Lists Home |
Date Index |
> -----Original Message-----
> From: Rick Marshall [mailto:email@example.com]
> 1. in a general sense we discovered there is no unique solution to some
> problems and the nature of the relationship between nodes is critical to
> a correct calculation (a full descending dom approach would probably
> avoid this problem, but it is not really general and when you factor in
> xquery you will run into very major problems)
I'm only vaguely familiar with XQuery, so I'd really like to know what major
problems you see. XPath works fine with my implementation, and I hear the
two are 'almost' identical!?
> 2. to work well you need to build in a calculus, and then programmers
> tend to fail as the system takes over (ref perma thread on recursion etc)
Note to self: add a recursion detection feature.
> to work, the rules will need to exist
> in the context of a schema. personally i think an extension of xsd is
> probably the correct way to do this.
I don't disagree. In fact, if I had a voice in the XML Schema working group,
I'd be pushing for such an extension. However, XML Schema is too complex
already; so even if we want to extend it, then IMO it should be done in a
separate working group and made to be XML Schema-friendly (or, better yet,
friendly to schema languages in general). I'm thinking something like the
'Composable' architecture of the WS-* specifications.
I've added a section in the tutorial about including XRules in the
context of XSD, and I consider this a *key* feature. It's only that,
anticipating that it won't be easy, I'm still trying to gather enough
courage to start implementing it.