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

On Wed, 17 Mar 2004, Herve Ruellan wrote:

> Mark,
>
> I think that this issue is somewhat linked to the main one from Yves' email.
>
> Yves's email was about how to restrict what is allowed in the SOAP
> infoset to ensure the best possible interoperability between SOAP nodes.
> The main point of his proposal is to restrict the content of a SOAP
> infoset to what is serializable with XML 1.0. However, he notes that due
> to the different character encoding allowed in XML 1.0, this might not
> be sufficient to ensure full interoperability between two SOAP nodes.

There are two different goals raised during the teleconference:
1/ what was intended by the WG when we decided to go for the infoset
2/ interoperability issue of XML 1.1 (extended to whole infoset range).

Point 2 is the main reason for restricting the range of allowed characters
in the infoset, and I was pointing out that even with infoset fulfilling
the "can be serializable using XML 1.0" requirement, interoperability is
NOT granted based on different characters encodings. However I agree those
are different issues, both under the hat of interoperability.

Noah made a great summary of the first issue in [1], and my comment is
about the statement in his option II:
<<<
So, one way we get universal interop, but are restricted to 1.0-style
content only.
>>>
Which is not universal per encoding issues.
If the WG choose to use option II, then only the HTTP binding will have to
be fixed, as allowing XML 1.0, 1.1 and different encoding is already a
source of interoperability for perfectly valid SOAP infosets.

Using the HTTP binding for a service might not always be the best option,
depending on the kind of MEP you want, the infoset you want to serialize
and so on.
Restricting to 1.0 only hoping that it will solve interoperability issue
_in general_ is wrong because it won't, even for the HTTP binding.

However, for interop of already deployed SOAP/1.2 stacks using the current
HTTP binding, made in the light of XML 1.0 only, the second question of
Noah's email should have option 1 as a reply.
<<<
Issue an erratum clarifying that XML version 1.0 MUST be the serialized
form used with application/xml
>>>>
And perhaps adding UTF-8 and UTF-16 as the only possible encodings as
well...

[1] http://lists.w3.org/Archives/Public/xml-dist-app/2004Mar/0024.html

-- 
Yves Lafon - W3C
"Baroula que barouleras, au tiéu toujou t'entourneras."

Received on Wednesday, 17 March 2004 08:03:09 UTC