Re: problems with forward invocation (gcc bug?)

[Helge Hess]

Richard Frith-Macdonald wrote:
> > This might not possible.  We know that -cString returns a (char *)
> > but the offending line said:
> >
> > [x cString];
> >
> > And if the compiler only knows that x is of generic (id) type.
> > How can it know what the return type of cString is supposed to be?
> 
> Good point ... I guess we can't call it a compiler bug then ... more
> of a language bug.  Now if we had a type encoding for a return type
> of 'dont care' the compiler could generate some info.

No, this is not a language/compiler bug. The [x cString] code with x
being id will compile correctly with gcc. There are only two situations:

a) there is only a single definition of cString - in this case the
return type is obvious and taken by the compiler

b) there is either no definition or multiple definitions of the selector
- in this case the compiler throws a warning (maybe it should throw an
error) and chooses some (usually unusable), default behaviour. If there
is no def it assumes 'id' for all selector parts.

Greetings
  Helge
-- 
SKYRIX Software AG - http://www.skyrix.com
Web Application Technology for Enterprises