[kepler-dev] current state of saved specialized actors and repository

[Edward A Lee]

1 & 2 are already in place in Ptolemy II.  For 3, the big
issue is how to mediate multiple contributions.  A while back,
we experimented with a peer-to-peer approach for this...
Kind of Napster for actors. I.e., you maintain a library and
publish its availability, and others can discover it through
a publish-and subscribe.  The implementation used JXTA (or maybe it
was Jini, I forget...).

The downsides, as I recall are:

  1) Getting the P-to-P mechanism to work across firewalls.

  2) Making it work for people like me, who work principally
     on laptops, and are only sporadically connected.

I think these are solvable problems... Suppose I could locally
maintain a library of "my contributions", and when my machine
is on the net, it uses P-to-P to update a local cache of
"global contributions"...  I don't know how to fix the firewall
problem, though...  Perhaps the access mechanism has to be HTTP...

Edward


At 11:59 AM 9/21/2004 -0700, Bertram Ludaescher wrote:

>Hi:
>
>A quick question on the current state of the art in "saving
>specialized actors in the user lib. and/or repository".
>
>Here is the application example:
>
>1. A parameterizable actor say the web service actor or the command
>line actor is specialized by the user to invoke a specific web service
>or cmd line tool (say R, with a specific R script as parameter)
>
>2. After the desired specialization the user wants to save back this
>actor as a "first-class citizen" (composite) actor back into the user
>library, so that it can be used as if it were a native actor
>
>3. When confident that a useful new actor has been so created, the
>user may wish to publish it in the central Kepler actor repository
>(which of course can be dynamically browser/searched from Kepler)
>
>
>If we have (1-3) nicely in place, we can start to have Kepler users
>contribute new workflows as new composite actors. Having ways of
>semantically describing inputs, outputs, overall function, and adding
>other annotations will be important as well for the functionality of
>the Kepler actor repository...
>
>Bertram
>_______________________________________________
>kepler-dev mailing list
>kepler-dev at ecoinformatics.org
>http://www.ecoinformatics.org/mailman/listinfo/kepler-dev

------------
Edward A. Lee, Professor
518 Cory Hall, UC Berkeley, Berkeley, CA 94720
phone: 510-642-0455, fax: 510-642-2739
eal at eecs.Berkeley.EDU, http://ptolemy.eecs.berkeley.edu/~eal