🚀 go-pugleaf

In article <4k17gf$gob@isoit109.bbn.hp.com>, Bruno RICHARD <bruno_richard@hp-france-om1.om.hp.com> writes:

-=)> What is the recommended declaration for a function that does
-=)> not take arguments and does not return anything:
-=)>    void myfunc();
-=)> or
-=)>    void myfunc(void);

The second one is definitely better, as the first one as different
meanings according to the C style you're conforming to :

- In K&R, () means "gimme anything as parameters, I won't care"
- In ANSI C, it should be avoided
- In ANSI C++, it means "this functions takes NOTHING"

As K&R is a bit obsolete and the (void) notation works in ANSI C as
well as C++, you'd better use this form that is BTW more explicit than
nothing and should be interpreted the same way by any modern C
compiler.

-=)> I remember that few years ago on Turbo C 2.0 the second form was better
-=)> optimized. So I got used to do so, but I just read the C FAQ and noticed
-=)> that there was some argument about main() returning int or void,

main returns an INT and takes zero or two arguments. Nothing else.
But well, Turbo C is pretty lame to produce a less efficient code in the
second case anyway. Use DJGPP, it rocks ! Your code will be at least 3 times
faster than with Turbo C ( Turbo, eh eh... ) and GCC is a bit more ISO 9899
conformant than Borland shits.

-=)> and
-=)> I am wondering whether there could be some portability problems around.
-=)> -bruno

Best regards, Frank.
-- 
.-=)>       Frank DENIS aka JEDI from SECTOR ONE ( j@nether.net )        <(=-. 
\   Serveur minitel RTC-One au prix d'une simple communication sur Paris :   /
 )  (1) 48.70.10.29 ou (1) 48.58.46.17 - Messagerie, HP48, PC, delires, ... (
(   Acces 28800 bauds au (1) 49.88.76.91 . Informations: email j@nether.net  )
 \_-^-_.                   Mon pere est un vieux con                  ._-^-_/