Paul Eggert wrote:
> > Assigning a 'char *' from a 'void *' still generates a diagnostic.
> 
> Is this because we might be using ancient C compilers, or something like 
> that? Because of course it's OK to assign char * to void * in C.

It's a long time since I wrote this code with mmap. Probably
at that time portability to some system was necessary that defined
caddr_t to 'long' or something like that. I can't tell for sure
whether the problems that led me to adding these casts are still
present. Not worth spending time on it.

> There are other places in Gnulib where we assign char * to void * 
> without a cast. If there's a reason we need the cast with mmap, I'd like 
> to add a comment

No comment needed, w.r.t. modern standards.

Bruno




Reply via email to