Re: Issue 501 closed

Dear Martin G. and the XMLPWG,

The I18n WSTF has discussed your comments and have the following 
response (inline):

Martin Gudgin wrote:

> Dear Andrea and I18WSTF,
> 
> You raised an issue, 501[1] regarding the SOAP Resource Representation
> Header specification[2]. Please note that this issue covers the first 4
> points in your e-mail[3]. The XMLP working group considered your points
> and has the following response:
> 
> Points 1-3: Yes, when using the resource representation header base64 is
> always a requirement, even for textual types. The SOAP envelope itself
> will always be in a single character encoding. The octet stream
> resulting from decoding some base64 text may well be in a different
> character encoding. This is not an issue. The character encoding in use
> for such data may be determined in a number of ways, including, but not
> limited to; specifying the charset as part of the xmime:contentType
> attribute (e.g. text/xml; charset=iso-8859-1 ), examining the XML
> declaration for XML based types (e.g. <?xml version='1.0'
> encoding='iso-8859-1' ?>, using the algorithm defined in Appendix F of
> the XML 1.0 Recommendation for XML based types, assuming a default
> character encoding defined by the specification of the media type.

While we who work in i18n every day are well aware of the ways of 
declaring charset, most people are not.  If it is not mentioned in the 
standard, the likelihood of it being done is very slim.  Since we 
believe there is a minimal amount of information necessary to process 
data, and that information is, for text, charset and in some cases 
language, we feel that this needs to be described in this context.

We also believe that forcing base64 encoding on readable text is a 
mistake which will introduce a number of problems, not the least of 
which is masking the fact that the inline data needs to be tagged.  We 
feel it should be strongly discouraged, if not disallowed.

The benefit of making this very basic information a standard is to allow 
ease of interoperability, which is the goal of standards in general. 
Text is the most basic of data exchanged, and without proper tagging 
it's not reliably processed.

We believe that for the reasons stated above, this should be covered in 
the test suite.

> 
> Point 4: xml:lang is not appropriate for use on the rep:Data element as
> base64 is not human-readable text. A SOAP message can carry multiple
> instances of the resource representation header and many such headers
> may carry representations of the same resource. Thus a given SOAP
> message could carry multiple representations of a given resource, each
> one in a different human readable language. The resource representation
> header has an extensibility mechanism that allows additional attributes
> to be specified. Such an attribute could be defined to indicate the
> human readable language of a text based resource. We note that there is
> an example of how to use this extensibility mechanism in Section
> 4.4.3[5] of the CR version of the Resource Representation SOAP Header
> Block specification[4]

We believe that at a minimum precedence rules need to be specified for 
determining the language of included data.  For example, do the MIME 
headers take precedence over the HTTP headers?

Again, mention of this issue and firm recommendations for how to handle 
it would enhance interoperability significantly.

Andrea Vine
Sun Microsystems
For the I18n WSTF

> 
> The working group does not expect to change the Resprentation header
> specification as a result of closing this issue.
> 
> Regards
> 
> Martin Gudgin
> Microsoft Corp.
> For the XML Protocol Working Group
> 
> [1] http://www.w3.org/2000/xp/Group/xmlp-cr-issues.html#x501
> [2] http://www.w3.org/TR/2004/WD-soap12-rep-20040608/
> [3] http://lists.w3.org/Archives/Public/xmlp-comments/2004Sep/0000.html
> [4] http://www.w3.org/TR/2004/CR-soap12-rep-20040826/
> [5] http://www.w3.org/TR/2004/CR-soap12-rep-20040826/#rep-http-headers
> 

-- 
The trouble with the world is that the stupid are cocksure and the
intelligent are full of doubt. -Bertrand Russell, philosopher,
mathematician, author (1872-1970)
[...shouldn't that end with "or maybe not?"]

Received on Wednesday, 29 September 2004 01:50:00 UTC