- From: Henrik Frystyk Nielsen <henrikn@microsoft.com>
- Date: Mon, 6 May 2002 15:16:08 -0700
- To: <noah_mendelsohn@us.ibm.com>, "Martin Gudgin" <martin.gudgin@btconnect.com>
- Cc: <xml-dist-app@w3.org>
I think the only need for introducing the notion of a root in the data model is to describe that whenever you have cross-references between serialization trees, you have a single graph. That is, I think the example below is valid, the response is unambiguous and it is part of a single graph along with the header block. Henrik ><s:header> > <x:e ref="a" s:encodingStyle="..soap encoding..."/> ></s:header> ><s:body> > <y:getStockPrice id="a" s:encodingStyle="..soap encoding..."> > <y:company>GizmoWare</y:company> > </y:getStockPrice> > ></s:header> > >getStockPrice is a struct with an inbound edge, hence not a >root. Yet, >it's still a good RPC, right? > >Bottom line: if we need and use a notion or root, then I >agree it belongs >in the data model, and I think you have the details right. >I'm not 100% >convinced we need it.
Received on Monday, 6 May 2002 18:20:25 UTC