Draft for review: ISO 8601 date work on IS 29500

MURATA Makoto (FAMILY Given) eb2m-mrt at asahi-net.or.jp
Sat Jul 24 11:29:23 CEST 2010

Dear colleagues, 

How about this approach, which is very common in standardization?

We introduce a reference model.  It is a collection of procedures.  It 
is introduced only for precisely defining the semantics of data.  It is
not meant to tie the hands of implementors.

Implementations are free to implement something very different from the 
reference model.  However, the final result is required to be the same 
as what the reference model provides.

An example of reference models is in the RELAX NG  specification.  In 
this reference model, schemas are always represented as XML documents 
and every normalization step is an XML transformation.  In
implementations, schemas are simply converted to internal data
structures from the beginning.  Even the processing order is different.  

In our case, I think that the initial conversion to serial values is a
step in the reference model and that implementations do not have to do
this step as long as results are identical.


More information about the sc34wg4 mailing list