RE: [ubl-cmsc] Schema Derivation -- Do We Need It?

[Matthew Gertner <matthew.gertner@schemantix.com>]

Title: RE: [ubl-cmsc] Schema Derivation -- Do We Need It?

Arofan,

 

My understand is that RELAX NG does not provide any extension mechanism, but that it is designed to enable this to be cleanly layered on top. There is already an effort that has produced some results along these lines, although I haven't had a chance to look at it yet. Eduardo managed to get some details from someone in the RELAX NG TC; I'll forward the mail to the list.

 

I see the idea of creating a clean extension mechanism for RELAX NG (or adapting the existing work) based directly on requirements derived from UBL to be a very appealing one. I would propose, however, that we complete work on the context methodology while remaining agnostic towards derivation mechanisms, and then take a closer look at the feasibility of this.

 

Cheers,

Matt

-----Original Message-----
From: Gregory, Arofan [mailto:arofan.gregory@commerceone.com]
Sent: Tuesday, December 11, 2001 9:52 PM
To: 'Matthew Gertner'; UBL CMSC Listserv (E-mail)
Subject: RE: [ubl-cmsc] Schema Derivation -- Do We Need It?

Matt:

I am in general agreement that derivation should be expressed in schema, for the reasons you express. Commerce One is an example of a company whose technology *does* leverage extensions expressed in schema for reasons similar to what you have outlined here, so I don't think this is entirely a futures discussion, but I agree that this will either become a standard use of XML or it won't. I also think that it will.

A point of correction: it turns out that RELAXNG has a way of self-extending which allow you to include an extension mechanism. We should ask Eduardo further about this, as he has looked into it more than I have, but from what I understand it *is* possible to have RELAXNG extensions. We might even want to look into the ability of extending a schema language as a way to get all the functionality we need. RELAXNG might support this. (Radical idea, eh?)

Cheers,

Arofan

-----Original Message-----
From: Matthew Gertner [mailto:matthew.gertner@schemantix.com]
Sent: Tuesday, December 11, 2001 12:39 PM
To: UBL CMSC Listserv (E-mail)
Subject: [ubl-cmsc] Schema Derivation -- Do We Need It?

I am attaching a discussion document that I have written in the hope of
sparking some discussion on one of the topics raised during our call last
week: do we actually need to use the derivation features of schema languages
like XSD when creating new types through the application of context. I have
modestly called it version 0.1, since I think it could probably use a lot
more discussion, work and fine-tuning. It is probably a decent start to the
discussion, however.

At this point I am primarily interested in knowing whether *anyone* out
there is with me on this. Am I a kook for believing that XML schema-driven
software development will be a popular practice in the future? Any and all
comments would therefore be most appreciated.

If there is some agreement that this is a valid issue and approach to the
problem, I have in mind some forums (like the XML-Dev mailing list) where we
could post a later version of this paper to get more valuable input from
technical experts who are not participating directly in UBL.

Cheers,
Matt

PS: Many apologies for sending out a Word document. I'll make a PDF version
tomorrow and send it to the list.