Re: DR 201: straightforward mapping, or Schema?

> IMO, and as mentionned by Stuart in [1], we should not consider
> language bindings (out of scope) in this requirement.

It is out of scope to create all the language bindings.  It is not out 
of scope, IMO, to require specification at some level of how testable, 
interoperable language bindings will be achieved.  If we are not willing 
to do this, I think we should postpone much of the spec.  Making a wire 
protocol that is not designed to bind interoperably would be a mistake, IMO.

> 
> Instead I propose that the last sentence of this requirement:
>               There will be straightforward mappings of the data types
> 
>               used to a wide variety of widely deployed programming
>               languages and object systems
> is either dropped, or replaced by something like:
>               XML Schema datatypes will be used to represent data
> types
>               most commonly used in popular programming languages
>               and object systems.

If this were really the case, then it would seem to me that most of the 
200's and 400's could be dropped generally, since XML messages in 
general are described well by schemas, and we don't need any special 
rules for that.  However, the requirements in the 400's for encoding 
nestable arrays, graphs, and "data based on datamodels not directly 
representable by XML Schema" leads me to believe that we are 
supplementing XML Schema with a standard XP encoding precisely because 
XML Schema datatypes do not directly represent certain models and types 
common in programming languages.

Ray Whitmer
rayw@netscape.com

Received on Wednesday, 15 November 2000 18:47:19 UTC