Re: XML Schema issues resolution (LC-55, LC-89. LC-115)

I guess you will have to mark me as unsatisfied with your described
resolution to LC-55, willing to concede LC-89 as insignificant (especially
relative to other schema location issues) and satisfied with LC-115 given
proper documentation.

LC-115: One thing that I am satisfied with is the behavior of unqualified
QNames (implied in your response but also a distinct issue LC-115), as long
as it is explicitly stated that they are bound to the default namespace (in
someplace more prominent that in the remarks of an example in the primer).
That isn't a given since attribute names in instance documents (or the value
of the name attribute of a <xsl:attribute> element) are QNames but are not
bound to the default namespace.

LC-55: There would be only a slight annoyance factor in separating built-in
datatypes into a distinct namespace from the schema definition elements.
Basically, a schema author would have to add a
xmlns:dt="http://www.w3.org/1999/XMLSchema-datatypes" to the xsd:schema
element use dt:integer instead of xsd:integer, and XMLSchema.xsd would
import instead of include part2.xsd.  I do not see that as a far-reaching
and disruptive impact.  The benefit would be the ability to reuse datatypes
in RDF and other applications without bringing over the entirety of the
schema definition language or, worse yet, creating a mirror type hierarchy.

> LC-89:
>   You observed that there was no way for an instance, as opposed to a
> schema, to appropriate a schema with no target namespace for use in
> validation of elements (and attributes) with qualified names.  You

Actually, my issue was the reverse  There is not a way for an instance that
uses unqualified names to reference a schema that has a defined target
namespace.  The only thing that prohibits an non-namespaced instance
document from using a namespaced schema is one sentence in section 6.3.2 at
the end of point 3.

>The second similarly provides a URI reference as a hint as to the location
of a >schema document with no targetNamespace [attribute].

However, the edge case of how a schema-aware, non-namespace aware processor
handles a hint may not be significant to get picky about.  Especially, since
there are much more significant issues with locating schema resources whose
resolution could make this issue irrelevant.

Received on Wednesday, 13 September 2000 12:59:32 UTC