On Thu, 2005-03-31 at 02:50 +0000, Firebird-Architect@yahoogroups.com
wrote:

> David Johnson wrote:
>
> >>>AFAICR, DEC invented UUIDs. Microsoft came up with
> >>>the name GUID I believe...
> >>>
> Nope. You made that up from scratch.
>
> UUIDs came from Apollo. They were picked up by the delusional prima
> donnas at OSF, who didn't do diddle squat. Microsoft knew a good
> thing
> when they saw it and claimed it as their own. DEC had it's last
> original idea in the early 1980s. The hardcore DECcies are hanging
> out
> in Hawaii waiting for TCP/IP to die so they can take over the world
> with
> DECnet ISO transport binding.
>
>

I was under the impression it was an IETF committee, but thanks for the
history. The snippet about DEC was an earlier contributor to the
thread, and I didn't register that I had included that bit.

> Paul Leach of Microsoft is the current "owner" of the
> >specification document.
> >
> Paul Leach of MICROSOFT!!!!!???? Tell me it ain't so!!!! Surely you
> mean Paul Leach of Apollo Computer.

from http://www.ietf.org/internet-drafts/draft-mealling-uuid-urn-05.txt

Network Working Group P. Leach
Internet-Draft Microsoft
Expires: June 1, 2005 M. Mealling
VeriSign, Inc.
R. Salz
DataPower Technology, Inc.
December 2004


A UUID URN Namespace
draft-mealling-uuid-urn-05.txt

<... snip ...>

I don't know where Mr. Leach was when the standard was built, but his memos since at least 1997 to present are all Micro$oft. Note that even under the M$oft name, the UUID specification is still called UUID. GUID is merely their implementation of the specification.

>
> UUIDs are necessary for systems lacking central management. With
> central management, you can do much, much better.

<snip .. out of order but related>

>
> Who care's about a machine not on a network?
>
>

I am building for systems with intermittent and/or unreliable communication that must be merged as external conditions allow, so by your standards a UUID is a requirement for my purposes. An example is a heavy construction management package for use in field and home offices. Since the cell tower goes up about 1 week after the construction crew has left the site, communications is an issue (manageable and intermittent, but still an issue). Since we are talking about tracing $1M to $20M per job over 1 to 12 months, the "who cares" is pretty **** important to someone. Not all applications run in a clean-room environment. Check out the communication infrastructure 40 miles north west of Zama Lake, Alberta. :o)

Although it would be nice, this does not "require" native UUID support in the database. For my purposes, a 36 character UUID representation performs nicely as the primary key on inexpensive WIntel hardware. (I don't keep the {}'s that M$oft adds to the string).

>
> In theory, MAC address are unique. Practice, however, trumps theory.
>

Agreed. It is a foundation of debate that reality reputes all logic. This is a significant flaw in the theoretical basis of dependency on UUID's.

>
> >Support for the data type, which is becoming more frequently used over
> >time, is not the same as endorsement.
> >
> Let's not confuse a problem with a perspective work around.

Agreed.