Re: glibc2.1 programming question

[Steve Greenland <stevegr@debian.org>]

On 24-Jun-99, 13:26 (CDT), Andreas Tille <tille@physik.uni-halle.de> wrote: 
> Hello,
> 
> under glinc2.0 worked the following:
> 
> typedef union {
>    char  *s;
>    int    i;
>    double d;
> } C_VALUE;
> 
> int func(C_VALUE val);
> 
> void main(void)
  ^^^^ 
<comp.std.c> That should be "int". "void main(<anything>)" is not valid
C.</comp.std.c>

> {
>   char *s;
> 
>   func(s);
> }

Yechh. The compiler is right -- that's why we have prototypes.

Replacing it with func((C_VALUE) s) is no better. Consider what happens
if func() decides to access member d. (Ok, you may know that func()
won't do that. But if it doesn't, why does it take a C_VALUE?)

As our fearless leader said, the code is evil.

Steve