- From: Mark Baker <distobj@acm.org>
- Date: Thu, 18 Jul 2002 23:45:47 -0400
- To: David Orchard <dorchard@bea.com>
- Cc: "'Champion, Mike'" <Mike.Champion@softwareag-usa.com>, "'Burdett, David'" <david.burdett@commerceone.com>, www-ws-arch@w3.org
On Thu, Jul 18, 2002 at 12:07:46PM -0700, David Orchard wrote: > RE: REST, Conversations and ReliabilityHeck, I wouldn't accept a WSA built > on RPC/OMG OMA principles either. There's this great stuff like SOAP, XML, > URIs, HTTP, the web, etc. out there ;-) No fine-grained tightly coupled > object/API based binary encoded interchanges and binary encoded addressing > formats that ignore the network and specify implementation details for me... You're talking about a particular configuration of CORBA with that description, not the OMA. Neither the OMA nor CORBA is necessarily binary-encoded. See XIOP; http://xiop.sourceforge.net/ OMA/CORBA also isn't any more fine grained than typical Web services. Nor is it any more tightly coupled, because interfaces are cleanly separated from implementation. So are you really sure you're not redoing the OMA? An easy way to tell would be to ask you to identify what you see as the architectural elements for the Web services architecture. i.e. components, connectors, and data elements per; http://www.ics.uci.edu/~fielding/pubs/dissertation/software_arch.htm Or pick another recognized taxonomy if you don't like that one. > Where the analogy to OMG OMA is useful is in describing the goals and > objectives of application to application integration, and learning from much > of what worked and didn't with OMG - because certainly OMG worked in many > areas. In particular, the separation of concerns between the application > and the services of the underlying network infrastructure, like reliability. > It's probably really useful in other areas that I just haven't thought of in > my current 2 minutes of thought on the topic. Well, I've thought long and hard about this for 4+ years, and I've yet to see any aspect of the OMA that doesn't have a superior parallel in REST, when evaluated for Internet computing. Perhaps there's something I've overlooked, but I doubt it, because OMA/CORBA was never designed for Internet deployment (despite some believing that it was); it was designed for Intranets, and curiously enough, that's the only place it saw any success (and by "Intranet" I include VPNs - anywhere where a single administrative domain exists, providing an out-of-band coordination channel between service provider and consumer). MB -- Mark Baker, CTO, Idokorro Mobile (formerly Planetfred) Ottawa, Ontario, CANADA. distobj@acm.org http://www.markbaker.ca http://www.idokorro.com
Received on Thursday, 18 July 2002 23:42:53 UTC