Lists Home |
Date Index |
Jeni Tennison wrote:
> Can you run that by me again? Are you saying that it's better to change
> the syntax of XML in order to allow annotation (e.g. by allowing
> attributes to contain structured values) rather than annotating an
> *Infoset* and then revealing those annotations by a processing
No, I am saying the converse. If the Infoset is the barest possible
abstraction from the syntax defined as XML 1.0, then in any case where we
are enlarging (or enlarging our understanding of) the Infoset we are
dealing with XML itself--that is, with the syntactic framework defined in
the Recommendation, and not with any one vocabulary or application of XML.
By the ostensible definition of the Infoset, the basis for any enlargement
of it (or of our understanding of it) must be found in the syntactic
framework defined by XML 1.0. Defining a new vocabulary or application,
however extensive, of that syntactic framework expands the Infoset
abstraction itself not one whit. The serialized infoset schema which you
cite, and quite rightly identify as only a single vocabulary or
application of XML, does not expand the Infoset nor add any new capability
to it. In fact, I am unsure what that vocabulary offers at all.
I must insist (as I have before:
http://lists.xml.org/archives/xml-dev/200209/msg00782.html) that XML is
the serialization; by the nature of the Infoset, no single vocabulary or
application of XML is broad enough to be posited as "the" Infoset
serialization. Indeed, the relationship of XML to the Infoset should work
in the other direction: the Infoset is what might be abstracted from the
syntactic possibilities of XML.
> (If you are, I think that I agree in a way, but I think there would be a
> lot of resistance to changes in XML syntax.)
I wouldn't agree, and I would put up just that resistance.