RE: [xml-dev] The triples datamodel -- was Re: [xml-dev] SemanticWeb permathread, iteration n+1

[Alaric B Snell <alaric@alaric-snell.com>]

On Tue, 8 Jun 2004, Paul Sumner Downey wrote:

 >>> > PNG files learnt from this; PNG chunks have a flag in them
 >>> > which indicates if they are necessary to decode the image
 >
 >>
 >> this sounds like the 'mustUnderstand' flag in the SOAP
 >> extensibility model.


Indeed. It's slightly more advanced due to the 'drop me if you modify the
image data' bit, but still inadequate for more complex data structures
IMHO.


 >> the nub of the argument here, AIUI, is whether extensibility
 >> is enabled generically by default in an XML processor (you
 >> must ignore what you don't understand) or if each and every XML
 >> vocabulary must define its own extensions and versioning model
 >> and get it right in the first version of that vocabulary.


Indeed!

There's much more chance for horrible errors if each vocabulary has their
own rules (particularly when you consider that many people making up
vocabs aren't really well educated in the implications of their choices),
but lacking a generic extension mechanism, it's all we have :-)

So, who's up for defining a generic toolkit for expressing how
applications should handle unknown elements (and attributes? Or can we
assume that *all* unknown attributes are ignorable?)

<element ext:ifUnknown="ignore" ext:dependsOn="...list of IDs...">
   ...
</element>

...or something similar? Probably could be a lot less verbose!

 >>
 >> Paul
 >>

ABS