W3C home > Mailing lists > Public > xmlschema-dev@w3.org > March 2002

RE: Attribute groups

From: Naren Chawla <naren_chawla@attbi.com>
Date: Fri, 15 Mar 2002 08:16:53 -0800
To: "Jeni Tennison" <jeni@jenitennison.com>, "Snow, Corey" <CSNOW@ddpwa.com>
Cc: <xmlschema-dev@w3.org>
Message-ID: <AOEGLCJGIFANMEODKIBGCEFLCCAA.naren_chawla@attbi.com>
Jeni

>
  - Express as much as you can in XML Schema, then use Schematron or
    another schema adjunct to express the co-occurrence constraint.
>

Is above a practical suggestion from a systems development perspective. Do
you have a small example of how to achieve this ?

I know there was discussion earlier about full support for "co-occurrence
constraints" and we wanted to wind down further discussions.

I thought I will squeeze in one small thought -

From a practitioner's stand-point, I feel even if XML Schema is not "Turing
complete", it should be "validation complete" (as Jeni suggested earlier).
Though latter might lead to former. I would love to get rid on one more
layer (validation layer) from my application.

It appears to me "constraining the contents of XML instance" is a major goal
behind XML Schema. Being able to achieve this successfully will lead us to
exciting ways to develop applications broadly termed as "Schema-based
programming".

--Naren











-----Original Message-----
From: xmlschema-dev-request@w3.org
[mailto:xmlschema-dev-request@w3.org]On Behalf Of Jeni Tennison
Sent: Friday, March 15, 2002 2:00 AM
To: Snow, Corey
Cc: 'xmlschema-dev@w3.org'
Subject: Re: Attribute groups


Hi Corey,

> What I'd like to be able to do is make the attribute *group*
> optional, based on whether one of the attributes in the group (the
> required one) is supplied- in other words, if the attrOne attribute
> is supplied, an attrTwo attribute may optionally be supplied. If
> attrOne is not supplied, attrTwo CANNOT be supplied, but the
> document remains valid if neither is supplied.

What you're describing is a "co-occurrence constraint" in which the
presence of one attribute affects whether the presence of another
attribute is allowed.

XML Schema is notoriously bad at co-occurrence constraints. The usual
advice in this situation would be to do one of the following:

  - Express the distinct attribute combinations as separate complex
    types both derived from the same abstract complex type; make the
    element have the abstract type as its type; use the xsi:type
    attribute to identify which of the two types is being used in the
    particular instance. (Note that this involves changing the look of
    your XML instance documents.)

  - Express as much as you can in XML Schema, then use Schematron or
    another schema adjunct to express the co-occurrence constraint.

If you want more details on either of these options, let us know.

Cheers,

Jeni

---
Jeni Tennison
http://www.jenitennison.com/
Received on Friday, 15 March 2002 11:16:19 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 11 January 2011 00:14:30 GMT