W3C home > Mailing lists > Public > www-ws-desc@w3.org > April 2003

Re: Proposal for Describing Web Services that Refer to Other Web Services: R085

From: Mark Baker <distobj@acm.org>
Date: Sat, 26 Apr 2003 01:18:56 -0400
To: www-ws-desc@w3.org
Message-ID: <20030426011856.K23133@www.markbaker.ca>

On Fri, Apr 25, 2003 at 02:21:57PM -0600, Champion, Mike wrote:
> [Off topic, but hey it's Friday!]

It's not off-topic at all.  It's directly relevant to this
requirement and proposal.

The current approach, as reified in many Web services specs
and this wsdl:binding proposal, is akin to;


where you could just as easily use;


and get all the benefits of URIs in doing so.

> Let's look at a concrete example.  Software AG's main number on our homepage
> is listed as  +49 6151 92-0  To actually call the receptionist there from a
> random spot in the world, you need more information than just the phone
> number. From my cellphone, I can actually enter +49 6151 92 0 . From my
> home, I would dial 0114961510.  From most businesses in the US, one would
> dial 9011496151920  From most places in Germany (as best I understand the
> system!) one would dial 06151920.  From within the building, you would just
> dial 0.  From a random hotel in Paris, or Hong Kong, I have NO IDEA!
> There's usually a little booklet next to the phone that explains the local
> conventions.

That's a problem with the phones; they haven't been upgraded to allow you
to dial "+49 ..." directly.  There's no reason why they couldn't accept
that string as-is.  In fact, my old cell phone (Motorola Timeport) did
just that.

> Worse, one often needs to know the "after the answer" protocol for dialing
> an specified extension, conference access code, or whatever.  As a practical
> matter, many of us also need to know the local access numbers for our
> calling card providers and the local convention for using it (for example, I
> can attest that the order in which you dial the account code and the desired
> number is different for MCI in the US and in Germany).  And if you  using a
> modem, you have the additional problem that physical dial tone signals are
> different in different countries, and you may have to pause dialing at
> various points to make sure that the phone network is ready for the next
> digit. I've spent long hours in hotels trying to figure out how to call an
> ISP to get email; I guarantee you I did not "only need an identifier"! 

I'm only talking about establishing the call.

The phone system uses late binding like the Web does.  All large
networks use late binding; it's the only known way to get integration
complexity to the magical O(N) level[1].

 [1] http://seanmcgrath.blogspot.com/2002_12_29_seanmcgrath_archive.html#90126406

Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca
Web architecture consulting, technical reports, evaluation & analysis
Received on Saturday, 26 April 2003 01:17:08 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:54:42 UTC