Request for comments: Peer-to-peer applications on the Internet.

I Find Karma (adam@cs.caltech.edu)
Mon, 16 Jun 97 18:04:58 PDT


Because I crave punishment, and because Ron has been so helpful with his
comments up until now, I've decided to expose three papers submitted to
various places in case anyone wants to offer constructive comments to
make them better.

The latest is

http://www.cs.caltech.edu/~adam/PAPERS/ndo/

which is about developing a peer-to-peer application running on the
Internet (in this case, a concurrent editing program) using the
infosphere Java libraries. This paper was written for Dr. Dobbs Journal.

Ron provided us with some useful comments on

http://www.cs.caltech.edu/~adam/jedi/paper/

which Jonathan is currently incorporating into that paper. JEDI is
about providing a dynamic method invocation layer in Java (something
which Ron disagrees with in principle since it provides programmers with
lotsa ways to shoot themselves in the foot, but which we believe is good
for smart programmers because it's both more efficient and allows the
plugging of new components into a system at run-time). This paper was
written for Douglas Schmidt's track on client-server computing for
HICSS-98 in January:

http://www.cs.wustl.edu/~schmidt/HICSS31/

Also, we have the paper

http://www.cs.caltech.edu/~adam/PAPERS/iscope/

which was submitted for consideration to the first International
Scientific Computing in Object-Oriented Parallel Environments (ISCOPE)
Conference will be held in Marina del Rey, California, in December.
This paper is about composing different distributed resource managers
(which may have different resource management policies) to create a
single, distributed resource manager. The call for papers for ISCOPE is
at:

http://www.acl.lanl.gov/iscope

We'd appreciate any constructive comments. Thanks!

As for me, I'm on to bigger things, like writing up my experiences with
Berna in predicting the performance for some supercomputing applications
that we did mucho months ago...

----
adam@cs.caltech.edu

An eigenvalue is a unique scalar which when it multiplies an
eigenvector, produces a resultant vector equivalent to the operator of
the eigen-equation applied to the same eigenvector. But you already
knew that.
-- Ron Resnick