Function pointers

[Christoph Bartoschek]

Am Sonntag, 6. September 2009 schrieb David Nolden:
> 
> I guess the better way would be turning that assertion back into a warning,
> and turning the other assertion a warning as well instead. C++ is very
> complex, and we never know what kind of strange code we will still be
>  facing, so it's always better not to crash..

If the abstract type stored in a FunctionDeclaration is allowed to be ever 
different from FunctionType, this would be better. But then it is not 
necessary to issue a warning.

However if you feel that the code should never exit because of an assertion I 
could change the patch.

Christoph