Dear,

Maybe a little background first: 
I am in process of delivering a XOM-like java library to XMLSchema called 
'endymion', that will be released under LGPL 
as part of the 'narcissus'-project. The goal of the API is to offer a heavily 
typed, java-like and simple access to the information
embedded in an XMLSchema.

At first I started of with parsing XMLSchema myself, but surrendered after a 
while becaause of the complexity it brough upon combined
with the apparent liberal interpretations of the XMLSchema spec of some other 
parsers. I decided on the XMLSchema infoset model (org.apache.xs.*) 
in Xerces because it is light, has active developers and tries to adhere to 
specs.

But, of course, I am encountering some issues (let's call them challenges :-))

* I would like to validate XML to specific complex types in XMLSchema's. (so 
without them being declared as XMLSchema instances...) I would also like to be
able to validate in memory DOM document trees / XOM document trees / JDOM 
document trees / SAX parse events with a non Xerces parser /... 
I think I am familiar enough with the Xerces internals to know that this should 
not be that difficult. For example for the SAX Parser
I would only need to simulate an XNI-pass by making an adapter class that 
implements DefaultHandler and translates all events to a XMLDocumentValidator 
as if in a a XNI-pipeline. This adapter class seems to be pretty generic (can 
make all XNI - XMLDocumentHandler act upon SAX events except for XNI events)
that I was wondering if it didn't already exist, or if somebody could give me 
some tips with regards to this.

* I see that in CVS there have been several changes / updates with regards to 
the visibility of fields and methods in several API classes. I was wondering 
if there is a certain goal or reason for permorming these changes and when they 
are finished / released. It would be really cool to have some public hooks
into the org.apache.xerces.impl.dv.xs.* - value-classes to be able to access 
the actual values of DecimalDV and others.

I am very eager to contribute back any bug fixes / changes that I will develop 
so even if you only have some guidelines as opposed to full fledged 
solutions I will be very happy...

Mik

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to