On Wed 20 Jul 2005 at 14:55:34 -0700, Jonathan Stone wrote:
> Of course, a CRC-32 isn't going to catch errors where xor'ing the
> original message and the damaged message gives you a polynomial
> that's  divisble by the generator polynomial. But how likely is that for
> "Naturally occurring failures", as opposed to enemy action?

Note that if, for instance, a single 32-bit word is set to all zeros,
that is NOT (in general) a single 32-bit error. Since some of the
original bits (in general) were zero to begin with, you have multiple
shorter errors.  Hence guarantees like "catches all single error bursts
of up to 32 bits" do not guarantee anything here since they do not
apply.

-Olaf.
-- 
___ Olaf 'Rhialto' Seibert      -- You author it, and I'll reader it.
\X/ rhialto/at/xs4all.nl        -- Cetero censeo "authored" delendum esse.