From: Johan Holmberg <holmberg@...>
Date: 2004-10-14T23:50:06+09:00
Subject: Re: Kernighan & Richie ---> prototypes ?


On Thu, 14 Oct 2004, Yukihiro Matsumoto wrote:
>
> |  What is holding back a conversion of the K&R code
> |  to use prototypes ?
>
> My preference.  I like prototypes.  But I dislike the new style
> argument list, especially when the list becomes large with type
> names.
>
> 							matz.
>

Oops. I had kind of hoped for another answer.

If the current source code was functionally equivalent to using
prototypes everywhere I wouldn't have minded. But if I'm not
mistaken you actually loose a lot of type-checking now.

I just ran PC-LINT on the Ruby source and got many messages like:

   ... call to function 'SOME_FUNCTION_NAME()' not made \
       in the presence of a prototype

All in all for 511 different functions.

I thought that the K&R style of function declaration/definition was
long ago obsolete. Looking in by bookshelf I see books such as
"Writing Solid Code", "The Practice of Programming" and "The Art of
Unix Programming" all mentioning the advantages of the ANSI style.

But I realize that it is your source ;).

I just had to make my argument.
This has been one of the main things I have wondered about while
reading the source (the other thing being the almost total lack of
comments about what functions do).

/Johan Holmberg