Survey for XML Schema CR-37: Default prefix for schema namespace

I just got a response (http://lists.w3.org/Archives/Public/www-xml-schema-comments/2000OctDec/0416.html) to CR issue 37 which was raised in response to 
http://lists.w3.org/Archives/Public/www-xml-schema-comments/2001JanMar/0240.html

Basically, the suggestion was that the default values ('') for the p and s parameter entities (schema definition suffix and prefix) in the DTD for schema were only appropriate for a small minority of
schemas (however that minority includes the schema for schemas) and that default values that were more appropriate could substantially reduce the need for schema authors to start their schemas with
boilerplate redefinitions of the parameter entities.

Definitely, having bad defaults is only a nuisance if you already know your way around parameter entities and duplicate ATTLIST declarations.  However, it does force you to introduce these concepts to
anyone that you are teaching how to work with schemas.

In the response which was a decision to maintain the current defaults, there was a mention that it was not know how many schema or schema authors would benefit from a change of the default values.
Which leads me to propose an little unscientific survey for those of you who have written schemas that are validated against the DTD for schemas.

Please respond to carnold@houston.rr.com and I'll summarize on xml-dev tomorrow.

1. What fraction of your schemas that have a document type declaration define an internal subset declaration along the lines of:

<!DOCTYPE xsd:schema PUBLIC ""-//W3C//DTD XMLSCHEMA 200010//EN" "XMLSchema.dtd"
[
<!ENTITY % p 'xsd:'>
<!ENTITY % s ':xsd'>
<!ATTLIST xsd:schema
	xmlns CDATA #IMPLIED
	xmlns:xsd CDATA #IMPLIED>
]
>

2. What fraction of your schemas that have a document type declaration and are not Schema for Schema, use the current defaults of '' and '' or would otherwise be adversely affected by the suggested
changes?

3. Would changing the default values of the p and s parameter entities to 'xsd:' and ':xsd' and adding declarations within the DTD for optional attributes on the schema element for xmlns, xmlns:xsd
and xmlns:x (for the XML namespace) eliminate the need for a internal subset in most of your schemas?  If not, what else would you need to eliminate the internal subset?

4. Would you have any objections to changing the default values for the parameter entities and adding definitions for the namespace declarations?

Received on Wednesday, 14 February 2001 14:37:23 UTC