W3C home > Mailing lists > Public > xmlschema-dev@w3.org > June 2001

Re: Semantic of the unique/key/keyref constraints.

From: Henry S. Thompson <ht@cogsci.ed.ac.uk>
Date: 19 Jun 2001 12:29:50 +0100
To: vdv@dyomedea.com
Cc: "xmlschema-dev@w3.org" <xmlschema-dev@w3.org>, David Fallside <fallside@us.ibm.com>
Message-ID: <f5bwv68btch.fsf@cogsci.ed.ac.uk>
ht@cogsci.ed.ac.uk (Henry S. Thompson) writes:

> I think you're right, this example is bogus (the complete schemas are
> all OK, but this out-of-line example is wrong).  It's clearly _part_
> elements that are (meant to be) unique, so the 'selector' should be
> "r:regions/r:zip/r:part".  If they're just supposed to be unique
> within a single _occurrence_ of the <zip> elt, then all that's needed
> is a single "field='@number'".  If they're supposed to be unique
> across multiple <zip>s, then as it stands with the restricted XPath
> expression subset the REC allows, I don't think you can do it.

I wasn't completely clear here.  The simple alternative offered above
requires <part> to be unique wrt @number throughout the
<purchaseReport>.  This means they're unique across multiple zips,
_whether or not_ the zips are different.  It's _allowing_ the same
part number under different zips that can't be accommodated by the REC 
as it stands.

  Henry S. Thompson, HCRC Language Technology Group, University of Edinburgh
          W3C Fellow 1999--2001, part-time member of W3C Team
     2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 131 650-4440
	    Fax: (44) 131 650-4587, e-mail: ht@cogsci.ed.ac.uk
		     URL: http://www.ltg.ed.ac.uk/~ht/
Received on Tuesday, 19 June 2001 07:29:49 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:55:52 UTC