Am no expert on character encodings, but are there a few issues with the
current draft and character encoding? Namely,
1) The fields dix:/message-type, dix:/membersite-url, dix:/signature and
possibly others require Latin characters in their contents. However
isn't a browser free to post the form with a character encoding that
does not permit these characters?
2) The Canonicalization Algorithm relies on the browser posting the form
in the same character encoding used by the server to generate it. With
the current draft this can this be guaranteed?
This article
http://www-306.ibm.com/software/globalization/misc/code_considerations/index.jsp
provides for a good overview of the problem, although may be a bit out
of date. Having read this what do folks think of the following proposal
to address the above issues.
The dix spec ensures that information is always exchanged (via the
browser) encoded as UTF8. It does this by stating that the character
encoding of the forms pages must be UTF8 and also mandates the following
rules for the html forms used for the data transportation.
1. The HTML head section of the form MUST contain <META
http-equiv="Content-Type" content="text/html; charset=utf-8">. There
needs to be a corresponding statement for XHTML.
2. The FORM element MUST contain the Accept-Charset attribute and it
MUST be set to 'utf-8'i.e. <FORM Accept-Charset="utf-8" Type= ...>
The Canonicalization Algorithm can also now probably be simplified a bit
by stating that the sort is a byte sort of the utf8 representation.
Also, does it not make sense to have the verify step be a REST style xml
web service vs. this current form POST. XML seems much better suited to
these character encoding issues than
"application/x-www-form-urlencoded". Who knows whether the libraries
for Ruby, Python, Java et al will be consistent with how they perform
their character encoding.
**Rob
p.s. also in the spirit of internationalization I notice references to
first and last name properties in the specification. Have you
investigated the possibility of using vCards xml namespace "
http://www.w3.org/2001/vcard-rdf/3.0#"; for identifying user properties?
_______________________________________________
dix mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/dix
