- From: Mark Baker <distobj@acm.org>
- Date: Mon, 21 Oct 2002 20:28:12 -0400
- To: "Burdett, David" <david.burdett@commerceone.com>
- Cc: www-ws-arch@w3.org
On Mon, Oct 21, 2002 at 03:18:31PM -0700, Burdett, David wrote: > Mark > > Your example involves a human, you, who gives permission to another human (I > think) It was meant to be a software agent. > who is your agent who does the search. Even if the agent isn't a > human and is software, then how did the agent software get, without any > human intervention: > 1. The intelligence to carry out and analyze the results of the search Let's just blackbox that; no matter how you implement this, you're going to need something that finds some service that matches what you're looking for. > 2. Understanding the semantics the schemas for the menu and order It's hardcoded to do that. > 3. Knowlesge that, after the receiving the menu, it could send an order. It's hardcoded to know that's possible when it sees the part of the schema that points to a menu-submitting processor identified by some URI in the menu. > I don't think your example qulaifies as "no human involved". Well, no humans are involved, which is the most important criterion in that determination I'd say. 8-) > The point I'm trying to make, is that if these things aren't standardized, > then you need to re-program them **every single time**. This won't happen as > it is too expensieve. I think we both agree that some more things need to be standardized. I think we disagree how much; I think not much more, because we've already got a suitable interface (GET/POST) and a suitable state engine (hypermedia). Unfortunately, most of the rest of the industry appears to believe that a *lot* more needs to be standardized. MB -- Mark Baker, CTO, Idokorro Mobile. Ottawa, Ontario, CANADA. http://www.markbaker.ca http://www.idokorro.com
Received on Monday, 21 October 2002 20:26:26 UTC