After lots of interesting (but fruitful) experiences with the XQuery engine in SQL Server 2005, I decided that I'd rather have an XML schema collection for everything than put [1] (that's the numeric predicate meaning "item 1 using 1-based ordinals") everywhere in the XQuery that I needed a singleton. It's a good idea to use "typed XML" anyway. Dan Sullivan thought a good non-trivial use-case would be Office 2003 docs and went forth to create SQL Server 2005 XML schema collections from the publically available schema documents.

First thing he ran into is that specifying processContents="lax" now causes an error when cataloging the XML schema collection. It used to catalog without error and then use processContents="skip" behavior instead in beta1. The Office schemas use this construct, so he's created some code to change it to processContents="skip" so SQL Server 2005 would "like" it. We've run into a couple more cataloging problems, repeated attributes on the same element, etc… Wonder where you report Office schema problems.

If you're wondering what this is about technically, pulling out my favorite XML Quick Reference (by Aaron and Gudge), they define processContents (on xsd:any or xsd:anyAttribute) as "Specifies whether a schema processor should validate the elements/attributes appearing in place of the wildcard". The default is "strict", meaning validate them. Lax means the schema processor *should* (but is not required to) validate if schema information is available. Skip means a schema processor should not validate these.

BTW guys, when's Essential XML Quick Reference Volume 2 due out? There *lot's* more specs now. Maybe you could publish an Essential XML Reference Encyclopedia?