Hi Cees.

You wrote: "UUID's shouldn't be big random numbers, they should be
*provable* universally unique. At the moment, the only *provable*
universally unique scheme that is standardized is to have the machine's IP
address, the process id, the current time and a sequence number in it."

You make good sense in your analysis of UUIDs, but I'd like to point out
that IP addresses are not necessarily unique either.  In my company, for
example, the I.S. department is always messing around with those things
(causing endless problems for me).  On many occasions I have discovered that
there are 2 (or more!) machines on our network that have the same IP
address!

So, while I agree that the IP address is a good starting point, I think the
current date/time as well as the process ID would probably also be needed
for uniqueness.

I'm not an expert in such things, but I have been bitten by this problem a
few times in my career.  ;-)

Cheers,

---==> Chris