Jonathan de Boyne Pollard wrote in a message to Mike Bilow:

  > I never disputed this.  Perhaps you have me confused with someone I
  > quoted?

 JdBP> No.  It was definitely you telling me, at great length,
 JdBP> things that I already knew.  (-:

Perhaps there was a misunderstanding, then.

  >   > 3. An int should be chosen to be most quickly manipulated on a
  >   >    target machine.
  >
  >  JdeBP> This is a common efficiency myth.  The ISO C standard in
  >  JdeBP> fact says no such thing.
  >
  > I'm not at all familiar with ISO C.  Both K&R and ANSI do recommend that
  > int should be whatever is convenient.  It may not be a
  > requirement, but what other possible basis could there be
  > for choosing sizeof(int)?  K&R goes further, saying that
  > the whole point of int as a type is to be quickly
  > manipulated.

 JdBP>   K&R says that int should be the type that is quickly
 JdBP> manipulated. ANSI C says that int should be whatever is
 JdBP> convenient.

That's a pretty fine difference!  As a practical matter, I think that
K&R and ANSI are saying essentially the same thing.  I guess this is a
matter of opinion, however.

 JdBP> ISO C merely says that the size of int is
 JdBP> implementation defined and gives a lower bound on the
 JdBP> number of bits that it contains.

 JdBP>   There's a trend here.  Do you see it ?  (-:

It is reasonable for the standards to leave increasing amounts of
flexibility in the hands of the implementors.  Probably a lot of this is a
result of MS-DOS, where 16-bit programs are running on 32-bit CPUs.
 
-- Mike


---