[
Lists Home |
Date Index |
Thread Index
]
John Cowan wrote:
> Bob Wyman scripsit:
>>It would have been wise if the WXS designers had noted existing
>>standards and used "SET" rather than "SEQUENCE" when defining WXS...
>
> There's a reason behind that: the schema-centric viewpoint on this point
> is opposed to the data-centric viewpoint. In general, if a schema says
> that the order is prescribed (a sequence), that means that there is no
> significance in the order. On the other hand, if the schema does not
> prescribe an order, then there is typically significance in the ordering
> that actually appears in the data. WXS takes a schema-centric viewpoint;
> ASN.1 takes a data-centric one.
As was pointed out earlier, the "schema-centric" view is missing the
ability to say, for elements, that no order is prescribed and order is
not significant. Same semantics as attributes.
OTOH, if a schema prescribes an order (a sequence) it is moot that there
is no significance in the order, as the prescribed order must be observed.
Further, the schema-centric viewpoint you describe is definitely not the
common usage for iterated choices, e.g., (a|b|...)*. In this case, no
order (or cardinality) are prescribed and the semantics is usually that
order is not significant.
Bob Foster
|
