In this work we covered a variety of different aspects concerning the
application of object-oriented software components for data
acquisition systems. As a result, we have also a clearer view of which
directions should be further pursued. Although a working toolkit for
cross object invocations has been designed, most of the code has to be
handcrafted to produce stubs and skeletons for an interface. Therefore
we will now concentrate on a precompiler that accepts OMG/IDL
definitions and produces runnable code in C++. As we found out that
the presentation layer can be a major bottleneck, emphasis will be put
on this issue. A configurable scheme that allows different
presentation layer encoder and decoders to be plugged in, has already been
investigated[GO00]. What concerns the building of
applications using such software components, research has been started
on how to plug together different objects that use our toolkit. It is
especially interesting to realise a scheme for efficiently multicasting
a message to many objects. It became clear to us that support for
distributed object invocation and communication abstraction raises a
need for generic addressing. The merger of different network
technologies in a single system makes it necessary to provide a
uniform scheme for addressing. It has to allow peculiarities of
different protocols, must be scalable and shall be independent of any
existing special solution, like the Internet addresses. This will be
one of the topics for future research. Another aspect that is
currently under investigation is how the upcoming 
(Intelligent
IO)[SIG97] can be used to offload the burden of server side
dispatching from the machines main processors to the IO
controllers. This work at hardware/software boundary will be
challenging because a new and promising device
technology is used in an object-oriented middleware product,
that does not present software between operating system and
application, but between the hardware and the application. It should
eventually make it possible to have the abstraction of true
interobject communication on any kind of operating system using any
kind of communication technology. Finally we will refine our
software component perrformance model and evaluate all components that
are produced in the scope of our future research and development.