- From: Tim Berners-Lee <timbl@w3.org>
- Date: Thu, 15 Jun 2000 15:29:15 -0400
- To: "James Clark" <jjc@jclark.com>, "Henrik Frystyk Nielsen" <frystyk@microsoft.com>
- Cc: "John Cowan" <cowan@locke.ccil.org>, "David Carlisle" <david@dcarlisle.demon.co.uk>, "David Turner" <dturner@microsoft.com>, <XML-uri@w3.org>, "Andrew Layman" <andrewl@microsoft.com>
-----Original Message----- From: James Clark <jjc@jclark.com> Date: Thursday, June 15, 2000 5:35 AM James wrote, >In the case of external entities, the only thing that an XML processor >needs to do with the system identifier is to fetch the resource it >identifies. It doesn't have to compare it for equality with other >system identifiers. In the case of namespace names, the namespace >processor doesn't have to fetch the resource it identifies, but it has >to compare it with other namespace names with possible different >contexts, in order to determine whether attributes are unique (in >accordance with section 5.3 of the namespaces Rec). Following the decision that the base URI for a URI reference in an extrenal entity is the URI of that entity, it is clear that an XML processor -- or at least a parser which creates an infoset which is passed (eg through the DOM) to an application must provide the information as to where a relative URI came from. So external entities cannot be expanded in place in the document tree without annotating them with this. I guess this is obvious. But the base URI can't be lost. (Of course, anything which feteches a resource in practice does to string comparison with URIs of resources already fetched, before fetching them again!) >James > >
Received on Thursday, 15 June 2000 15:30:38 UTC