[Pipet Devel] Re: Ideas for network distributed objects

[Jean-Marc Valin]

> very intersesting post from Jean,
> 
> >
> > A document is a file and can contain many networks (it's the equivalent of a .c)
> 
> > A network if the C equivalent of a function. Each document must contain a MAIN
> > network.
> 
> Logical.

The only difference with a C program is that if you need many .n (that's how we
name the overflow files), you need one MAIN in each, while in C you only need
one main. Only the MAIN network is accessible from outside.

> 
> >
> > NetOutput means which node (and which output of that node) will be used as the
> > output of the whole network. The same applies to NetInput and NetCondition.
> > Questions?
> >
> 
> Can I find a overview somewhere in the Overflow code?

What kind of overview?

> I've been looking in it, gives me the impression every node type is a separate c
> source.
> But the structure is clear, and I like it :)

Yes, all node is in a separate .cc (If it starts with the NODE_INFO macro, it's
a node). Actually, all directories, except data-flow and vflow only contain
nodes (the two ohers contain nodes, but also other stuff). Lots of the node even
have the class definition in the .cc, since the code never explicitly uses that
class (the NODE_INFO simply puts a Factory for that class in a global
"database"). The only nodes that need to have the class in a .h are the ones
that are inherited. For example, we have a Probe node that's a bit like a
debugger (Next, Stop, Continue) and we have a TextProbe and PlotProbe node that
add a display (text of plot) to the Probe.

	Jean-Marc

-- 
Jean-Marc Valin
Universite de Sherbrooke - Genie Electrique
valj01 at gel.usherb.ca