Re: Proposal for issue 277 - part 2

Marc Hadley wrote:
> On Monday, Oct 14, 2002, at 11:03 US/Eastern, Herve Ruellan wrote:
> 
>>
>> (v) Remove http://www.w3.org/2002/06/soap/mep/ namespace and 
>> http://www.w3.org/2002/06/soap/mep/FailureReasons/ namespace.
>> <rationale>
>> They are not used anymore.
>> </rationale>
>>
> They aren't used as XML namespaces as such, but they are used as 
> absolute URIs within the spec and, rightly or wrongly, we use a notation 
> of absolute:relative (just like prefix:localname) when referring to 
> properties. I think we need something like this otherwise the spec will 
> just be full of long URIs that will reduce readability.

Your right. My proposition was probably not clear enough. My intent is 
to remove the use of http://www.w3.org/2002/06/soap/mep/ and 
http://www.w3.org/2002/06/soap/mep/FailureReasons/ as namespace, but to 
keep them as URIs, and to keep their usage as base URIs for some 
property values.

 From my reading of the spec, the absolute:relative notation still 
refers to QNames, as it is mainly used for property names and that 
properties are still named with QNames (see 5.1.1). But we can change 
that (this is item (vi) in the proposal).

 From a practical viewpoint, the intent of my proposal is to remove both 
those URIs from Table 1 in part 2.
However, I just did a check in part 2, and those URIs are only used in 
Table 1. So we can fully drop those URIs.

> 
>> (vi) Define properties using URIs and not QNames (see first part of 
>> proposal), and remove 
>> http://www.w3.org/2002/06/soap/bindingFramework/ExchangeContext/ 
>> namespace, http://www.w3.org/2002/06/soap/mep/request-response/ 
>> namespace and http://www.w3.org/2002/06/soap/features/web-method/ 
>> namespace which are only used for defining properties.
>> <rationale>
>> It seems better to identify properties with URIs than with QNames.
>> </rationale>
>>
> Same comment as above.

For those URIs, my propsal is to not use them anymore as namespaces 
(removing them from Table 1), but to use them as base URIs for defining 
property names.
We could probably use a notation such as prefix/relative to make the 
spec less verbose. I we use such a notation, I would be nice to have all 
the URIs involved in this notation summarized in a table, but I think it 
would be better to have a table distinct from Table 1.

Regards,

Hervé.

Received on Thursday, 17 October 2002 08:01:49 UTC