PATCH: IA64 floating-point types

[Joseph S. Myers]

On Mon, 8 Sep 2003, Mark Mitchell wrote:

> These are additional IA64-specific builtin types.  The psABI specifies
> __float80 and __float128.  Both are 128-bit types when stored in
> memory, but __float80 is IEEE Double-Extended and __float128 is IEEE
> Quad.  On HP-UX, where long double is IEEE Quad, __float128 and long
> double are the same type.
> 
> The __fpreg type is a special floating-point type; it is also IEEE
> Double-Extended, but using __fpreg means that when an 80-bit

Where are these types specified from a language extension point of view?  
(For example, when any of the types isn't the same type as a standard C 
type, what conversions apply?  I take it that the specification doesn't 
allow for _Complex to be used together with these names (since it won't 
work with built-in typedefs)?)

-- 
Joseph S. Myers
jsm@polyomino.org.uk