designed and how we created convention to work on them.

so, in this convention, for technical reasons, the linker needs to know, where is the actualy instruction part that the machine should read to run the program.
and by convention, that is the “main”, as a symbol in binary, searched by another program of so-called loader.

2025-10-04

grok ai explains linker, main. and in compiled languages sans main, how do they do linking? e.g. Julia, Ocaml, Clojure, Swift.
https://x.com/i/grok/share/Eio3LlbW5NdMubClCvXweFWsM

2023-11-07

Here's a sketch of history of where did C get its main.

ok. main, came from C (1972). and its daddy B (1969).
and the concept may came from BCPL (1967), which uses start().
the main() do not exist before that. e.g. not in fortran (1957), lisp (1960). A decade before the C virus.

u see, they talk about object files, and refer to each other as symbols, and addresses.

so, to run a program, the program in the end is bunch of machine code in disparate parts in the hardware.
this is how things are , actually, in the era of electronic computers, or so i think.
now, this is opposed to, modern times, or earlier times. for example, in modern times, we think of a python script, or ruby, bash, js, as something like a natural language, we read it from top to bottom. The meaning of the code is accumlated sequentially in order, like english.
from a high level of view, a language, not for human but for computers, is like this. similar to human langs.
and this is indeed so, for most high level langs, or scripting langs eg python ruby bash pwsh, or functional langs eg lisp wolfram, or math langs eg apl matlab.
or, in more primitive times, eg punch cards, or hand written instructions on how to solve math problems, algorithems r written in a script, to be executed from top to bottom in order.
but in the electronic computer era, eg the times of c, it created this... aberration, due to the concept of 'compilation'.

in this scheme, u do not start by writing a program of step by step instructions of an algorithm. instead, u need to setup the machine hardware, so each component needs to contain certain code. and in order to do that easily, u created the concept of c, that 'compiles', into these diff 'objects', to be placed in diff places in the hardware.
so, the source code of c, is not really, meant to be sequential spec of something. rather more like, a bunch of modules. as georrg said.