- From: Williams, Stuart <skw@hplb.hpl.hp.com>
- Date: Sun, 17 Mar 2002 09:05:32 -0000
- To: "'noah_mendelsohn@us.ibm.com'" <noah_mendelsohn@us.ibm.com>, "Williams, Stuart" <skw@hplb.hpl.hp.com>
- Cc: Martin Gudgin <marting@develop.com>, "'Jean-Jacques Moreau'" <moreau@crf.canon.fr>, xml-dist-app@w3.org
> -----Original Message----- > From: noah_mendelsohn@us.ibm.com [mailto:noah_mendelsohn@us.ibm.com] > Sent: 15 March 2002 19:51 > To: Williams, Stuart > Subject: RE: Rework on SOAP 1.2 Part 2: Section 2 and 3 > > > Stuart Williams writes: > > >> Firstly, i think that from the discussion it has become > >> clear to me (I think) that in this data model nodes are > >> *not* labelled, only edges, which I think is what I missed. > > Exactly. Not sure how Gudge would feel about this, but I > could easily live with: > > <orginal chapter 2> > The SOAP data model represents information as a directed, edge-labeled > graph of nodes. > </orginal chapter 2> > > <proposal> > The SOAP data model represents information as a directed graph of nodes > and edges. Edges in the graph carry optional labels; nodes are > unlabeled. > </proposal> I could mostly live with either - in fact I think the term "edge-label graph" pretty much captures it and the clarification that nodes are unlabelled would avoid others making the mistake I made. > >> Is it the case that terminals are all (possibly typed) > >> literals (I think). > > Yup. In an interleaved thread there is discussion of the need to make > clear that non-terminals never have a lexical value. Terminals have > lexical values, except for the case modeled in the serialization of > xsi:nil. > > >> I think what would be really clear would be > >> to say that edges are represented by element > >> names and that nodes are represented by element > >> content, non-terminals have only elements as > >> content and that terminals have typed literals > >> as content. (of course turned into infoset'ise). > >> Maybe that is indeed what it says... I'll read it again. > > The part about element names representing edges is always true. The part > about content representing nodes doesn't seem quite right, which is why > I'm not sure I agree with your suggestion. First of all, we can't use > "element content", as that doesn't catch attribute children. xsi:type and > xsi:nil also represent the node, as opposed to the edge. So in 3.1 case 1 > we would have to say something like: Hmmm interesting... do we have three things to represent here? Node identity, type and value? > <original> > If the element information item representing the edge does not have a ref > attribute information item ( see 3.1.4.2 ref Attribute Information Item ) > amongst its attributes then the element information item is said to > represent a node in the graph and the edge terminates at that node. > </original> > > <possibleRevisionImNotSureILike> > If the element information item representing the edge does not have a ref > attribute information item ( see 3.1.4.2 ref Attribute Information Item ) > amongst its attributes then, the children of the element information item > represent a node in the graph, and the edge terminates at that node. > </possibleRevisionImNotSureILike> > > I think it's a bit more precise, but I'm not sure that readers will find > it more effective in conveying what we intend. > > The point you raise that I think we do need to do something > about is to clarify that: > > <p:looseEdge soapEnv:encodingStyle="...encoding URI xlmns:p="pUri"> > <p:a>123</p:a> > <p:b>456</p:b> > </p:looseEdge> > > > Results in an edge with no source: > +------------------+ > |---------------------| pUri,a | NodeType: Simple | > | |----------| Value: "123" | > | | | | > | | +------------------+ > pUri,loosedge | | > --------------| Nodetype: struct | > | | +------------------| > | | pUri,b | NodeType: Simple | > | |---------| Value: "456" | > | | | | > | | +------------------ > ---------------------- > > I think we do need a clarification on that. Gudge and I are both aware > that serialization roots also aren't discussed in this version, and I > suspect loose edges should be covered at the same time. I think, if you buy the view that nodes are unlabelled, SOAP Encoding starts at 'loose edge' rather than a node. > ------------------------------------------------------------------ > Noah Mendelsohn Voice: 1-617-693-4036 > IBM Corporation Fax: 1-617-693-8676 > One Rogers Street > Cambridge, MA 02142 > ------------------------------------------------------------------ Thanks, Stuart
Received on Sunday, 17 March 2002 04:06:17 UTC