> > Do the lines really depict network connections or rather data "pipes" > > which may or may not be accross a network? > > (Even though I just submitted the abstract, these comments will be helpful for > writing the poster.) > > That's something that we have to work out. I imagined that they would all be > 'network-based' data pipes, if that makes any sense. So, if the GUI and the > analysis program both reside on the user's computer, the connection will just > loop back. URI: If you are using CORBA connections, you need not to worry about such things. At least ORBit uses local domain sockets ( == mem mapped ) if both end connections reside on the same machine. :) I think mico does this too, but I'm not sure. > This way, Loci never has to distinguish between a remote locus and a local one: > everything works via TCP/IP, Internet socket/port, or whatever. But you guys > are more familiar with this sort of thing. Maybe there's a major flaw in that > idea. Has anybody been checking out some multicast technlogy that could be useful to loci? Since loci should be general - purpose and network wide, you might end up at least sometime with a large number of loci that works with the same data. Using TCP/IP in such an application would be very inefficient. And unfortunately, the CORBA standard does not support multicasting. It would at least be very hard to inplement it. :( Cheers // Liss