[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [xml-dev] Re: determining ID-ness in XML
"Champion, Mike" wrote:
> How about
> 5) Refactor the various specs so that they layer cleanly, and add some sort
> of "features" or "profile" string (in the XML declaration? As a PI?
> namespace URI pointing to a RDDL document?) to negotiate the feature set
> between the processor and the application.
> I know this is easier said than done, but seems like the right conceptual
> solution to me.
I think that, in any practical implementation, this amounts to (4). I would not
advocate refactoring because 1) we are looking for stability, and recreating
specifications which (even assuming that any consensus can be found once they
are reopened) is too high a price to pay in uncertainty and churn; and 2) one
man's clean is another man's dirty, which is my basic point. The constituent
specs look cleanly layered, or conversely thin and uncoordinated, only in light
of a processing model which projects a particular conceptual whole onto them.
In short, we cannot expect 'document-centric' processing if by that we mean
processors which are obliged to carried out the intent of a document creator as
stated in some canonical form within the instance document. Processing is
processor-centric. A 'negotiation' between a document and a processor consists
of some portions of the instance document providing the evidence on which a
processor determines, by its own functional rules, what that document is in
respect of its capability to process it.