Minutes of RPC Task Force telcon, 8-9.30a PDT, July 20, 2001



Identify and discuss issues

Action: David will arrange another call on Wednesday July 25, details to follow in separate email.

Action: Jacek will identify RPC related issues from existing issue list, by COB Monday July 23.

-- Jacek, Frank and Chris expressed concern about relation between RPC and encoding. Jacek thinks the relation is underspecified. Frank believes there is ambiguity in the spec as to whether or not section 7 requires section 5.

Action: Frank will identify a list of section 5/7 dependencies, a preliminary list will be ready by the Wednesday TF telcon.

Frank, Jacek and Chris also expressed concern over the modelling of RPC as structs (another case of dependency between sections 5 and 7), perhaps an Infoset description would be a way of describing RPC in a more neutral manner?

Action: Frank to send email to Martin Gudgin to ask his opinion on the utility of an Infoset description for section 7.

David noted that issues 16 and 78 on the issues list pertain to void return types and the ordering of structs. It was suggested there was a fair amount of agreement on how to handle void return types. Frank has already written up proposed resolutions for 16 and 78 (for the Dinard f2f). These will be discussed on the Wednesday TF telcon, and the TF will have the option of deciding whether to recommend these resolutions to the larger WG, possibly to be decided on next wednesday's WG telcon.

Action: Frank to send David pointers to issue 16 and 78 resolution texts.

Action: David to put these texts as proposed resolutions into the issues list.

Frank noted that people have asked for boxcarring (in reference to section 7) although section 1.1 states this is not part of the design. He suggests we may want to add wording to section 7 to further say that boxcarring is out of scope.

Chris brought up the issue of providing correlation when the underlying transport does not provide any correlation (or that feature of the transport is not taken advantage of). It was noted that this is issue 44. The TF discussed the implications and scope of a correlation extension. For example, what happens when a correlation id is provided in the message itself, and is also provided by the transport -- should there be the notion of precedence? It was decided to proceed by evaluating a proposal for a correlation extension.

Action: Chris to draft a correlation extension, to be posted by Tuesday July 25 in time for TF members to read before the Wednesday TF telcon.

Jacek wants to differentiate between instances of objects on a server, for example, as a means to provide statefulness. It was pointed out that section 1.1 explicitly precludes object references in "core" SOAP, and so the TF thinks it is inappropriate to spend time designing any sort of extension mechanism to provide such a facility.

Frank expressed concern over the clarity of wording and ambiguities in section 7.2. For example, he finds the ability to reference headers (which may be changed by intermediaries) counter to a basic notion of RPC.

Action: Frank to send an email to the TF with a pointer to a summary email on this topic.

Jacek questions whether RPC is an application or an extension.

Action: Jacek will send an email describing this issue.

Valid CSS! Valid HTML 4.0!
Revision: $Date: 2001/08/02 09:42:24 $