Void main(void) - the Wrong Thing | Hacker News Reader

Void main(void) - the Wrong Thing

users.aber.ac.uk

42 pointsgnosis21 comments

I remember in high school my C teacher made us write it that way, even after most of the kids in the class tried to convince him against it. In his defense, he didn't know C (he was taking night classes at a local college and then teaching us what he had been taught the night before).

Your learnt C in highschool? Thats pretty great. In highschool we had a very talented and enthusiastic teacher who rewrite the vb6 examples in quite a bit of the terms curriculum in Python and introduced us to that first.

I actually predominantly learned COBOL and RPG in High School (which makes me feel old, even though this was in 97). I thought COBOL was useless at the time, but amusingly have had to do two separate source code audits of COBOL code in the past year (where I was the only person in my company who could do it).

My senior year, they decided to pilot a C programming class. None of the teachers knew C, but one Math teacher volunteered to take night classes and teach us what he was learning.

I would say in about two months, every student in the class was significantly more knowledgeable than the teacher; but he was very cool about it.

The only bad part was he spent about four months on graphics programming; which wouldn't have been bad, except we learned Borland C Graphics (getpixel, putpixel, etc.) which were pretty much useless.

The year culminated in us sending a team to a C Programming Contest being held at a nearby university. I always thought it was amusing that the grand prize (which we didn't win) was a copy of Visual Cafe (the Java IDE).

For the last recruit tests I gave, it was the easiest way to discard applicants without proper C knowledge. A wrong answer predicted a bad score on pratical (and lenghty) tests like code commenting, finding bugs, etc.

Anecdote: When corrected, a guy who allegedly coded in C for 10 years even asked what was the use for main to return something :-(( So we talked about unix return values, $? in bash and he had a spark in his eye. I had a tear in mine.

seriously? i knew that shit when i was 15. taught myself C on an amiga 500.

Besides, int main(void) is shorter to type. And, horridly, return isn't necessary in main, so you don't even have to be bothered to type that.

int main(void)--for the lazy AND correct!

> And, horridly, return isn't necessary in main,

Only true for C++.

    but it works for me!

Works On My Machine(tm) http://www.codinghorror.com/blog/2007/03/the-works-on-my-mac...

Also true in C99 5.1.2.2.3: "reaching the } that terminates the main function returns a value of 0."

For the even lazier:

    main(void)

Though I vaguely recall something about this being deprecated in C99.

Yes, that's deprecated. And it was horrible before.

Wouldn't int main() be even better. Shorter, and acknowledges that main actually takes parameters (command line arguments).

GCC (as an example) turns () and ( void ) into the exact same output. I have some memory of long, long ago entering the world of C in the hands of the DICE compiler for the Amiga, that being clear in the declaration of functions is The Right Way: say void, or say inputs. Semantically I consider ( void ) to simply say "in this case, we're not going to be using any input vectors", rather than "this function can never have inputs; it's not supported". Clarity is never wrong.

main is a special case, but generally () and (void) are different, void meaning that the function cannot have any parameters and empty allowing any number of arguments. Therefore imho is clearer to use main() rather than main(void), indicating that there may be parameters, but we don't care.

http://codepad.org/UGLG7yRm and http://codepad.org/w2GyTkB1

I think it's worth mentioning that in the Plan 9 dialect of C, main returning void is absolutely correct. Instead of returning an integer, you call exits() which uses a string argument. The fact that Ken Thompson and Rob Pike made this design choice for Plan 9 is interesting. This is just a related fact, not an attempt to argue that void main is acceptable in standard ansi/iso C.

Also see:

http://jk-technology.com/ctips01.html#int_main

and

http://www.eskimo.com/~scs/readings/voidmain.960823.html

Just to be a devil's advocate, void main(void) can be correct when programming microcontrollers & other embedded applications.

With GCC, you can even set `void main(void) __attribute__((noreturn));` .

(For the lack of a better source, http://www.nongnu.org/avr-libc/user-manual/porting.html )

I see void being passed as a parameter list to main in a lot of examples. All of the things I've read suggest that the only difference between having a void parameter list and an empty parameter list is in function declarations, and that the two are equivalent in function definitions. Why is it used so frequently in the main function definition, if it serves no purpose?

You've probably been reading C++ material. Only in C++

  int f();

and

  int f(void);

are identical.

In C, however,

  int f();

Declares a function taking any number of arguments. In C++, that would be denoted as

  int f(...);

Which is unfortunately a syntax error in traditional C as the ellipsis is only allowed following some explicit parameters. (It's not that big a deal; you'd have to

  #ifdef __cplusplus
  extern "C"
  #endif

in the header of the C library for C++ compatibility anyway (to prevent function name mangling and thus link errors), so you can #ifdef the () and (...) too.)

FWIW, C++ does support the legacy

  int f(void);

syntax, but its use is discouraged except for headers which are also to be used for C code.

I don't understand what the discussion is, the standard says the main function should have a return type of int, that should be the end of it, if some random beginner argues "but it works for me", there's not really much you can do, people will continue to write incorrect programs if they want to.

Reply on news.ycombinator.com