Rob Eamon wrote: > --- In [email protected] > <mailto:service-orientated-architecture%40yahoogroups.com>, Michael > Poulin <m3pou...@...> wrote: > "an access to a business functionality does not make it a service; it > is just one of many access channels: an interface does not change the > core of the things." > > I read more into that comment than was there. I agree that just adding > an interface is not sufficient to expose a particular capability as a > service. > > Where we may diverge is in how much effort there is to making it a > "proper" service. IMO, there isn't much more to it. Making it > discoverable, describing its behavior, listing its operations and > interfaces, etc. can be accomplished in any number of ways--none of > which are overly difficult.
One of the cool things about Java Mobile code, is that if you have an interface, which for example, used to be used locally in the JVM, and it represents the functionality that you want to make public, you can optimize away problems with methods which might be "iterator" based rather than "collection" based, by using a smart proxy to move some of the "service's" behavior into the client's proxy so that batching gets done behind the scenes by the smart proxy to make an interface which doesn't work well when used directly over the network, to be usable because it simplifies the clients view of the service. This can create some difficulty in moving the local operations into a service oriented view, but I tend to agree with Rob that the 1-3 steps are actually quite easy to do and it's really the other stuff that matters the most. Gregg Wonderly
