[relaxng-user] conformance and datatype libraries

[David Tolpin]

> like the RELAX NG spec to define an abstract mechanism.  Such an abstract mechanism 
> helps implementors to design APIs, especially because the RELAX NG spec already provides 
> datatypeEqual and datatypeAllows.  Is this your point?

Almost. I would like RELAX NG spec to either refrain from specification of services
a datatypeLibrary provides, or to specify them fully. 

Actually, RNV has exactly on point of plugging a datatype library
in, and reports all errors (that is, it says that a certain type
has incorrect parameters or name or value), except that type errors
are delayed till the validation stage. This lazy evaluation of
grammar is not fully conformant, but allows use of datatype libraries
with the same services provided as described by the spec.

> 
> >   2) Section 4.16 does not specify whether a schema with a value that is not allowed
> >   by its datatype is valid or not.
> 
> This is true.  I think this is a bug.
> 

A bug in which direction? Should a schema with a QName value with namespace prefix not
declared in the schema be rejected?

David Tolpin