Re: Another try at HTTP binding

Hi Philippe,

This is looking better.  I like the trend towards being more of a forms
language, but it doesn't appear this proposal has quite embraced them.
Perhaps that's ok, I'm not sure; we're breaking new ground here, trying
to blend description languages and forms ...

Consider example 4.1;

  <http:operation
      location='temperature/{town}'
      method='get' separator='&'/>

When interpreted as a form delivered at runtime, that's asking for the
client to provide a "town" so as to complete the URI which could have
GET invoked upon it.  That differs substantially from the use in the
example though, as the other data (date, unit) is not used.  What I
think is really going on here, is an order-of-operations problem; that
example assumes a document exists and needs to get to an endpoint
somehow, while the form view say that form itself determines the
document that is sent.

One way forward would, I suppose, be to describe how to use XForms (or
form technology in general) with Web services.  Dunno, just tossing out
an idea.

Or perhaps somebody else has some other ideas?

Mark.
-- 
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca