Re: And the DTD says, "I'm NOT dead yet!!"

[John Cowan <jcowan@reutershealth.com>]

Bullard, Claude L (Len) wrote:

> Umm... the formal specification is 
> informative?  That seems backward.  
> Can you or anyone who cares to explain 
> why that is the case since informative 
> descriptions are typically non-binding? 

For a mixture of reasons theoretical and practical, I think.
Formal descriptions are harder for many people to read, so they
get fewer eyeballs.  The informative RDF Schema description
in the Infoset was (IIRC) only reviewed by two people within the WG
other than myself (who wrote it), whereas the normative prose
was reviewed by a great many people.

Furthermore, the formalism is rarely able to do it all;
there tends to be some normative prose somewhere, unless the
formalism is Turing-complete like the van Wijngaarden grammar
used in the Algol 68 Revised Report.


> When doing a validation, I need to use 
> the ROA that is binding.  Perhaps it 
> is a legal tangle where in one process 
> for applying the record (reading an ROA 
> to determine rules for the implementation) 
> the prose is binding,

Yes.

> but in another process, 
> (determining if a transaction content 
> conforms), the informative description 
> becomes normative for the transaction. 

Not really.  If the formal description disagrees
with the prose, it must be changed to agree with
the prose.


-- 
There is / one art             || John Cowan <jcowan@reutershealth.com>
no more / no less              || http://www.reutershealth.com
to do / all things             || http://www.ccil.org/~cowan
with art- / lessness           \\ -- Piet Hein