Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [stp-dev] SC subproject issues

I have several issues with this subproject first of which is the need for the subproject. It is not very clear at all and has a great deal of overlap with what is already defined in the core subproject.

I tend to think of the SC/Core/SOAS subprojects as supplying the good for
the three development phases of a SOA implementation. First, you create
service interfaces, implementations and policy choices, then you assemble
all of the services together to create deployable modules, then you
coordinate a deployment of potentially many modules to a suite of containers
or runtimes.

Maybe what is causing the confusion is the fact that there are already
existing ways to create the services - using plain old java, using the
WTP tools - that you can then put together in the assembly and that all
you need to provide is an introspector extension. However, there
are some approaches to making services that aren't yet supported by
Eclipse tools, for example JAX-WS and C++. I think
that the service creation project is where these tools should go when
they are created - it seems the most suitable place - and having all of
these capabilities in the same place means that we can make the most of
the work done, by sharing code.

However my biggest problem is with the exemplar contribution. Why in the world would we choose to use Celtix as the exemplar implementation? First of all Celtix is a specific runtime technology and is not a simple implementation type. Second we should be using standards for our exemplar. In the core subproject we are planning to contribute the Java implementation as the exemplar. That seems much more reasonable than Celtix. If you want a more complex implementation type we could look at Java EE 5. I think having Celtix integrated in this framework is fine but not as the exemplar.

Ok, this is a misunderstanding - there's no intention of committing
Celtix as a runtime technology into this project! What we are aiming
to do is to put together JAX-WS service creation tools that will
integrate with the core and the deployment framework. JAX-WS is a
standard, and up until very recently we were very limited in terms
of compatibly licensed JAX-WS runtime choices - Celtix was the one
we could use. Now, with IBM's contrib of JAX-WS code to the Axis2
project getting voted on as I type, it looks like there will be
two runtime targets.

Does this make more sense? If you think the web pages are causing
some confusion, we'll address it.


Back to the top