RE: [xml-dev] Were these derived from a logical data model: XSLT, XML Sc

["Michael Kay" <mike@saxonica.com>]

> 
> XPath 2.0, XSLT 2.0 and XQuery 2.0 do use a logical 
> (abstract) data model to describe XML.

They use a data model to describe the inputs and outputs of the language,
but not to describe the language itself, which is I think the question that
Roger was asking. 

No-one ever sat down and drew a UML diagram showing the relationship between
an xsl:choose instruction and its xsl:when and xsl:otherwise children.

XML Schema comes closer, in that there is an abstract model of a schema that
is distinct from the concrete syntax of a schema document. But even there, I
think it would almost certainly be unhistorical to suggest that the abstract
model came first and the language was then designed around it.

It's worth noting that a vocabulary like XSLT is synthetic: unlike a banking
transaction or a tax return, it is not trying to model something that exists
in the real world. UML and similar modelling techniques are helpful in
developing a shared understanding of the objects and relationships that
exist in the real world (or in an existing system of some kind). XHTML is
arguably derived from an analysis of the nature of documents in the real
world (involving decisions such as whether a table is to be regarded as a
kind of paragraph) - but I strongly suspect that the analysis was not done
using any formal or semi-formal method.

Michael Kay
http://www.saxonica.com/