Re: Type and Structure Re: ASN.1 and XML

[James Clark <jjc@jclark.com>]

Rick Jelliffe wrote:

> The grammar-based schema languages are trying
> to interpose some type system between markup structure
> and the model, without any guarantee that their type systems
> will be any easier to map to a model than the original structures
> were. Witness XML Schemas, which is conceived in terms
> of "components" (its model) but has no analog of "component"
> in either its type system or its markup. TREX and RELAX
> seem to snuffling for diamonds in the same sandpit, with as
> much chance for success.

I do not see either TREX or RELAX as trying to interpose a type system
between markup structure and the model (I assume you mean a
conceptual/semantic model).
In fact, this is one of the big differences in philosophy I see between
TREX and RELAX on the one hand, and W3C XML Schema on the other.

TREX and RELAX both avoid modelling features such as inheritance.  The
focus of TREX and RELAX is, like that of XML, syntax.  TREX and RELAX
are markup-centric: they allow you to describe the syntax of an
XML-based language. They don't build complex information structures
(like the PSV infoset) that attempt to provide a richer model than that
which is inherent in well-formed XML.  TREX and RELAX don't impose
anything on XML other than what it already inherent in well-formed XML. 
That's why, unlike XML Schema, TREX and RELAX don't modify the infoset.

I don't believe the use of TREX and RELAX interposes a type system
between an XML-based language and a semantic model, any more than the
use of regular expressions interposes a type-system between a
string-based language and a semantic model.

James