W3C home > Mailing lists > Public > www-ws-desc@w3.org > October 2004

Re: Minutes, 14 Oct 2004 WS Description WG

From: Hugo Haas <hugo@w3.org>
Date: Mon, 18 Oct 2004 12:35:00 +0200
To: Jonathan Marsh <jmarsh@microsoft.com>
Cc: www-ws-desc@w3.org, Arthur Ryman <ryman@ca.ibm.com>
Message-ID: <20041018103500.GH11709@w3.org>
* Jonathan Marsh <jmarsh@microsoft.com> [2004-10-14 13:30-0700]
> 13. Issue LC21: Multipart Style and {direction}=out [.1]
> 
> [.1] http://www.w3.org/2002/ws/desc/4/lc-issues/#LC21
> 	
> 
> 		Discussion:   
>                 
>                 Asir: Multipart style direction currently restricted to
> IN, Why...??
> 		      Proposal to resolve included with the comment.
>                       (remove direction restriction language phrase: see
> comment text [.1])
> 
> 		JMarsh: Anybody remember why we restricted to IN
> directions only.
> 
> 		Arthur/Asir: Maybe just a simple editorial bug in the
> spec?
> 
> 		JMarsh: Restriction based on the HTTP 1.1 binding
> restrictions?
> 
> 		Discussion about need/use multipart style for OUT
> messages
>                 (file upload, query parameters, image returns etc.)and
> if this
>                 restriction is needed.
> 
> 		Asir/Hugh: Style constraints may not be granular
> enough..
> 
> 		Asir: Part 1:  2.4.1.1 states the style applies to all.
> 
> 		Dbooth: That may be a general statement and specific
> styles may put
> 		        further restrictions....
> 
> 		Asir: If that's true...need to add a statment like:  The
> Style property
>                       may constrain both input and output, however a
> particular style may
>                       constrain only one direction. ...Agreed....Add Ed
> TODO for Part 1...
> 
> 		Further discussion of the use for output direction......
> 
> 		JMarsh: As a WSDL consumer, is knowing this limitation
> going to make a
>                         difference...??
> 
> 		Hugo: Helps to know what binding to use for a consumer
> or to
>                       "prepare" or design to receive this ....
> 
> 		Jmarsh: Is there a need for independent style
> restrictions (IN and OUT)?
> 
> 		Arthur/Asir...yes.
> 
> 		Jmarsh: Suggest either leave style at the operation
> level and create a
>                         Multi-Part IN and Out Style (can have a list of
> styles that produce
>                         a Union of constraints).....or move style to the
> message reference
>                         level (large change/risk??)
> 
> 		Arthur: Need to make sure the style constraints don't
> (aren't allowed to)
>                         result in a null set.
> 
> 		JMarsh: Suggest separating Multipart style to both IN
> and OUT...
> 
> 		Hugo: Might the "better" (not necessarily easier) to
> apply the style
>                       property to the message reference level instead of
> patching this
>                       specific style?  Is this same issue going to apply
> to other style
>                       constraint definitions.??
> 
> 		Suggests the ability to put style constraints on BOTH
> operation level and
>                 message reference level..
> 
> 		JMarsh: Is this really necessary? e.g. URI style on
> OUT??
> 
> 		Arthur: Are we confusing IN/OUT vs Request/Response
> these aren't
>                         necessarily equal?
> 
> 		JMarsh: Summary of Hugo's proposal: Push style to the
> message reference
>                         level for granularity and have "global" default
> at the operation level.
> 
> 		Dbooth: Do we really need to have style properties at
> the operation level
>                         (reduces the default
> scoping/composition/override problem)?
> 
> 		Discussion ....resulting in the dropping of the
> Multipart IN and Multipart OUT option..
> 
> 		Asir: Request review of options we are considering.
> 
> 		JMarsh:Options for Straw Poll.
> 
> 			1. Keep style properties on both components.
> Add text to merge them when
>                            a processor interprets the component model.
> 
> 			2. Syntactic default.  Move style properties
> from operation to message
>                            component, change the name to styleDefault,
> single style property on
>                            message reference component.
> 
> 			3. Move style property completely to the message
> level with no
>                            defaulting.
> 
> 		J-Jaq Question: Can faults have style properties??
> Currently NO...
> 
> 		Straw Results: Option 1:1 vote, Option 2:4 votes, Option
> 3:3 votes.
> 		No objection to recording concensus on Option 2.
> 
> 
> 		Resolution: See Straw poll results above.

Just to make sure that we don't lose it, Arthur raised during this
discussion that the HTTP binding should probably also support Out-In
MEPs. I don't think that we resolved this during this discussion.

Regards,

Hugo

-- 
Hugo Haas - W3C
mailto:hugo@w3.org - http://www.w3.org/People/Hugo/

Received on Monday, 18 October 2004 10:35:01 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:33 GMT