[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: DTD and Schema Teasers

• To: "Bullard, Claude L (Len)" <clbullar@i...>
• Subject: Re: DTD and Schema Teasers
• From: Webb Roberts <webb.roberts@g...>
• Date: Wed, 17 Dec 2003 10:58:05 -0500
• Cc: "'xml-dev@l...'" <xml-dev@l...>
• In-reply-to: <15725CF6AFE2F34DB8A5B4770B7334EE03F9F08B@h...>
• References: <15725CF6AFE2F34DB8A5B4770B7334EE03F9F08B@h...>
• User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4

Bullard, Claude L (Len) wrote:
> Two only slightly related questions:
> 2.  What are good reasons for having multiple schemas for the same 
>     instance?

This is something we're running into in the justice data model 
effort (justicexml.gtri.gatech.edu).  We have a fundamental 
conflict between two schools:

1. Shared schema:  This provides a common definition for syntax 
and use of the data dictionary.

2. Document-specific schemas: These state document / 
implementation-specific constraints on a document.

It's a fine line to walk.  (1) is trying to solve the problem of 
allowing machines to say things that can be understood, and (2) 
is trying to prevent machines from saying things outside a 
desired data space.

We've found that when attempting to apply document-specific 
constraints, you pretty much have to throw out the concept of 
common schemas and reused types.  For example, if one part of a 
document defines a type one way, another part of the document 
can't define that type another way.  And there seems to be no way 
in XMLSchema to manage common / reusable schema components in 
tandem with application-specific constraints.

One solution might be to disallow use of type-referencing in 
instance documents (use of xsi:type=), and pass over instances 
with validation against multiple schemas.  That makes reuse of 
global element definitions difficult, among other things.

We're starting to lean towards a solution like schematron for a 
solution, but that steps away from the desired minimum toolset, 
and requires a good bit of sophistication from the users.

Webb Roberts (webb.roberts@g...)
Research Scientist
Georgia Tech Research Institute

• References:
  • DTD and Schema Teasers
    • From: "Bullard, Claude L (Len)" <clbullar@i...>

• Prev by Date: Re: DTD and Schema Teasers
• Next by Date: Re: DTD and Schema Teasers
• Previous by thread: Re: DTD and Schema Teasers
• Next by thread: Re: DTD and Schema Teasers
• Index(es):
  • Date
  • Thread