RE: Transformations

At 11:25 99/10/08 -0700, John Boyer wrote:
 >a) Section 4.3.1 Location does not say whether it will allow fragment Ids
 >after the the URI.  It would probably make a better data model if we did
all
 >partial document work in the transformations (and this could be treated as
a
 >special transformation that is required as opposed to recommended).

I think I agree with this. In the HTML context that which comes after a URI
is merely a "view" on the whole document, and thinking that way in our
context can be dangerous; I like transformations. But I also understand
Don's point that we still need to refer to objects elsewhere in the
document. However, I think this can be satisfied by saying the location must
be a URI-clean or IDREF [1].

<John>
I suppose the point I am making (which Don also made in a recent email) is
that semantically there is no semantic difference between putting the ID
fragment in a transformation versus putting it after the URI.  If you want
to refer to something in the same document, you could represent this using

<Location></Location>
<Transformations>
	<Transformation Algorith="urn:dsig:xpointer">SomeID</Transformation>
</Transformations>

or with

<Type>text/xml</Type>
<Location>#SomeID</Location>

The latter is shorter, but it mucks up the data model.  The former is longer
but cleaner in so many ways.
John Boyer
Software Development Manager
UWI.Com -- The Internet Forms Company

</John>

This seems like an abitrary restriction though, and if someone puts and XPtr
in their URI and also specify a XSLT, we need to define what happens, or say
that the signature engine will throw an error.

[1] http://www.xml.com/axml/notes/FindingIDs.html


_________________________________________________________
Joseph Reagle Jr.
Policy Analyst           mailto:reagle@w3.org
XML-Signature Co-Chair   http://w3.org/People/Reagle/

Received on Thursday, 14 October 1999 10:58:04 UTC