> 
> but how *else* do you do this correctly, for it to work with both
> k&r compilers and "strict ansi" ones ?
> 
> i know it's ugly, but, i know of no other way to get the benefits
> of prototypes while still allowing the code to compile with non-ansi
> compilers (which some code that i maintain requires).
> 

Is that NetBSD code (kernel/userland), or your own code?  I don't see
the problem if you have to maintain your own code, as gcc and most any
C compiler (well, unix C compilers), will eat K&R C code for breakfast.
However, the following should make it quite abundently *CLEAR* that we
need to use either lint, or ANSI-C.  Nuff said.


> To:      netbsd-gnats@wizardz.com
> From:    Daniel Hagerty <hag@wizardz.com>
> Subject: port-hp300/1918: mis-declaration of hilqfree in hil.c
> 
> >Number:         1918
> >Category:       port-hp300
> >Synopsis:       mis-declaration of hilqfree in hil.c
> 	
> >Description:
> 	hilqfree from arch/hp300/dev/hil.c is not declared
> properly. The argument 'p' doesn't have a declaration and is assummed
> int. This doesn't show up in kernel's that don't have DEBUG set.
> >How-To-Repeat:
> >Fix:
> 	This patch should fix it.
> 
> diff -c -r1.1.1.2 hil.c
> *** hil.c	1995/12/13 13:26:49	1.1.1.2
> --- hil.c	1996/01/09 08:27:04
> ***************
> *** 1012,1017 ****
> --- 1012,1018 ----
> 	hilqfree(hilp, qnum, p)
> 		register struct hil_softc *hilp;
> 		register int qnum;
> + 	struct proc *p;
>  {
>  
>  #ifdef DEBUG
> >Audit-Trail:
> >Unformatted:

--Toby.