RE: Draft minutes of TAG Teleconference of 23 April 2007

In the minutes it says:

| Marc de Graauw wrote an article on xml.com, spurred in part  
| by our earlier discussions. He proposes you give not a single 
| version, but indicate each
| version that you believe the document conforms to.

One correction: the gist of what I'm saying is not indicate each version
sender believes the document _conforms_ to, but each version (or more
general: language capability) the sender _requires_ the receiver to
understand. Receiver may then ignore the rest. (In my examples, the docs
conform to the versions listed in the examples, but sender could very well
require a language capability which describes only a small part of a doc.)
Conformance IMO is not the main point, it is the question whether a receiver
may safely process a doc which receiver does not fully understand. 

Note that I'm speaking from a background of digital exchanges (healthcare
and justice amongst others) which are embedded in a legal framework, and
processing an incoming document is not without obligations. (This situation
is very different from HTML, where 'sender' often has little or no control
over what 'receivers' do with docs.) I appreciate the forward compatibility
mechanism proposed in the Versioning Finding, but believe there need to be
fine-grained mechanisms to overrule 'IgnoreUnknown' behaviour for such
digital exchanges, and I tried to explore one principle for it: list all
capabilities required in the instance. Since receveivers will know all
capabilities implemented in the receiving processor, most situations can be
resolved, whether receiver of sender is the later version.

Regards,

Marc de Graauw

http://www.marcdegraauw.com

Received on Monday, 23 April 2007 20:02:14 UTC