Comment on WS-I18N WD

These two sets of comments were previously posted to other lists.  I'm
re-posting them here to make sure that they don't get lost.

Mary
====================================================================
From: Chris Ferris <chrisfer@us.ibm.com>
Date: 09/16/2005

1. <IBM Chris Ferris> There is no normative (or non-normative for that
matter) reference to SOAP or WSDL.  Is there a reason for this?  I think
that the spec should try to provide a binding to both SOAP1.1 (possibly
non-normative) and SOAP1.2 as well as to both WSDL1.1 and WSDL2.0 so as
to be most practically useful given that at present, SOAP1.1 is most
commonly used for interoperability and that WSDL2.0 may endure a rather
protracted roll-out given its complexity, and also given that MSFT seems
dis-inclined to target its adoption for Longhorn.

2. <IBM Chris Ferris> It references WS-Routing, yet the actual reference
in the references section is to the W3C WS-Addressing specification.
WS-Routing is defunct as far as the set of WS-* specs is concerned.

3. <IBM Chris Ferris> Secondly, there seems only to be a mapping to
WSDL2.0 features and properties.  IBM (amongst others) doesn't support
the F&P aspect of WSDL2.0 [2].  Basically, it directly competes with
WS-Policy.

4. <IBM Chris Ferris> IMO, the spec should be leveraging the SOAP1.2
soap:relay attribute [3] and should use the
http://www.w3.org/2003/05/soap-envelope/role/next role for i18n
processing to achieve the desired effect (or so I would imagine) of
having all processing on the message be performed using the desired
locale.  At the very least, this should probably be recommended by the
spec.

5. <IBM Chris Ferris> I'm a little concerned about the use of the "$"
for certain of the values e.g.  <locale>$default</locale>.  My concern
is that with certain scripting environments such as PHP, that it may be
confused with variable substitution and will require escaping.  I guess
I don't understand why the value space for the locale isn't a URI.

6. <IBM Chris Ferris> I think the spec needs a little more specificity
with regards to the content model of the <i18n:international> element.
Specifically, it should be described using XML Schema (IMO) and also
should probably be described in terms of the infoset...  similar to the
way that SOAP's elements and attributes are defined in the SOAP1.2 spec.

====================================================================
From: Jonathan Marsh <jmarsh@microsoft.com>
Date: Thu, 27 Oct 2005 16:15:43 -0700

Based on discussion by the WSDL WG based on an initial review by Tony
Rogers, we'd like to make the following comments about this spec.

In general, this draft is in good shape.

Some minor points:
1. although it refers to WSDL repeatedly, starting from the third line
of the introduction, it does not include a reference to WSDL in the
References appendix.

2. also in the References appendix, it has a reference to WS-Routing,
but the spec referenced is WS-Addressing.

3. Example 4 could be confusing.  It contains two instances of a
preferences element, despite the first item in section 3.4 stating that
only one preferences element is permitted.  This example could lead a
less careful reader to believe multiple preferences elements are
permitted.

4. Example 5 makes two references to a language spec of "de_DE" - should
this be "de-DE"?

5. Says "here are some document examples", but gives only one.  Perhaps
more will be added shortly?

6. Section 3.2 lists, in the first paragraph, special language tags
"$neutral" and "$default".  Should it also list "$user"?  "$user" is
covered in Section 4, but should probably appear in 3.2 as well.

7. In Example 6, the constraint element needs clarification.  The value
shown, "locale:$user", is not a valid value for this element (should be
a valid QName pointing to a schema type, ideally illustrated within the
spec as well).

------------------------------------------------
Mary K. Trumble
Tel: (512) 838-0094; T/L 678-0094
mtrumble@us.ibm.com

Received on Tuesday, 24 January 2006 13:57:31 UTC