Jean-Marc Valin wrote: > > Well, maybe we should have more than one GUI, or at least a GUI with a couple > "modes". There are many applications that require a different interface: > -a GUI equivalent to something like "ls -l | awk '{print $2}' | sed 's/aa/bb/' > | sort -u > somewhere" (a la Loci) > -a system to run complex distributed tasks (a la GMS) > -a signal processing framework a la Matlab/Simulink/Lab View (a la Overflow) > > I don't see how the same interface can handle all those correctly. Ahhh, but this is where we use XML to define GUIs! As I mentioned earlier, XML documents can be arranged according to the arrangement of nodes in a network. Then the XML can be translated into a Gtk interface using something like libglade. What I haven't brought up yet is that 'windows' can switch between views. At one moment, a window can show a network of nodes. The next moment, this window can be switched to show Gtk widgets arranged as I mention above. > I prefer workspace de document (which is too generic). I prefer Node though, a > month ago I had never heard the word Loci/Locus (sorry, engligh is my second > language). Loci/locus is actually a Latin word, meaning location :-) > I don't mind Link of Connector (link is shorter, bout that's all)... > pipe, maybe?.. Pipe may be confused with Unix pipes, which are local. Link is okay, and is reminiscent of HTML links. > As for Network... we have to be careful... I Overflow (though it's not > implemented exactly that way), you can see "Network" as a base class and > "subnet", "iterator" and "threadedIterator" (possibly more later) as subclasses > of Network. So, what are nodes a subclass of? Jeff -- +----------------------------------+ | J.W. Bizzaro | | | | http://bioinformatics.org/~jeff/ | | | | BIOINFORMATICS.ORG | | The Open Lab | | | | http://bioinformatics.org/ | +----------------------------------+