W3C home > Mailing lists > Public > www-ws-arch@w3.org > January 2004

Re: Proposed replacement text for Section 1.6

From: Mark Baker <distobj@acm.org>
Date: Sun, 25 Jan 2004 15:40:42 -0500
To: "Champion, Mike" <Mike.Champion@SoftwareAG-USA.com>
Cc: www-ws-arch@w3.org
Message-ID: <20040125154042.O3790@www.markbaker.ca>

On Sun, Jan 25, 2004 at 02:35:51AM -0500, Champion, Mike wrote:
> > I don't think it's useful to mention the uniform interface 
> > there, since it doesn't distinguish it from the Web where the 
> > uniform interface constraint is pervasive (GET and POST 
> > specifically).  IMO, the biggest difference between the style 
> > of the architecture of the Web, and REST, is cookies, by a long shot.
> 
> Well ... I guess I was thinking of the fact that HTTP POST is used by an
> awful lot of sites as more or less a generic data tunnel to a server-side
> process and not a way to request that a server "accept the entity enclosed
> in the request as a new subordinate of the resource identified by the
> Request-URI."

*shrug*.  I see those as pretty much identical.

>  The list of examples of what POST is supposed to be used for
> in RFC 2616 seems awfully limited compared to what is actually done --
> essentially RPCs, in many cases even before SOAP was invented. XML-RPC and
> SOAP were originally intended more to bring order to the chaos out there in
> the world of hand-coded RPC over CGI than to add new RPC capabilities,
> AFAIK.  So, it seems highly contrived to argue that the Web As It Is uses a
> constrained interface, but SOAP-RPC is a Bad Thing because it does not.

I don't want to get into value judgements here because I've had to do
RPC-like stuff on occasion.  I'm not saying that RPC doesn't exist on
the Web either, only that it's extremely uncommon, and so much so that
it can effectively be ignored for the purposes of an architectural
examination of the relationships between elements.

Consider that there's O(1000000000) Web resources which use GET[1],
probably something like O(1000000) which use POST[2], and just O(100)
which use RPC[3].

 [1] http://www.google.com
 [2] 1 out of 1000 pages using forms seems a good ballpark figure, give
     or take
 [3] http://www.markbaker.ca/2002/04/WebServicesGrowth/

>   I
> do agree that cookies are an even more glaring example of the difference
> between REST and the Web As It Is.

Ok, good.

> > Also - and I think I commented on this before - "CRUD" is 
> > only an example of another constrained interface, not one 
> > constrained to have uniform semantics.  I'd suggest just 
> > saying "-- GET, PUT, POST, DELETE, and any other method which 
> > can be implemented by all resources".
> 
> I'd be happy to adopt your suggestion here.

Great, thanks!

Mark
-- 
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca
Received on Sunday, 25 January 2004 15:41:30 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:25:25 GMT