Re: ETF: Issue 47 (SOAP Data Model)

 Marc, please see my replies inside.



On Tue, 11 Dec 2001, Marc Hadley wrote:

 > >  Your text keeps the mention about so called multistructures
 > > which, AFAIK, are not described elsewhere in the spec and no
 > > implementation supports them. Multistructures can be easily
 > > viewed as structures with array members (the members that are
 > > multi) and I think they should be serialized as such for the sake
 > > of simplicity. So I'd like us to remove multistructures from the
 > > data model.
 >
 > My text was written before the F2F so doesn't reflect any
 > decisions made there. However, it looks to me like it is
 > still consistent with the text in the latest editors copy.
 > Do we need to open a new issue to address your concern or
 > is it covered by one of the outstanding ETF issues ?

It has been already raised (finally) in my email [1] so I don't
think you need to do anything more on this. 8-)

 > >  Generally, I'm not sure the data model should speak about
 > > locally and universally scoped names,
 > You would prefer item 4 to remain in the encoding section then ?

Yes, I would.

 > > I'm also not sure we can
 > > actually have any single-reference values if you define
 > > multi-reference values as such that _can_ be referenced from
 > > multiple locations.
 >
 > Not sure, it's certainly legal in the encoding to have a
 > "non-inline"  element referenced only once within the
 > serialisation tree. Also, any element with an id attribute
 > _can_ be referenced from multiple locations. Perhaps we
 > need to drop the multi/single reference terminology and
 > just go with inline/non-inline ?

Your wording was "if only one accessor can reference it, an
instance is considered 'single-reference'". What I meant was that
in the data model there is IMHO no instance that _can_ only be
referenced by a single accessor. I see all data as potentially
multi-reference.

To keep your text, I think you need to change it from "can
reference it" to "references it". With such a change, the text
would make sense.

 > >  Finally, I'd certainly not mention the RPC section from the
 > > data-model section as RPC seems just too high above the data
 > > model.
 > >  Also we might want to add a paragraph or two about when this
 > > data model is used (actually here a pointer to RPC might be
 > > appropriate)
 > The mention of the RPC section in the proposed text was
 > intended to fulfill your latter point - do I take it that
 > you withdraw your initial objection or is some rewording
 > necessary ?

I think the last sentence needs to be reworded, for example
like:

"See section 5 (Using SOAP for RPC) for an example of use of this
data model."

to emphasize that other applications can use this data model and
to clarify the non-normativeness of this reference. I wouldn't
change the preceding sentence in this way, though, to avoid the
impression that people will be welcome if they create forty-two
new encodings for this single data model.

Best regards,
Jacek



[1] http://lists.w3.org/Archives/Public/xml-dist-app/2001Dec/0128.html

Received on Wednesday, 12 December 2001 10:28:54 UTC