Name mangling

[Martin von Loewis]

> >   void
> 
> I thought void was not strictly a type. Well, here, I guess we may  
> consider it a type.

It appears as part of a composite type, e.g. "pointer to void" or
"pointer to function returning void", so it is certainly relevant for
mangling.

> >   function expecting type*, returning type
> >   pointer to member function of type, expecting type*, returning type 
> >   pointer to member of type T, having type U (i.e. "U T::*")
> 
> As Daveed pointed out, the throw spec should be part of the function  
> types in C++.

It would only appear on functions (and methods), not on function
types, right?

Also, it is not strictly necessary, is it (being in Daveed's
nice-to-have list)?

> Regarding pointers, for anytbody supportingf ILP32 (is that just  
> us?), we may have a need for some "long pointer". This could be a  
> __far modifier. In any case, in ILP32, there is to be a 'long long'  
> type (64 bit) in the ABI.

That seems like a useful feature. Is it reasonable to consider __far
in the same category as the cv-qualifiers?

Regards,
Martin