>>We're definitely trying to get Xerces up the 12/17 draft (and successive >>drafts). The holidays have contributed to some slowness on this. The other >>thing is >>that lots of parts of the spec have changed, and in ways that will force major >>changes in the way that validation currently works.
I don't envy the amount of work involved. I'm currently doing an initial read of the 12/17 draft since I am finally home after a non-stop month of travel starting with XML '99. It definitely looks like a radical (and desirable) change from the September draft, though there are definitely aspects of the May 99 draft that I still miss (like the choice element and separate elements types for references and declarations). I am curious about some of the approaches or resources that you are thinking of using on parts of the effort. I haven't seen these appear as proposals or in the status, but they seem like things that should be published. a) How are regular expressions for datatypes being considered to be implemented. Is there a tolerable open-source implementation that could be incorporated or is the intention to write a custom regex evaluation engine. b) The use of XPath's (especially ancestor selectors) for the key's would seem to make an event-based validation of keys problematic. There is a editor's note expressing a willingness to consider restricting the XPath's used in keys. Is anyone lobbying to simplify this, or is there some trick I'm missing that would make this easy. c) I think there Xerces should provide some hook when resolving schema locations so that the application can redirect to a local resource (local XML file or some in-memory representation).
