Lists Home |
Date Index |
Bullard, Claude L (Len) wrote:
> But both are just means of association with respect
> to semantics. And we don't have a standard means
> of describing semantics.
Seems to me that semantics are most naturally
related to types. ... But we're not allowed to
use that word here, huh?
> No matter how we slice this, James is right. If we
> want to associate semantics/behaviors, something
> more layered has to be at the other end of that
What do we call the set of instances that can be
mapped to a specific base architecture through a
single architectural form? What do we call the
set of all instances amenable to processing by a
single DSDL? If we had names for these things, we
might find we had a nice hook to which processors
could attach semantics.
I'm becoming convinced that all XML documents have
an important property and that we don't have a
good name for that property. "Document type"
comes close, but I think the term carries too much
baggage to carry a sufficiently precise or a
sufficiently general meaning. Root element
namespace+localname is too imprecise.
I'll try to describe this property below using the
decidedly un-catchy term "representational form"
an alternate term might be "lexical type".
Documents are created with the intent of
communicating some information. The document
creator encodes the information according to the
rules or conventions of some representational
form(s) in order that some processor(s) can get at
Representational forms define sets. A document
instance either is or is not a member of the set
defined by a particular representational form. An
instance is often a member of many sets -- it has
multiple values for for its
"representational-form" property. There is
usually (always?) one predominant representational
form that most precisely matches the instance
These sets may nest (subset through restriction
constraints) and intersect. All XML documents are
members of the "well formed XML" representational
These representational forms may be used as
contracts. Document instances can be tested to
see if they conform to the rules. These rules may
be expressed in any number of schema languages,
procedural code and prose documentation.
We often define specific processing (i.e. XSLT
stylesheets, DSDL validating) for specific
representational forms. Which processing
we choose depends largely on our local context.
It is useful to be able to give distinct
representational forms distinct names.
It is often useful for a document to assert, in
the instance, that is a member of the set defined
by a representational form. We might deduce the
form's name through an instance's DOCTYPE
declaration or the root element's namespace +
local name. There are problems with both of these
What I don't know:
* Is the representational form
property intrinsic, extrinsic or emergent?
* Is this property fixed for the life of the
document, or does it change over time?
* Could this property be also be obtained for