W3C home > Mailing lists > Public > www-archive@w3.org > September 2005

Re: Query forms should be resources, not operations

From: Mark Baker <distobj@acm.org>
Date: Mon, 5 Sep 2005 14:20:33 -0400
To: Kendall Clark <kendall@monkeyfist.com>
Cc: Dan Connolly <connolly@w3.org>, www-archive@w3.org
Message-ID: <20050905182033.GX17056@markbaker.ca>

Hi guys,

I assume you won't mind that I've CCd www-archive...

On Mon, Sep 05, 2005 at 09:27:09AM -0400, Kendall Clark wrote:
> On Sun, Sep 04, 2005 at 08:36:22PM -0400, Mark Baker wrote:
> 
> > Much cleaner, IMO.
> 
> Mark, here's a quick response from me, not the group. You don't take into
> account the complexity costs of implementing yr design on the client side --
> now clients have to sniff or fully parse query strings to know *which* URL
> to dispatch to. -Shudder-

I don't see that.  They may know which URL to POST to by discovering
descriptive information about that resource at the time they discover
the URL.  For example, they might learn about the URL via a triple like;

<http://mystuff.example.org/asker/> rdf:type dawg:AskProcessor .

(FWIW, that's the basic discovery/interaction model behind my RDF
Forms[1])

But they needn't even do that, since the application might just drive
them there, as would be the case with the use of XForms I described.

> > As for other advantages, my approach also avoids the security problems
> > of tunneling operations through HTTP POST.
> 
> I don't believe this is true either. In yr design, we'd still have to
> specify POST equivalents for the case where SPARQL queries are too long to
> be serialized into GET. Which is the precise (and only, as far as I know)
> reason we have a POST binding now.

I agree that's the only reason you'd use POST right now (since all the
forms are "asking questions" and therefore better off using GET), but
I'm not sure what that has to do with tunneling operations.

> > FWIW, I looked for the discussion around this issue to try and better
> > understand the group's concerns, but couldn't find any, just the
> > resolution to contact me;
> > 
> > http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JulSep/0243.html
> 
> We discussed yr design and, as I recall, no one on the WG was motivated by
> its properties. And we discussed a design similar to yrs a few months back,
> arising from some ideas Dan Connolly had.

I wasn't aware of that.  Pointers?

> Thanks for the feedback and I suspect the WG will have another response.

Sure.  Thanks.

FWIW, if it wasn't clear, I consider this a Web architecture issue; that
the protocol hides four different resources behind a single URI and so
is counter to the advice of TAG's webarch doc when it says;

"Assign distinct URIs to distinct resources."

 -- http://www.w3.org/TR/webarch/#pr-uri-collision

 [1] http://www.markbaker.ca/2003/05/RDF-Forms/

Mark.
-- 
Mark Baker.  Ottawa, Ontario, CANADA.          http://www.markbaker.ca
Coactus; Web-inspired integration strategies   http://www.coactus.com
Received on Monday, 5 September 2005 18:18:42 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 7 November 2012 14:17:53 GMT