Yes, we are heading our way to utilize and support various distributed computing environment in Ptolemy II. In my oppinion, there are two directions, and we are doing experiments on both of them. 1) Passing messaged through the network. This gives the possibilities that Ptolemy models can get information from the network, and place information onto the network. The approach we are exploring is the persistent spaces model (like those in Linda, JavaSpaces, and T Spaces). There will be domain polymorphic publishers and subscribers that know (or discover) an information space. Publishers put tokens to the space, and subscribers get information from the space. A subscriber can register to the space that it interests on a certain kinds of token, and the space can notify the subscriber when the token is available. We are experimenting JINI/JavaSpaces implementation of it now. 2) Making components available through the network. The idea is that any Ptolemy component can be places onto the network. An application can discover what components are available, check that they fit for its purpose, and load the component (or run it remotely). This is more like a EJB model of encapsulating components. But the EJB model certainly lack of models of computation in it. We have done some experiment using the CORBA object model. We define a component as a service and execute it remotely. Java-RMI is another choice of the underlying mechanism, but I think CORBA is more flexible. The work we have done so far are all on the technology enabling phase. We haven't seriously consider the performance yet. But from what I have seen from the CORBA example, running CORBA (Java-IDL) on NT has a big performance pitfall. I haven't tried other ORBs. But I think the commercial ones have better performance. I am glad you are interested in this issue. If you have any comments and suggestions, please let us know. For the Matlab questions, please refer to Christopher's mail below. Regards, Jie > > > This appeared on the comp.soft-sys.ptolemy newsgroup. > > If you could give a response about your Corba work, then that would be > great. > > You could basically ignore the matlab question, or refer the > poster to Brian Evans' and Edward's posts about Matlab and DE at > http://www.egroups.com/group/ptolemy-hackers/1404.html? > http://www.egroups.com/group/ptolemy-hackers/1403.html? > > -Christopher > ------- Forwarded Message > > > From: "Sylvain D'HOINE" <[EMAIL PROTECTED]> > Subject: Ptolemy II in a distributed environment - Performances > Date: Thu, 9 Mar 2000 23:32:43 +0100 > > Has anybody developped a distributed application using Ptolemy II ? > What did you use ? CORBA, EJB, Java RMI, JINI ... ? What about the > performance ? > > Also, has anybody made some performance comparison between Ptolemy II and > Matlab + Simulink ? > > Regards, > > Sylvain > > > > ------- End of Forwarded Message > > > _______________________________________________ > Ptolemy maillist - [EMAIL PROTECTED] > http://www.gigascale.org/ptolemy/listinfo/ptolemy ---------------------------------------------------------------------------- Posted to the ptolemy-hackers mailing list. Please send administrative mail for this list to: [EMAIL PROTECTED]
