NM>
  > What's wrong with:
  >
  >     int
  >     main(void)
  >     {
  >         printf("Unsigned Ints have %u bits.\n",
  >                sizeof(unsigned int) * CHAR_BIT);
  >
  >         return 0;
  >     }
NM>

  Strictly speaking, it's not portable.  Although it is generally agreed
  that in ISO Standard C there are no "hidden bits" in an unsigned char
  (in order to satisfy some of the semantics of memcpy()), it is *not*
  guaranteed that other integral types will not have hidden bits.

  In other words, the number of available bits in any integral type may
  be *less* than the number of `char's wide that the type is (i.e. sizeof)
  multiplied by the number of bits in a `char'.

  Of course, this doesn't apply to Intel platforms, and I don't believe
  that it applies to PowerPC either.  However, it is allowed by the ISO
  Standard because it *may* apply to other platforms (imagine a platform
  where all integers have a hardware parity bit, for example).

  But as far as Intel OS/2 and PowerPC OS/2 are concerned anyway (let's
  get back on topic here), `unsigned int' has 32 bits.

  > JdeBP <
___
 X MegaMail 2.10 #0:
--- Maximus/2 3.00