I would like to share some thoughts with you on future directions
of computing. Basically, I see the software industry relying
heavily on Component Object Modeling, as successfully marketed by
Microsoft. What this entails is the development of small modules
which are linked at run-time and behave as an interconnect
software module. Some of the greatest advantages of this approach
are that it allows a component update to be easily implemented
without having to update the entire executable. While I was
working at Intergraph on their CAD/CAM product SolidEdge, I saw
the advantages of this approach when development team embarked on
setting a standard called OLE for Design and Modeling
(www.dmac.org). SolidEdge concentrates on an assembly-centric
modeler which allows the updates of various components. SolidEdge
demonstrated the value of this approach by changing its
underlying boolean engine from ACIS to ParaSolids going from
release 3 to release 4. This became the start of a trend in the
CAD industry where not only SolidEdge, but its competitor,
SolidWorks, began to set standards which could be used by third
parties (which developed analysis tools) to interact with the CAD
modeler. Furthermore, users could extend the capabilities of the
modeler by writing Visual Basic code which would control the
modeler and perform automated tasks. One early example was to
illustrate motion analysis by showing how the part would
interfere with parts in an assembly. Microsoft supports the
branching of these standards into industry specific solutions. I
know that there are other COM standards in the banking industry.
Similarly, I believe that such standards can be set in the
biotechnology industry by far reaching visionaries. Since so much
progress has all ready been accomplished by the CAD industry, it
is possible to build on top of all ready made progress because
many of the ideas coincide. By publishing and exposing
interfaces, we could set the standard in the biotech industry
just as Microsoft has set the standard in document architecture.
One advantage OLE has over Java is that it is a binary interface
standard which uses fast platform specific code without having to
be byte code interpreted at run-time. As I learned during my
tenure at Intergraph, one difficulty with COM is to slice up the
problem into such an arrangement where the server lag does not
render the approach unfeasible.
Please let me know of what you think of my ideas.
Sincerely,