Implications of XML 1.1 for WSDL 2.0

This note completes my action item of 2004-04-29.

The WSDL 2.0 spec is written in terms of the XML Infoset. The Infoset spec 
was recently revised [1] to cover XML 1.1. There are at least two changes 
in XML 1.1 that affect the Infoset:

1. The set of legal values for character data has increased. However, on 
closer reading of the Infoset spec, it defines a character code as any 
number in the ISO 10646 range of 0 to #x10FFFF, even though not all of 
these are legal XML codes. So in this case there is no problem since the 
Infoset spec is general enough to describe even documents that are not 
well-formed XML 1.1, i.e. that contain illegal character codes.

2. The set of legal values for characters that may appear in element 
names, attributes names, enumerated attribute values, targets of 
processing instructions, etc., has increased. The Infoset spec doesn't 
define these, instead saying that it takes whatever an XML parser hands 
it.

There are also changes to Namespaces that are reflected in the new Infoset 
spec.

My conclusion is that the Infoset spec is loose in the sense that it 
doesn't nail down every aspect of an Infoset. That is fine for the Infoset 
spec, but I do NOT think it is fine for WSDL 2.0, since in practice 
systems will be interchanging concrete documents using XML 1.0 or XML 1.1. 
If we are not clear on this, then we will probably hit interoperability 
problems when XML 1.1 gets adopted.

I think there are two extreme approaches and probably some intermediate 
cases.

Case 1. We state that WSDL 2.0 MUST use  XML 1.0 only. This means WSDL 2.0 
documents MUST use XML 1.0 and that any imported or included document MUST 
also use XML 1.0.

Case 2. We state that WSDL 2.0 MAY use either XML 1.0 or XML 1.1 and that 
conformant processors MUST support both, and any combination of versions 
for all imported or included documents.

Case 1 has the advantage of simplicity and ease of implementation. 
However, it denies the benefits of XML 1.1 which may be significant to 
some users. For example, ease of authoring on IBM-compatible mainframes 
(NEL), and the flexibility to use more international characters in names.

Case 2 has the disadvantage that it is harder to implement since there are 
not a lot of XML 1.1 parsers in production yet. It has the benefit of 
making the features of XML 1.1 available.

I suspect that even if we adopt Case 2 that WS-I will issue a Profile that 
specifies Case 1.

[1]  http://www.w3.org/TR/xml-infoset/

Arthur Ryman,
Rational Desktop Tools Development

phone: +1-905-413-3077, TL 969-3077
assistant: +1-905-413-2411, TL 969-2411
fax: +1-905-413-4920, TL 969-4920
mobile: +1-416-939-5063
intranet: http://w3.torolab.ibm.com/DRY6/