2008/9/5 Rob Eamon <[EMAIL PROTECTED]>: > --- In [email protected], JP Morgenthal > > <[EMAIL PROTECTED]> wrote: >> >> Interesting timing of your question Rob. I pose the blog entry >> Gervas posted from me as a partial response to your question. >> However, I would be careful about the statement "Services do not >> have UIs" as all services have interfaces. I'll go so far as to >> say that a service could have a human interface in addition to a >> UI. > > I guess we disagree. The nature of UIs (or human interfaces) is such > that they would tend to make the service stateful and too context > specific. Those aspects are better managed, IMO, by service clients > which are typically built for a given audience and context.
But those service clients are, from the perspective of the user, a producer and arbiter of access to capabilities. In other words they provide a service. This is why I put the concept of "Virtual Services" into the methodology piece as it recognises that there are some services that do not provide capability directly themselves but merely act as a re-direct to the underlying capabilities. >From the consumers perspective it walks like a service, talks like a service and acts like a service. Steve > >> That said, I would be careful to differentiate platform from >> architecture. > > Agreed. Now if Linthicum would refrain from claiming that a platform > will drive SOA, that'd be cool! :-) > >> To me Chrome is the platform and SOA is the architecture that >> something running on that platform might access. > > This one gives me the heebies a little. A platform doesn't "access" > an architecture, IMO. Chrome, or any browser, would be the > implementation detail of a particular architecture component. It > isn't separate from the architecture. It is a part of it. It is the > host platform of a service client. > > I subscribe to the view that SOA isn't an architecture at all. It is > a style. The architecture, which will follow SO and other principles, > is the EA, the AA, etc. > > To help clear up confusion, would it be helpful if we stop referring > to SOA as a thing that is instantiated/implemented? SO is a > characteristic of a BA, EA, AA, etc. It isn't an architecture itself. > Would promoting this view help move things away from equating SOA > with specific technologies/implementations? > > -Rob > >
