On Thu, Jun 27, 2002 at 10:07:50PM +0200, Jaromir Dolecek wrote:
> Would using
> 
> 	switch (retval) {
> 	case ERROR:
> 	    {
> 		char buf[256], *errbuf = &buf;
> 
> 		/* do error handling */
> 
> 		break;
> 	    }
> 
> suffice?

Well, it's still on stack. I'm not sure that elimination of stack is
a justifiable goal, stack memory is much faster then malloc'd one and 
it less error-prone(no need to free, no memory leaks). If I'd worried
about stack size I'd use unmapped red-zone as vax/sparc ports do first.

	Andrey