Lists Home |
Date Index |
On Sat, 22 Feb 2003 19:41:16 +0000, B2ll de hÓra <firstname.lastname@example.org> wrote:
> Well maybe, but the very first thing to do there is determine a way to
> negotiate features between processors.
Yup. I think Sean McGrath suggested something like this years ago, and
nobody seems to have picked up on the idea. I don' remember Sean's syntax,
but it could be something like:
<?xml version="2.0" encoding="UTF-8" features="this that something" ?>
The features and strings might be defined in some future XML spec, e.g.
EE="expand external entities" DA = "default attribute values", etc. I'm
probably horribly naive to think that the enumeration of features would
relatively easy to get agreement on :-)
So, a parser would "advertise" the list of features it supports its
documentation and maybe via some standardized API call, and instances would
declare which features it requires; mismatches could be detected
immediately, or apps could call whichever parser is optimal for the
requested features, or whatever.
Reasonable people can disagree on whether the features in XML 1.0 are so
minimal that they should all be one "feature," but this kind of thing
would make a real difference when considering the implications of all the
other stuff that might be rolled into or out of XML n+1, such as DTDs,
namespaces, xml:base, xml:id, not to mention XSDL, XPointer, etc. etc. etc.