Reworking of HTTP serializations following F2F

Fulfilling my action item, I have reworked the section defining HTTP
serialization formats.

The result is at:

  http://www.w3.org/2002/ws/desc/5/09/wsdl20-adjuncts.html#_http_serialization

with a diff-marked version agains the current editors copy at:

  http://www.w3.org/2002/ws/desc/5/09/diff.html#_http_serialization

In CVS, it's implemented in the http_serialization_reworking branch.

Now on to the details.

I was asked to implement the table at [1], which was going to include
the following fixes:
1. allow POST as application/x-www-urlencoded
2. when using POST as application/x-www-urlencoded, allow the instance
   data to be entirely serialized in the message body
3. replace the {foo/} notation by a property

1 and 2 have been accomplished. While doing this work, I realized
that 3 was actually not needed if we serialize all the instance data:
either the remainder goes in the request IRI, or it goes in the
message body. So the syntax is simplified.

When fixing the application/x-www-urlencoded POST serialization, we
lost the ability to serialize part of the instance data in the URI (as
application/x-www-urlencoded was incorrectly serializing POST as
application/xml).

That could be fixed by having the templating mechanism of
application/x-www-urlencoded be applicable to the other serialization
formats, when the IRI style is in use.

I have not done this as it was going further than what the WG had
decided during the F2F, and I am wary of doing design changes at this
point. However, the removal of this feature while fixing the bug may
be an issue (a substantive change). So we should have more discussion
on what we want to do about that.

Cheers,

Hugo

  1. http://www.flickr.com/photos/psd/47263522
-- 
Hugo Haas - W3C
mailto:hugo@w3.org - http://www.w3.org/People/Hugo/