- From: Brian E Carpenter <brian@hursley.ibm.com>
- Date: Fri, 14 Dec 2001 18:18:19 +0100
- To: Mark Baker <distobj@acm.org>
- Cc: Keith Moore <moore@cs.utk.edu>, janssen@parc.xerox.com, discuss@apps.ietf.org
Mark, I worked in process control systems for ten years or so, and my answer is: no, mechanisms such as you describe aren't good enough. You need transactional reliability. Brian Mark Baker wrote: > > > I can't see any basic reason why a thermostat should be accessed as if > > it was a hypertext document. > > I would do it for the same reason that anything ends up on the Web; > > - I can manipulate it from the same app I use to manipulate so much of > my life already; the browser > - it can leverage existing and yet-to-be-developed extensions such as > WebDAV so that access to it can be locked, versioned, etc.. > - if I forget its URL, Google can find it for me > - if my house network is unreliable, an existing cache on the network > can let me know what the last cached state of my thermostat is, and > when that state snapshot was taken > - authentication for free > - content negotiation permits my french-speaking serviceman to > also manipulate and debug it remotely through the same URL > > etc.. > > If that isn't reason enough for you, then do it just to prevent the > APPS area from needing 10 area directors by the end of the decade. 8-) > > > I would expect building-services control > > messages to get layered directly over a transport protocol in due > > course. I certainly want thermostats to be accessed by a highly reliable > > mechanism that survives disconnected operation, which is where this > > thread started. > > You didn't think my explanation of reliability in the context of > thermostat control was sufficient? Can you elaborate on what else > you'd think would be needed? > > MB > -- > Mark Baker, Chief Science Officer, Planetfred, Inc. > Ottawa, Ontario, CANADA. mbaker@planetfred.com > http://www.markbaker.ca http://www.planetfred.com
Received on Friday, 14 December 2001 12:18:59 UTC