Hi Frank,
Generally, I think most everyone working on Kepler eventually realized
that an OSGi-ified Kepler/Ptolemy was the best possible end result for
modularization of the system. Unfortunately that realization happened
too late and the feeling was that we would not be able to achieve that
goal with the existing resources in a consistent and backward compatible
manner. The system that was produced meets most of the goals of the
original Kepler Core proposal.
Christopher! Glad to hear you have some funding to continue OSGi work.
Let me know if any more becomes available :) I have no doubt we could
find a good technical solution within the OSGi framework that maintains
an "open system". Reflection is such a nasty business...
Aaron Schultz
On 5/10/2011 10:46 AM, Christopher Brooks wrote:
Hi Frank,
My understanding is that some research was done and it was
found not to work out for Kepler.
Try using google to search for
site:http://lists.nceas.ucsb.edu osgi
Some good things did come out of that effort though. One thing
is that I cleaned up a bunch of unneeded dependencies in the
ptolemy tree.
My current understanding is that a big issue is with how MoMLParser
instantiates actor classes by using reflection. Apparently there
is an issue with OSGi where OSGi needs to know in advance what
classes are in the classpath. The problem is that MoMLParser does not
know in advance what classes will be instantiated. I don't find closed
systems to be that compelling for research. A closed system is useful
for deployment. Here, closed means that only a predefined set of classes
will *ever* be loaded.
I'm a little foggy on the details, but I believe Eclipse has a buddy
annotation that could help.
In other news, I have funding to do some of the OSGi work, but
it is not scheduled until next year. I think I blocked out about
6 months at 30% time.
In other, other news, I noticed that Taverna is in the process
of releasing an OSGi, see
http://www.mygrid.org.uk/dev/issues/browse/TOSGI
_Christopher
On 5/9/11 6:34 PM, Frank White wrote:
Hello All:
What ever came of the Kepler OSGI initiative? I see there were some
OSGI investigations back in 2008, but I can't seem to find anything that
documents a final decision. Appreciate any insights you might provide.
Thanks!
Frank
_______________________________________________
Kepler-dev mailing list
[email protected]
http://lists.nceas.ucsb.edu/kepler/mailman/listinfo/kepler-dev
