- From: Anthony B. Coates <abcoates@TheOffice.net>
- Date: Wed, 18 Sep 2002 12:15:23 +0100
- To: xmlschema-dev <xmlschema-dev@w3.org>
** Reply to message from "Joe Warwick" <joe@icon101.com.au> on Tue, 17 Sep 2002 10:38:44 +0800 > I would love to know if there is an industry standard way of mapping an > inheritance typed object instance tree into a relational database while > maintaining a high level of normalisation. No. Not a good workable one, anyway. > What gets me is how to choose the point in the tree where flattening the > objects is most efficient, yet causes least xml overflow blobs. > Has anyone managed this task? It seems like it would be best to build an > object oriented system on top of sql server through translation stored > procedures. This would only be suitable for small-scale usage, certainly not on an enterprise system. You would be coupling your database schema directly to an XML Schema, and if that Schema is an external one, version changes could be unpredictable. Also, you would probably need to support multiple Schema versions simultaneously (2 at least), and my experience is that relational databases don't make it particularly convenient (i.e. transparent) to use multiple DB schema versions in the same database. > I know that SQL server can create a database from an XML Schema, but nowhere > does it say how it handles extension and restriction types. Both SQL Server, Oracle, & DB2 are moving to native XML support. I think this will negate any momentum to define a really good XML to relational mapping (if it can be done at all). No idea yet how "native XML" support in relational databases will compare to purpose-built native XML databases, though. Cheers, Tony. ==== Anthony B. Coates, Information & Software Architect mailto:abcoates@TheOffice.net MDDL Editor (Market Data Definition Language) http://www.mddl.org/
Received on Wednesday, 18 September 2002 07:16:07 UTC