Re: [soapbuilders] Proposal for naming the SOAP RPC return value

Jacek Kopecky wrote:
> 
>  Hello all. 8-)
>  The RPC Task Force has considered the issues #16 [1] and #113
> [2] and the feedback to the message [3] in the soapbuilders' list
> (summarized in [4]) and proposes the following rewording of the
> appropriate three bullets in SOAP version 1.2 section 7.1 [5]
> (editors' draft) and adding the following examples and the ending
> note.
> 
>  (...)
>
> The name of the return value accessor is "result" and it is
> namespace-qualified with the namespace identifier
> "http://www.w3.org/2001/06/rpc". The name of the struct is not
> significant. However, a convention is to name it after the
> procedure or method name with the string "Response" appended.

Does any existing Sec 5 implementation with
direct mapping to native types support
universally scoped members of componound values?
Wouldn't a simple convention "the name of the 
result element should be 'result' (unqualified)" 
sufficient? Let's keep the promise of the 'S'
in SOAP and keep things simple. Although 
Sec 5 explicitly talks about globally scoped parts of 
complex values, I have never seen one, probably 
because most existing programming languages
just do not support this kind of things. 
Promoting usage of exotic features 
will probably raise new compatibility issues.
At least for kSOAP, supporting xyz:result
would mean a lot of extra effort. Since
kSOAP is supposed to be used in mobile devices
this really scares me.

I suggest to apply Occams razor and
to remove the universally scoped accessors
from componound types completely in 1.2.

Best,
Stefan


-- 
Stefan Haustein             
Univ. Dortmund, FB 4, LS 8   tel: +49 231 755 2499
Baroper Str. 301             fax: +49 231 755 5105
D-44221 Dortmund (Germany)   www-ai.cs.uni-dortmund.de

Received on Monday, 20 August 2001 11:35:15 UTC