RE: WSDL, app protocols, URI schemes

Mark,

I look at it a bit more wholistically than just looking at what you call the
methods.  To make a true comparison of wsdl documents to browser centric
constructs, one also needs to include an html form and the html form
encodings section.  So you are just looking at the "method", but I would
like to look at all the data necessary to invoke a service.  Which includes
methods and data formats.

In the trivial, you are correct.  Given www.google.com and the HTTP URI
scheme, I can invoke google.  But there's little value.  Where the value
comes is from filling in the form that google returns.  And that form is
defined by the google authors, who also obey the html form encodings because
browsers also follow the form encoding.  To figure out how to invoke
"http://www.google.com/search?hl=en&ie=ISO-8859-1&q=foo", a human has to
look at the form and enter the data.  Now I know that you'll argue, "ha! you
just include the form result, my point is proven"  But I still had to enter
the form data!  At the end of the day, a human had to do the entry.  And if
Google ever changed it's interface and you tried the link, you're browser
would break.  Of course, you as a human are smart enough to figure out what
to do when it breaks, but we'd like computers to know more definitively.

The combination of an HTML form, HTML form encodings section and URI Scheme
is the equivalent of the WSDL operation, WSDL port, and URI scheme.

I argue that looking at just the URI scheme isn't very useful.  And I
further argue that it is XML that causes WSDL et al to happen.  Humans can
enter character data - we have keyboards.  So HTML Form encoding (name value
pairs) makes a perfect sense for browsers.  But XML means that we can pass
arbitrary structures around.  So the solution is to create a better model
for describing interfaces.  And comparing WSDL to URI schemes makes WSDL
seem less expressive, when it's actually far more expressive than pre-wsdl
technologies.

Cheers,
Dave

> -----Original Message-----
> From: www-ws-arch-request@w3.org [mailto:www-ws-arch-request@w3.org]On
> Behalf Of Mark Baker
> Sent: Thursday, September 26, 2002 5:47 PM
> To: David Orchard
> Cc: www-ws-arch@w3.org
> Subject: WSDL, app protocols, URI schemes
>
>
>
> On Thu, Sep 26, 2002 at 02:20:45PM -0700, David Orchard wrote:
> > This is comparing apples to oranges.  The relationship
> isn't between URI
> > schems and wsdl files.  On the web, we don't need wsdl
> files because humans
> > know how to fill out forms.
>
> Hmm, no, that's comparing apples to oranges.  I know we disagree about
> the capabilities of a uniform interface, but I'm surprised we disagree
> about this.
>
> With both Web services (well, our definition at least) and the Web,
> there are "objects" identified by URIs, and those objects all expose a
> set of methods which can be invoked over a network.  Right?  The
> methods of a Web service are the port types defined in a WSDL
> document.
> The methods of a Web resource are defined in the application protocol
> specification(s) associated with the URI scheme (if it's resolvable).
>
> So, the URI scheme is analogous to a name for a WSDL document, and a
> WSDL document is analogous to a specification of an application
> protocol ... hence my comments here;
>
> http://lists.w3.org/Archives/Public/www-ws-arch/2002Aug/0306
>
> MB
> --
> Mark Baker, CTO, Idokorro Mobile (formerly Planetfred)
> Ottawa, Ontario, CANADA.               distobj@acm.org
> http://www.markbaker.ca        http://www.idokorro.com
>
>

Received on Friday, 27 September 2002 13:05:52 UTC