Introduction by Frank. The core of the issue is that there is nothing to distinguish the starting point (request or response element) of processing in the body of a message. He also notes the problem exists in headers as well as the body. The solutions proposed have included define element order, somehow use WSDL, use the "root" described in (a poorly worded) sec 5.6 of the spec. Various practices are used in today's implementations, and part of the challenge of issue 78 is to accommodate current practices. Such consideration has led to a proposed solution using the "root" attribute to indicate the starting point element, and to assume the request element is the first element in the body if "root" is not present. The order criteria is based on the perception that it represents the most common ordering practice.
Part of the resolution includes a proposed rewrite of sec 5.6 (see June dist-app message #0110) and sec 7.1 (see July dist-app message #0262, "v5"). The latter also proposes a number of Glossary terms.
Jacek takes issue with a "serialisation root" in the proposal, which he believes is the last tie between sections 5 and 7 (and the RPCTF has been moving toward removing ties between secs. 5 & 7).
Jacek also takes issue with the placement of the root attribute on child elements of <body> because it requires a scan of all child elements to determine which is the starting point, or to determine that the root is not present.
Discussion followed toward the end of which, the beginning of a proposal emerged:
It was also discussed that such an attribute might be useful in situations other than RPC, in which case it would be more appropriately be part of the "core spec". Furthermore, it was discussed that such an attribute might be equally applicable to headers as well as the body. This raised concerns about the distinction between the body (just another header?) and other headers. These further points of discussion were to be taken to dist-app, perhaps as further discussion points for the WG telcon when discussing issue 78/16.
Action: Frank to send an email today outlining the new proposal
Action: David to add this proposal to the WG's telcon agenda