Re: Wide_[Wide_]Character

[Georg Bauhaus <rm.dash-bauhaus@futureapps.de>]

Dale Stanbrough schrieb:

> 
>> Isn't this just like the RM not specifying the bit layout of
>> numeric objects?
> 
> I'm not sure what the point of Wide_Wide_Character is if not to deal 
> with Unicode (or ISO-10646:2003).

Sure, Wide_Wide_Character deals with ISO-1646:2003, the normative
reference is listed in the LRM; you get I/O of those characters,
and compilers will document the external encodings you can use.

I also got to know how to pass Wide_Wide_Character objects into
and out of my program in case I must (that's the Interfaces[.C] part).
But why and when should I wonder what the internal bit layout of
Wide_Wide_Character objects actually is?


> You could invent your own 32 bit Character code (or use the one the 
> vendor gives you), but playing in your own backyard doesn't seem very 
> productive.

Why not? If it is faster to use 64 bit words for Wide_Wide_Character
operations, if this does not waste too much first level cache,
then it seems like a good idea for a compiler to use 64 bits for
Wide_Wide_Character.



> To me the only point is if it implements the code.

Why?

--
Georg Bauhaus
Y A Time Drain  http://www.9toX.de