der Mouse  (mouse@Rodents.Montreal.QC.CA) wrote:

>>> I don't know what's done by implementations where int and char are
>>> the same size.  seebs? :)
>> In the 8-bit int case,
>
>...you aren't using C, as seebs pointed out.
>
>I was actually thinking of an implementation I heard spoken of (I think
>it was for a DSP or some such CPU) that made char as wide as int, which
>was the size of the machine's registers, because addressing individual
>bytes was either impossible or prohibitively difficult.

The TI C3x DSP uses a 32-bit int as a char; the machine can only address 
32-bit words.  The Cray XMP/YMP is also a word addressable system, but 
the "C" implementation used a 8-bit quantity for a char and did the 
manipulation in software.  I'm assuming both are "correct" as far as the 
standard went at the time and that the size of a char was left to the 
implementor.

-bob