[dsdl-discuss] Re: DTLL / RELAX NG "libraries"

From: Alex Brown <alexb@griffinbrown.co.uk>
Date: Fri Mar 03 2006 - 10:09:05 UTC

Rick hi

> Because there is no general processing model for XInclude
> that forces an XML parser to do an inclusion, it becomes an
> application-specific action to perform the inclusion. In
> that way, XInclude doesn't offer anything more than home-made
> include statements.

What you are describing here is the sort of 'dumb' include that is close
to DTLL wants, I think. Jeni's draft spec says:

"It is as if the content of the included document (the children of the
<datatypes>element) is inserted into the datatype library in place of
the <include> element."

(Because of the "one namespace=one instance" alignment though, this is
being changed to allow inclusion of other things. However, I'm coming to
believe that in fact DTLL does not need any inclusion mechanism - on
which I'll post later).

XInclude does have the advantage of grasping the many nettles around
'dumb inclusion' (e.g. encoding, Namespaces, onward URL resolution).
 
> Take our own DSDL for example: do we actually in any DSDL
> spec say that XInclusion has to be performed on a document or
> schema before other processing?

No we don't. but I sort of got the feeling (perhaps wrongly) that this
was wanted. But if so it certainly hasn't bubbled through into the
documents.

> So I think XInclude is currently doomed (as a replacement for
> entities) because it requires application/DTD level enabling
> (rather than being part of a dependable infrastructure and
> therefore in any way assumable)

Like XML Namespaces?

Like Namespaces support I think of XInclude support as a parser
'feature' that applications have to specify is required. Because Xerces
supports it it might get some traction ...

> and because element inclusion
> without additional semantics is not an interesting facility
> for people.

We have a number of clients using it, and finding it useful as a 'dumb'
inclusion mechanism.

> (And having
> any additional semantics warrants a different namespace.)

Yes, and maybe name too. I think part of the problem is that the word
'include' is used to mean everything from '#include' to 'smart-merge'.

Ironically, I think it may be some of the mechanisms in DSDL itself
which will make Xinclude more usable. DSRL to patch xml:base support
into a schema maybe? NVDL to hive-off xml:base elements maybe?

- Alex.

--
DSDL members discussion list
To unsubscribe, please send a message with the
command  "unsubscribe" to dsdl-discuss-request@dsdl.org
(mailto:dsdl-discuss-request@dsdl.org?Subject=unsubscribe)
Received on Fri Mar 3 11:08:49 2006

This archive was generated by hypermail 2.1.8 : Wed Apr 12 2006 - 14:48:02 UTC