- From: <paul.downey@bt.com>
- Date: Thu, 10 Jun 2004 13:15:45 +0100
- To: <jean-jacques.moreau@crf.canon.fr>, <mark.nottingham@bea.com>
- Cc: <www-ws-desc@w3.org>
my (probably confused) thinking: 1) this puts 'style' firmly into the realms of an optional extension. someone not interested in a given namespace can more easily ignore it. we can't mark extension attributes as 'required', no? This demotion of style is a good or bad thing depending on your POV - i think it's a good thing. 2) it actively encourages other style related extensions to be provided as other attributes in the 'foo' namespace, e.g.: <operation foo:rpc="true" foo:debug="true" foo:camelStyleNames="true" foo:useInOutParameters="true" .... this is again, good or bad depending on your POV. - i think good! 3) it allows for multiple styles to be easily attached to an operation, as in Mark's foobar example. - good in anyone's book, surely. So that's a big +1! Paul -----Original Message----- From: www-ws-desc-request@w3.org [mailto:www-ws-desc-request@w3.org]On Behalf Of Jean-Jacques Moreau Sent: 10 June 2004 08:27 To: Mark Nottingham Cc: www-ws-desc@w3.org Subject: Re: Issue 217: Syntax for multiple styles Talking about style, I guess you really meant foo:rpc="true"? ;-) Yes, let's separate the semantics (and have XML do the parsing for us). +1. (I guess it's all backwards.) JJ. Mark Nottingham wrote: > > in his proposal for this issue [1], Jonathan suggests that it has > already been addressed by issue 98 [2]. > > The resolution to 98 does address the bulk of the concern I had here. > > However, from a stylistic standpoint (no pun intended), I would > prefer that such things be flagged with separate attributes, e.g., > instead of > <operation style="http://some/uri/that/says/this/is/RPC > http://some/other/uri/that/says/this/is/PUT"> > something like > <operation foo:rpc="1" bar:webMethod="PUT"> > seems preferable. However, this isn't critically important, and if > the WG prefers a URI, so be it. > > I would note that the resolution to issue 98 hasn't yet been > incorporated into the draft (apologies if this is known to the > editors; just want to make sure it doesn't get lost). > > 1. http://dev.w3.org/cvsweb/%7Echeckout%7E/2002/ws/desc/issues/wsd- > issues.html#x217 > 2. http://dev.w3.org/cvsweb/%7Echeckout%7E/2002/ws/desc/issues/wsd- > issues.html#x98 > > > P.S. Stepping back for a moment, I notice something curious in > relation to issue 221, regarding QNames vs. URIs. WSDL has chosen to > use QNames as the primary means of identifying components (for which > many use cases include references from outside the document), while > choosing URIs to identify operation styles, a mechanism with a purely > local semantic. > > It seems to me that this is backwards; URIs are more useful for > things that might be referenced on the greater Web, whilst QNames are > safer and more useful in a specialised, controlled contexts. > > But that's a discussion for another thread, perhaps. > > -- > Mark Nottingham Principal Technologist > Office of the CTO BEA Systems >
Received on Thursday, 10 June 2004 08:15:54 UTC