- From: David Booth <dbooth@w3.org>
- Date: Tue, 13 Jul 2004 17:50:00 -0400
- To: "Martin Gudgin" <mgudgin@microsoft.com>, "Sanjiva Weerawarana" <sanjiva@watson.ibm.com>, "WS Description List" <www-ws-desc@w3.org>
At 08:44 AM 7/13/2004 -0700, Martin Gudgin wrote: >. . . I'm happy with anything that is OPTIONAL, I'm just not happy mandating >that the service expose internal implementation details... . . Are they REALLY internal implementation details? If so, then I see no problem. But if the requester agent needs to follow the same convention C as the provider agent in order properly use the service, then those implementation details emphatically are NOT "internal". The key question is whether convention C is engaged with the provider entity's explicit knowledge and intent, such that the provider entity is likely to document its use in the application semantics documentation. If convention C is enabled automatically or transparently by the provider entity's toolkit, then it is unlikely that the application developer will know to document the need for it in the application semantics documentation, thus leading to the interop issue described in Scenario X. Furthermore, conveying such implementation details (i.e., details that both parties need to know about) in a machine-processable form is one of the main purposes of WSDL. Even if such information *does* appear in the application semantics documentation, it would be NICE if the toolkit would indicate the use of such details in the WSD, so that other toolkits could notice it and automatically engage it for requester agents, rather than requiring the requester entity to perform manual steps to engage it. References 1. Scenario X: http://lists.w3.org/Archives/Public/www-ws-desc/2004Jun/0300.html or: http://tinyurl.com/4krve -- David Booth W3C Fellow / Hewlett-Packard Telephone: +1.617.253.1273
Received on Tuesday, 13 July 2004 17:50:04 UTC