RE: Some comments on the July 17 XInclude draft

> -----Original Message-----
> From: d96-mst@d.kth.se [mailto:d96-mst@d.kth.se]

> In the second paragraph in section 3, it says external entities with
> multiple top-level elements cannot be included from. Why not? It seems
> to be that such inclusion can be useful.

The Infoset specification does not define an infoset for external entities
as standalone objects.  Thus we can't transform to or from them.  The WG
discussed whether it would be useful to define an infoset for this case, and
decided not to.

My view on this (which may differ from the entire group's) is that:
- External entity documents are not designed to be parsed separately.  Many
implementations will have trouble parsing an external entity document out of
the context of an XML document.
- Since XInclude supports XPointer, there is no need to have external files
that are not well-formed.  We should in the long run discourage having two
types of first class XML-type documents.
- There is a workaround even for documents not under your control - you can
create a wrapper document and include parts or all of that.

> About the issue XInclude-31-which-namespace. It seems like a good idea
> to use the "xml" namespace, however an inclusion instruction like
> 
> <someelement xml:href="foo.xml"/>
> 
> looks a bit strange and doesn't suggest that it's an inclusion.
> However, if you revert to the element syntax, then
> 
> <xml:include href="foo.xml"/>
> 
> would be very clear.

Thanks for your input.  We are clearly leaning back towards a return to
element syntax.

Received on Tuesday, 8 August 2000 13:05:38 UTC