[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [sca-assembly] SCA Eventing Pub/Sub Contribution FAQ
From: Estefan, Jeff A (3100)
[mailto:jeffrey.a.estefan@jpl.nasa.gov] Martin et al., Frankly, I’ve been tracking this
thread for a while and I don’t get all the fuss around trying to add
event processing capability to a services paradigm programming model.
With the exception of insuring that the SCA programming model can support
multiple message exchange patterns (MEPs) for service invocation and
alternative response (whether it be support for asynchronous queuing,
request/response, request/callback, request/polling, batch processing, or, more
to the point, “event-driven” pub/sub), why are we attempting to add
support in SCA for the event processing paradigm in general? It’s
not a service-oriented paradigm; in other words, it is not capability- or
functionally-driven at its core. [<MartinC>]
In our experience, its very hard for an application or sub-set of an
application to fall neatly and exclusively into one of a client/server or a
pub/sub paradigm. In general, a mix of both is required to build complex
applications. The heart of SCA, IMHO, is the component nature of the model and if
one acknowledges that a component may offer or use services, and/or may generate
or consume events, then we need a way to describe and configure these facets. It would seem to me that we would only need
to insure that that SCA-Assembly spec support the notion that some entity that,
upon occurrence of certain events of interest, triggers the invocation of a
service. In that way, such an entity would simply be another SCA client
and communication between the consumer/client and provider entity (service)
could be any of the supported MEPs. [<MartinC>]
We tried to build support in such a way, but we found the client/server and
pub/sub paradigms to be too different to fit one on top of the other. So my point is, why not add an entry to the
eventing pub/sub FAQ list that questions the support for event processing
at all? [<MartinC>]
What would the question be? This is in contrast to the current list of
FAQs, which focus on the assumption that such support is required and what
needs to be changed in the SCA programming model to support it. As an aside, I would like to participate in
your Wed discussions on this topic but it conflicts with another TC meeting I
have. I’ll do my best to follow the thread and respond accordingly. Regards… - Jeff E., NASA/JPL |
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]