W3C home > Mailing lists > Public > xml-dist-app@w3.org > March 2004

Re: [XML11TF] (on the "restrictive" option)

From: Yves Lafon <ylafon@w3.org>
Date: Mon, 22 Mar 2004 15:45:22 +0100 (MET)
To: noah_mendelsohn@us.ibm.com
Cc: Herve Ruellan <herve.ruellan@crf.canon.fr>, Mark Nottingham <mark.nottingham@bea.com>, "'XMLP Dist App'" <xml-dist-app@w3.org>
Message-ID: <Pine.GSO.4.58.0403221513260.24294@gnenaghyn.vaevn.se>

On Wed, 17 Mar 2004 noah_mendelsohn@us.ibm.com wrote:

> I guess I see this as a much smaller concern.  RFC 3023 basically defers
> to the XML Recommendation(s), and XML 1.0 says [1]
>
> "All XML processors MUST be able to read entities in both the UTF-8 and
> UTF-16 encodings."
>
> ...both of which are encodings that are capable of conveying the full
> range of characters allowed by the {char} production.  While it's true
> that one can attempt to send encodings that either (a) would not be
> universally understood or (b) don't convey the full range of {char}, I
> think it's fair to say that our existing recommendations taken together
> provide fairly clear guidance on how to avoid such interoperability
> concerns:  send UTF-8 or UTF-16 and every legal Infoset will faithfully
> transmitted to a receiver that MUST be capable of parsing it.  I suppose
> we could add a note reiterating the use of other encodings may break
> interop.
>
> I see the XML 1.1 infoset question as much deeper for the reasons already
> stated:  introduce a fully conforming SOAP 1.2 intermediary (coded a few
> months ago to the then-current recommendations) into a path where you're
> sending XML 1.1 content and your envelope will not get to or past the
> intermediary.

Well, the intermediary will have first to understand a message coming from
a binding requiring knowledge of XML 1.1 (or more) in order to recostruct
the infoset properly. Failure to do so is at the same level of not being
able to handle a mustUnderstand.
If this intermediary then needs to retransmit the message using another
binding that don't have the proper capabilities, then we are in the same
situation as being a sender wanting to send an infoset with no available
binding.

The "failure to reconstruct an infoset" may only happen with an
unspecified binding, like our current HTTP binding.

Should we add also in part 1, 4 (binding framework) that a binding SHOULD
describe the infoset range that can be serialized/deserialized using this
binding?

-- 
Yves Lafon - W3C
"Baroula que barouleras, au tiéu toujou t'entourneras."
Received on Monday, 22 March 2004 11:02:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:59:16 GMT