Re: Proposed text for issues rec20 and rec22 - Part 2

Michael:  I think this is an excellent start, but I wonder if we don't 
need to give some more attention to the data model, encoding and RPC 
sections. 

======================
For example, 2.1.1 says[1]: 

"An edge label is an XML qualified name."

Can the new names appear as edge labels?  If so, what are the 
serialization rules if using an XML 1.0 infoset and/or serialization.  We 
know that some SOAP nodes can't handle 1.1 at all.  Is there a separate 
encoding-level error that represents failure to reconstruct a data model 
with 1.1 edge names?  The easiest way out of this is to say that even if 
SOAP part 1 allows for XML 1.1 or later, the data model does not.   I'm 
not sure whether that's acceptable, but I think we at least need to 
consider the options.
======================
Similarly 2.2 [2] and 2.3 [3] claim that graph nodes can have a "lexical 
value".  This could be an erratum, but the advent of 1.1 highlights the 
need for clarity.  Can the new 1.1 control characters be there?  Can 
non-XML characters be there?
======================
3.1.3 [4] describes the encoding of complex values by saying somewhat 
vaguely 

"For a graph edge which is distinguished by label, the [local name] and 
[namespace name] properties of the child element information item together 
determine the value of the edge label."

Well, they may determine the edge label, but how?  Whether we have a 1.1 
issue here depends on whether the data model allows 1.1 content, which is 
the issue raised above.  If so, then we need to set out the encoding 
rules, and to decide whether you can ever be in the situation of encoding 
a 1.1 data model into a 1.0 infoset (and if so what error is reflected.)
=======================
3.1.4  [5] discusses computing type names.  We might want to make clear 
that, unless XML schema is revised, type names cannot have >XML 1.0 names. 
 Ref to schema recommendation.
=======================
4.2.1 [6] on RPC invocation defers to appendix B on name mapping [7], 
which in turn refers to the [8] 1999 version of namespaces.  Probably not 
broken in an XML 1.1 world, but arguably surprising or confusing. Consider 
an RPC method name that has a character that is not a legal name character 
in the 1999 version of namespaces, but is legal according to the newer and 
also current "Namespaces in XML 1.1" [9].  Per our recommendation, that 
method name should be escaped, even if you are using XML 1.1.  As I say, 
not broken but confusing.
=======================

So, I think at least some of those need a bit of consideration.

Noah

[1] http://www.w3.org/TR/soap12-part2/#edgelabels
[2] http://www.w3.org/TR/soap12-part2/#graphnodes
[3] http://www.w3.org/TR/soap12-part2/#values
[4] http://www.w3.org/TR/soap12-part2/#complexenc
[5] http://www.w3.org/TR/soap12-part2/#enctypename
[6] http://www.w3.org/TR/soap12-part2/#rpcinvocation
[7] http://www.w3.org/TR/soap12-part2/#namemap
[8] http://www.w3.org/TR/soap12-part2/#XMLNS
[9] http://www.w3.org/TR/xml-names11/

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------






        <michael.mahan@nokia.com>
        Sent by: xml-dist-app-request@w3.org
        04/27/2004 08:50 PM
 
                 To: <xml-dist-app@w3.org>
                 cc: (bcc: Noah Mendelsohn/Cambridge/IBM)
                 Subject: Proposed text for issues rec20 and rec22 - Part 
2



I took an action item in last weeks teleconf to extend Noah's efforts in 
Rec20/22 
and look at part 2. I believe Noah was correct when he stated that he did 
not think 
much was needed in Part 2. [1]. 

I found 2 areas of text which MAY need attention

--------------------------------------------------

<current>
The lexical value of a graph node representing a simple value is the 
sequence of 
Unicode characters identified by the character information item children 
of the 
element information item representing that node. The element information 
item 
representing a simple value node MAY have among its attributes a 
'nodeType' attribute 
information item (see 3.1.7 nodeType Attribute Information Item). Note 
that certain 
Unicode characters cannot be represented in XML (see XML 1.0 [XML 1.0] 
>,[ref to XML 1.1]<). 
</current>

<proposed> little change, just wanted to make this section visible to all 
and verify. </proposed>

--------------------------------------------------

<current>
7.1.4 HTTP Media-Type
Conforming implementations of this binding:

1. MUST be capable of sending and receiving messages serialized using 
media type 
"application/soap+xml" whose proper use and parameters are described in A. 
The 
application/soap+xml Media Type.

2. MAY send requests and responses using other media types providing that 
such media 
types provide for at least the transfer of SOAP XML Infoset.
</current>

<proposed>
7.1.4 HTTP Media-Type
Conforming implementations of this binding:

1. MUST be capable of sending and receiving messages serialized using 
media type 
"application/soap+xml" whose proper use and parameters are described in A. 
The 
application/soap+xml Media Type.

2. MAY send requests and responses using other media types providing that 
such 
media types >MUST< provide >the transfer of SOAP message Infosets that are
serializable as XML 1.0, and MAY provide for transfer of SOAP message
Infosets that are representable using other W3C recommendation-level
versions of XML (e.g. [link to XML 1.1]).<
</proposed>

[1] http://lists.w3.org/Archives/Public/xml-dist-app/2004Apr/0012.html

Received on Monday, 3 May 2004 21:54:03 UTC