W3C home > Mailing lists > Public > www-ws-arch@w3.org > July 2002

Re: Generic/specific connectors

From: Mark Baker <distobj@acm.org>
Date: Mon, 22 Jul 2002 19:54:10 -0400
To: "Champion, Mike" <Mike.Champion@SoftwareAG-USA.com>
Cc: www-ws-arch@w3.org
Message-ID: <20020722195410.Z1586@www.markbaker.ca>

Hi Mike,

On Mon, Jul 22, 2002 at 04:49:39PM -0600, Champion, Mike wrote:
> This does seem to be the crux of the  debate.  The REST/Web camp advocates
> applications taking responsibility for many of the details of state
> management, notification, reliability, security, etc.  The other side (OMA /
> GXA / whatever) believes that the infrastructure should handle as many of
> these details as possible.
> 
> Would you agree Mark? 

Yes, but.  Yes, it does make it more complex, but we've had that licked
since the very first Web server was developed.  It takes care of the
dispatching of a particular message to your app based on some aspect of
the message itself (URI, media type, headers, etc..), depending upon how
you configure it.

But even if it were still slightly more complex, what you gain as a
result is interoperability.  And on the Internet, that is required, not
just a "nice to have".  As I said before, every successful system on the
Internet has made this tradeoff (a round of beers for the WG if anybody
can find a counter-example 8-).

I'm also not saying that the infrastructure can't take care of some
things, just not all of them.  Reliability is one.  Security is another
(but I don't think people will disagree with that).  Notification, can
actually be handled by the infrastructure - we have done that - but it's
not as good as if it were handled by the app (though still *really*
useful).

> I've been wondering about this too.  In the  HUMAN CENTRIC hypertext Web,
> i.e. the web we know and love, this is not much of an issue because a human
> can look in a directory or search engine and follow "interesting" links, and
> look at the text and either use it or ignore it.  Do we need a lot more
> "semantic" infrastructure before this works reasonably well for
> program-to-program communications?

Yes, definitely.  We need a way for a document to say "this invoice
is subordinate to this order", so that an order processing app can know
which URI to invoke GET on to retrieve the information it needs.  The
Semantic Web gives you the technology you need to do that.

The weak link in machine-to-machine processing on the Web, isn't the
protocol, it's that most of the content is HTML, a data format that
doesn't exactly define many e-business-y semantics (except "address", I
suppose 8-).

MB
-- 
Mark Baker, CTO, Idokorro Mobile (formerly Planetfred)
Ottawa, Ontario, CANADA.               distobj@acm.org
http://www.markbaker.ca        http://www.idokorro.com
Received on Monday, 22 July 2002 19:42:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:25:02 GMT