- From: <Noah_Mendelsohn@lotus.com>
- Date: Fri, 15 Jun 2001 13:04:26 -0400
- To: "Venkateswar Wunnava" <wvsvenkat@worldnet.att.net>
- Cc: jeffrafter@definedweb.com, wason@mindspring.com, xmlschema-dev@w3.org
A few thoughts on the global/root question: First of all, there is a brief treatment, adapted from a note I sent months ago, in the schema FAQ [1]. There are really at least two questions here: designating roots, and globals vs. locals. The second question is relatively easy: globals are the sorts of declarations you have in DTDs. If you do the obvious conversion of a DTD to a schema, then every element is global. Period. Locally scoped elements allow you to model cases like: <BOOK> <TITLE>Moby Dick</TITLE> <AUTHOR> <TITLE>Mr.</TITLE> <LASTNAME>Melville</LASTNAME> </AUTHOR> <BOOK> Clearly the book title and the author title are different beasts; the latter probably is an enumeration of Mr./Ms., etc. Locals let you express this; DTD's (globals) do not. Don't try to read more into globals/locals then that. The FAQ covers the root question, as does Tom, but I'll reiterate a bit. Not only are there times when you want to allow multiple root names from a given namespace (maybe it's OK to root at either BOOK or BOOKLIST), what's more important is that you might want to do partial validations. Maybe you only want to validate an <AUTHOR> or a <SHIPPINGADDRESS>. This is a very common case for editors. If we had explicit roots, then things would be more complicated. I'd need two validation modes: whole documents and fragment. As it stands, there's just one: you can validate from any element, top level or not, using any matching element declaration or complex type. So, leaving out root as a distinguished concept let us cover more cases with a single mechanism. [1] http://redrice.com/schemavalid/faq/xml-schema.html#d6 ------------------------------------------------------------------------ Noah Mendelsohn Voice: 1-617-693-4036 Lotus Development Corp. Fax: 1-617-693-8676 One Rogers Street Cambridge, MA 02142 ------------------------------------------------------------------------
Received on Friday, 15 June 2001 13:09:14 UTC