- From: Glen Daniels <gdaniels@sonicsoftware.com>
- Date: Thu, 14 Apr 2005 11:55:06 -0400
- To: <www-ws-desc@w3.org>
In fulfillment of my long-standing AI: ------------- 0. Introduction There are currently two formal objections against the WSDL 2.0 specs in the area of Features and Properties. One, by Microsoft, IBM and SAP [1], objects to the inclusion of Features and Properties markup in the WSDL specification. The other, by Oracle, Sonic, Sun, and IONA [2], objects to the unreasonably quick dismissal of the "compositors" proposal. This note suggests a compromise which might enable both of these objections to be resolved by giving up something from each "camp". 1. The Compromise in Summary The tradeoff would be as follows: 1) The formal objection to F&P gets resolved by F&P turning into an optional extension. Thus, the objectors would not need to implement it, and yet the F&P supporters would still have an available non-proprietary form of capabilities metadata in the W3C specs. 2) The formal objection regarding compositors gets resolved by allowing a limited (<and>, <choice>) version of the compositors proposal to move forward in the WG. Thus, the objectors would be allowed a chance to finish their work in a more tightly scoped fashion, and to implement it in the F&P section of the spec - which as per the above would be optional and therefore less uncomfortable for those who voted it down. 2. Details 2.1. Optional Features and Properties Several companies do not feel comfortable with the inclusion of Features and Properties in WSDL 2.0. The primary reason for this seems to be that it overlaps the extant work in non-W3C specs such as WS-Policy [3]. On the other hand, many companies don't feel comfortable using specs like WS-Policy which, although certainly well done, have not yet been vetted by a standards process, and are subject to both control issues and potential IP issues. Some companies also feel it is important for WSDL to directly support the SOAP 1.2 extensibility framework, and more importantly that WSDL 2.0 contain some "built in" (even if optional) mechanism to support the expression of capabilities and requirements metadata. The working group has voted on the inclusion of the F&P mechanism several times, and each time there was clear support for leaving it in. A solution to this impasse could involve refactoring the F&P work as a normative extension to the WSDL core. This would still give the F&P proponents a W3C-supplied technology, aligned with the SOAP 1.2 extensibility model, for expressing constraints/capabilities - but it would not require the objectors to implement it in order to be "WSDL 2.0 conformant". It is true that this might cause interoperability issues, but no more so than any other optional extension within the W3C (the HTTP binding for WSDL, or even SOAP) or without (WS-Policy, etc). This work would involve: 1) Introducing a new section to the "Adjuncts" document, and moving all description of <feature> and <property> markup into that section. 2.2. Limited Compositors The authors of the original compositors proposal felt that the decision to shut down work in this area was inappropriate in light of a) some of the political overtones, and b) the fact that although they were more than willing to produce a revised version of the proposal, there was no opportunity to do so before a vote was forced. Therefore the authors would like to see the group reconsider a much more narrowly scoped version of the proposal. Although this does not represent a fleshed-out proposal, and also does not represent the opinions of the companies involved, my guess for this more limited work would involve: 1) Introducing <and> and <choice> elements 2) These elements would be placed around markup such as <property>, <soap:module>, and <feature>, and may be nested 3) <and> would entail engaging all the immediate child elements 4) <choice> would entail selecting one of the immediate child elements to engage This would enable WSDL authors to express combinations of valid settings which "cross-cut" multiple extensions, enabling greater control over composibility than can be achieved at present with the property constraint syntax. This functionality would also live in the F&P extension described above, and would therefore not be required. Clearly a detailed proposal for this would need to be written up, but the intent of this note is to gauge support for the compromise idea before that work occurs. 3. Conclusion The two formal objections in the area of F&P are currently on track to impede WSDL 2.0's progress through the W3C process. They will need to be reported to the director and will certainly cause delays. If the group can reach a consensus on these issues such that they can be withdrawn, it will be better for both the WG itelf and the W3C as a whole. If both "camps" are amenable to the solution outlined in this note, or something like it, this would be a good road forward. Note that either of these solutions COULD also be implemented individually, but since there are two distinct "camps" across these two related objections, it is the author's opinion that this represents an opportunity to do some "horse trading" in order to resolve both at the same time. [1] http://lists.w3.org/Archives/Public/www-ws-desc/2004Jul/0375.html [2] http://lists.w3.org/Archives/Public/www-ws-desc/2004Jul/0371.html [3] http://msdn.microsoft.com/library/en-us/dnglobspec/html/ws-policy.asp ------------- Thanks, --Glen
Received on Thursday, 14 April 2005 15:55:10 UTC