On 10/30/17 3:03 PM, Brian E Carpenter wrote:
On 31/10/2017 07:44, Miles Fidelman wrote:
...
somehow application layer developers don't think in terms of protocols,
they keep coming back to client-server models & software-oriented APIs,
rather than protocols
If an API is well documented and implies a well-defined state machine,
I would have no problem referring to it as a stable reference.
Brian
Well sure, but an awful lot of APIs aren't defined in terms of
well-behaved state machines, much less synchronized state machines. A
procedural call or method invocation (local or remote) is a different
conceptual model than an exchange of PDUs and keeping state machines
synchronized.
Miles
--
In theory, there is no difference between theory and practice.
In practice, there is. .... Yogi Berra