W3C home > Mailing lists > Public > www-ws-desc@w3.org > March 2002

Re: Reqs DR033/DR034

From: FABLET Youenn <fablet@crf.canon.fr>
Date: Thu, 07 Mar 2002 14:45:41 +0100
Message-ID: <3C876F05.9050907@crf.canon.fr>
To: Jonathan Marsh <jmarsh@microsoft.com>, www-ws-desc@w3.org
CC: moreau@crf.canon.fr
  Here are some suggestions that may help:

It must be possible to specify what type of description (types, 
types+messages, ...) is contained in an interface without interpreting 
the whole interface

WSDL must describe rules that allow or disallow the import of a WSDL 
description in another description (depending for instance on the "type" 
of the interface)

WSDL must allow that any element in a WSDL description can be referenced 
and extended by another element of the same type (i.e. a portType can be 
built upon another portType, idem for types...).

WSDL must describe rules that specify how the import of a WSDL 
description in another description will work (one may say: when 
importing A in B, just paste the desc of A in B to have the final desc 
C. Personally, I may prefer (but I am not sure of all the implications) 
to have a finer import model, which let for instance overwrite/extend in 
B elements already declared in A).

I agree that the last two reqs might be hard to reach if we choose to 
have results quickly (c.f. WS-I). But I think that they are interesting...


Jonathan Marsh wrote:

>Can you suggest more precise text?
>-----Original Message-----
>From: FABLET Youenn [mailto:fablet@crf.canon.fr] 
>Sent: Wednesday, March 06, 2002 4:58 AM
>To: www-ws-desc@w3.org
>Cc: Jean-Jacques Moreau
>Subject: Reqs DR033/DR034
>DR033: Support abstract interfaces
>DR034: Support interfaces derived from abstract interfaces
>Ideas behind these reqs.
>Reqs DR033 and DR034 are related to the import functionnality in WSDL1.1
>(which could somehow be used as a derivation mean).
>WSDL1.1 already supports abstract interfaces and derivation of 
>interfaces, but a more precise model should be described.
>I think that the import functionnality is not well explained (what if 
>you redeclare the same portType in file A and file B, and file A is 
>imported in file B 
>for instance (is this example covered by the "Names 
>within a name scope MUST be unique within the WSDL document" in WSDL1.1 
>paragraph2.1.1?)). It might also become hard to maintain and understand 
>WSDL files if all imports are possible.
>It can be interesting to be able to qualify the level of abstraction of 
>an interface, like saying "this interface contains only types, this 
>interface specifies only types and messages....". An interface that 
>declares itself as containing only types and messages should for 
>instance not be able to import anything else than types and messages 
>This could allow good programming rules to be enforced.
>I think that we need to fix and/or clarify the use of the import 
>functionnality. If we do so, we could take the opportunity to go a step 
>further and add some OOP functionalities, like extending a portType (is 
>it DR042?) by derivation (using or not the import functionnality) for 
>ance. I do not think that it will cause a great overhead to do so.
>I do not think that DR109 is sufficient to describe these ideas, we 
>should maybe make DR033/DR034 more precise.
>  Youenn
Received on Thursday, 7 March 2002 08:47:30 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:54:37 UTC