> Matti Aarnio <matti.aarnio@sonera.fi> wrote:
> > Jakub Jelinek <jj@sunsite.mff.cuni.cz> wrote:
> > ...
> > > long long is defined in ISO C9X, your above definition is no longer true.
> > >
> > > C9X clearly states in 6.2.1.1:
> > >
> > > - The rank of long long int shall be greater than the
> > > rank of long int, which shall be greater than the rank
> > > of int, which shall be greater than the rank of short
> > > int, which shall be greater than the rank of signed
> > > char.
> >
> > Interesting, so this tells that:
> >
> > rank(long long int) > rank(long int) > rank(int) > rank(short int) > rank(char)
> >
> > While following was the previous definition:
> >
> > rank(long long int) >= rank(long int) >= rank(int) >= rank(short int) >= rank(char)
> > Systems do exist where *all* of those have same size, but they are
> > usually programmed in Fortran-9X, require liquid cooling, and recognize
> > name "Cray" ...
>
> I don't see a precise definition of "rank", but I do see that there is a bit
> of confusion about "rank" vs. "size". To quote from 6.3.1.1 (not 6.2.1.1):
A dictionnary should be enough. I understand that they must be considered
as strictly ordered even on target where some have same representation.
In other words:
- Never assume that 2 of them have the same representation if you want to
write portable C code.
- A nice compiler will warn you when your code seems to assume such thing,
even if the target platform has actually same representation for the 2
types.
> -- No two signed integer types shall have the same rank, even if they have
> the same representation.
> -- The rank of long long int shall be greater than the rank of long int,
> which shall be greater than the rank of int, which shall be greater
> than the rank of short int, which shall be greater than the rank of
> signed char.
Apply these sentences to army, and if it still isn't clear, then
learn French language. ;-)
> The 1st bullet indicates that rank is independant of representation and if
> you look at the definitions within <limits.h> (5.2.4.2.1) you'll see that
> signed char is at least 8 bits long
> short int is at least 16 bits long
> int is at least 16 bits long
> long is at least 32 bits long
> long long is at least 64 bits long
I donnot remember 'int' to ever have been 16 bits large on real
machines. Is this spec. that old? ;-)
> There is no prohibition against all of the integer types being the same length
> as long as the minimum requirements are meet.
Example on GCOS8:
char 9 bits
short 36 bits (optionnally 18 bits)
int 36 bits
long 36 bits
Seems both Bill and Bull are still compliant to C, btw. :-)
> John Cochran
Gérard Roudier.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/