Treating namespace URIs as schema names

 From: MURATA Makoto <>

> In his note entitled "XML Notation Schemas", Rick pointed out
> "two severely broken non-conformances", namely:
> - treating namespace URIs as schema names.
> I believe that Rick is right.

I should point out that I have no objection if a namespace URI points to *a*
schema, as long as there is no idea that it is *the* schema for the data.

I believe the schema group should define a PI for linking to schemas, just
like the stylesheet PI, allowing multiple schemas (expecially from competing
vendors) each
of which is appropriate in different roles.

    role="for data entry"
    namespace=""  ?>
    role="for extra checks on attributes"
    namespace=""  ?>
    role="for client-side acceptence"
    namespace=""  ?>

The role attribute should have some kind of controlled vocabulary to select
when each kind of schema is appropriate.

With this mechanism in place, I don't think there will be much resistance to
then saying that the namespace URI may *also* point to a schema (i.e., the
schema the author thinks is most useful for the general users of the data),
as a shorthand for having a PI. (Similarly, the DOCTYPE declaration can be
defined as an equivalent form to the last PI.)

But I think you will find tremendous resistance to anything that  people may
will tie their data to particular schemas, particular schema languages and
especially to particular schema tools. I think the principle of
extensibility demands that different schema types be allowed (e.g. just as
different stylesheets languages are allowed).

The XML Schema WG might also find this attractive in that it might allow
them to
go ahead with the current minimal schema proposals without as much flack:
by making a mechanism by which other people can develop other kinds of
schemas, the XML Schema WG does not need to justify that the current
draft does not meet some of the criteria from the user-requirements
(in that it adds some syntactic sugar to parameter entities, and a nice
veneer of inheritance,
but it does not provide any improvement to users in the targets of
validation; people
need a new schema language because of more than the supposed syntactic
deficiencies of
XML markup declarations.)

Rick Jelliffe

Received on Tuesday, 25 May 1999 00:37:49 UTC