Chris Smith smith at
Tue Mar 24 07:09:16 GMT 1998

On Mon, 23 Mar 1998, Michael Kay wrote:

> I have been using IDREF to represent relationships within a
> genealogical data set, but with considerable reservations
> wasn't so full of hyperlink behaviour. I want to model a 
> relationship between two objects that happen to be in different
> datasets. All the XLL stuff about the manner in which links are
> followed is to me entirely presentation-oriented and I don't
> want it in my data thankyou very much. From that point of
> view I prefer the IDREF style of specification which says
> nothing about the behavioural characteristics of the link.
> What I really need is a "distributed IDREF".

On OpenTradingProtocol, we are using ID/IDREF in much this manner,
with the additional riders - IDs, once attached to an element, are
permanent because the entity may be authenticated (including
attributes) - and (the tough one) elements may be picked up from
several previous documents ("messages") and assembled in a new
document with a guarantee that the unchanged IDs will not conflict!

We only achieved this by clearly specifying the generation process for
the *content* of the IDs.

I think it was worth the effort, because it allows all the pieces of
the transaction to be readily accessible. It should even be possible
(although not required) to store all the components of an OTP
transaction (many messages) in a single document for receipt purposes
without the IDs conflicting.

 Chris Smith                                          <smith at>

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at
Archived as:
To (un)subscribe, mailto:majordomo at the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at

More information about the Xml-dev mailing list