UUIDs and OIDs

[greggy]

So I went to "http://www.itu.int/ITU-T/studygroups/com17/oid/X.667-
E.pdf", because it was recommended to me (from another thread) and as
I understand it, 2.25.<UUID in decimal form> is a valid OID that can
be used as a prefix for UIDs.

Also, according to this document, I can form a valid UUID from a
completely random number and form: 2.25.<my entirely random UUID in
decimal form>.

Is my understanding correct? If so, then...

1) The document is making an assumption that a UUID is truly unique,
by stating that UUIDs that are partially or fully composed of random
values (sections, 12.5.3, 12.5.5, and 15) are fully legal UUIDs.

2) The OID 2.25.<uuid as a long integer> is assumed never to collide
with any other.

So it seems to me that one could just generate UIDs in the form
2.25.<uuid> each time a UID is needed. It would be perfectly legal,
since the UID would never collide with an OID (according to the
assumptions made by the document). Registration should not be
necessary, since no one could ever generate that same UUID value.

Or am I missing something?

[Robert Horn]

On Jun 30, 6:29 pm, greggy <gregofi...@yahoo.com> wrote:
> So it seems to me that one could just generate UIDs in the form
> 2.25.<uuid> each time a UID is needed. It would be perfectly legal,
> since the UID would never collide with an OID (according to the
> assumptions made by the document). Registration should not be
> necessary, since no one could ever generate that same UUID value.
>
> Or am I missing something?

The only restriction is that you must generate that number with
substantial care. Look at the details of the algorithms for
generation of UUIDs. It's much more than a simple random number. But
you are otherwise correct. 2.25.<uuid> is a valid OID.

[greggy]

So can anyone tell me why I would not want to use a UID of the form
2.25.<uuid>?

Is there a time or place not to use this form of UID?