xml-dev - Re: Options in XML 1.0

[Jonathan Borden <jborden@mediaone.net>]

Simon St.Laurent wrote:
>
> Eliminating DTDs is _not_ my solution, nor is 'something to that effect'.
> I asked a question about the current layering (or non-layering) approach
> within XML 1.0 parsers, pointed out that it has significant
> interoperability implications, and suggested that maybe this is worth
> closer examination.
>
> Acknowledging problems is the first step toward creating solutions, I
> think, and we're getting there - if you'll drop the overheated rhetoric.

Perhaps my statements are a bit overheated on rereading. I do agree that
there is a logical distinction between the text substitution function of
entities and validation as both are used in DTDs. If you aren't suggesting
eliminating DTDs altogether, you seem to suggest the elimination of many
uses of the internal subset.

In terms of interoperability and in light of John Cowan's comments on
parsers returning portions of the Infoset, let me posit that true
interoperability is enabled by parsers which return the entire Infoset (and
recognize that the Infoset is a subset of the full XML grove).

You may argue that the Common XML subset of the XML grove is a more
appropriate subset, and a priori there is no way to determine what the one
and only proper subset is. On the other hand if we look to SAX, the DOM and
XPath as a baseline, the object model of these already widely implemented
and used technologies roughly fits with the XML Infoset and so I agree that
it is a reasonable baseline to expect parsers to conform to. Another way of
saying this is that I expect my parsers to report all infoset items and
don't mind if they fail to report XML productions which aren't in the
infoset. Suppose we have two parsers each which return the full infoset,
would there be an interoperability problem?

>
> > I say that such a solution does not "fix" XML 1.0 so much as
> >redefine XML into something else than XML. XML is XML warts and all.
>
> Anything which 'fixes' XML has this problem, and there's no way around
> that.  That doesn't mean it isn't worth considering.  For example, the W3C
> seems rather intent on fixing various other parts of XML 1.0, with
> Namespaces/XInclude/Schemas/etc., often in ways which don't work play very
> nice with XML 1.0.  I don't see a lot of complaining about that.

Actually I think there is a surprising hesitence to use recs like XML Names
and on. Personally I don't mind x,y, or z as long as it doesn't limit my use
of XML 1.0 as it stands today  ... or as sung by Dired Straits "I want my ..
PE" :-))

Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org