OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

ubl message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: Re: [ubl] UUID and GUID

The ISO object identifier (OID see X.660 in reference cited below) has the benefit that all IDs are guaranteed globally unique and traceable. Any owner of an ID instantly becomes the naming authority for numbers under its branch. This approach has many advantages over the "pseudo" unique approach provided by UUID. UUIDs are only statistically unique and there is some finite risk of an ID collision. Given a UUID there is no way to trace its origin. Given an OID, you can trace the one and only one worldwide tree of naming authorities to find the "owner".
We have used OIDs in ANSI metering standards for meter identifiers, serial numbers, customer identifiers, and network identifiers. They are also extensible and compressible.
In a message dated 2/9/2006 1:46:48 P.M. Eastern Standard Time, jon.bosak@sun.com writes:
Hello UBL TC,

At the Manhattan UBL TC meeting, I took the AI to check on UUID
vs. GUID and figure out which of these was proprietary.

The answer to that question is: GUID is a proprietary Microsoft
implementation of UUID.  We can't be referencing proprietary
specifications here, so we have to use UUID.  (The OASIS ebXML
Registry Standard has always used UUID, by the way.)
Consequently, we will have to change all occurrences of GUID in
the schemas to UUID.  I'll enter this in the issues list once we
settle on a UUID reference.

The reference question is not as simple as one would hope.  For
background, see the Wikipedia entry on UUID.  Here's my take after
reading that article and doing a little bit of checking around:

- The original 1997 UUID spec from The Open Group is not what we
   want, despite the fact that this is the spec referenced in ebRR
   3.0 as

      [UUID] DCE 128 bit Universal Unique Identifier

- Possible UUID specs include:

    - ISO/IEC 11578:1996, which is not freely available online

    - IETF RFC 4122, which is available at


    - ITU-T Rec. X.667 (2004) | ISO/IEC 9834-8:2005, which is
      based on RFC 4122 and can be found in "prepublished" form at


Ordinarily I would go with the RFC simply because it's freely
available.  However, the ITU-T/ISO/IEC spec is also freely
available as an exception to the usual rule, and I'm informed that
an agreement between ITU and ISO will keep it that way.  So my
recommendation is to reference that one unless someone sees a
reason not to.

Please post any thoughts you might have on this, and let's aim to
resolve the issue in next week's Atlantic TC call.


To unsubscribe from this mail list, you must leave the OASIS TC that
generates this mail.  You may a link to this group and all your TCs in OASIS


[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]