Re: Start node (Re: [Gzz] PEG model_management--tjl)

[Benja Fallenstein]

Tuukka Hastrup wrote:
> On Fri, 4 Apr 2003, Alatalo Toni wrote:
>
> > > There can be other reflexive assertions in addition to fenfire:home, and
> > > we might need to get into them as well. In that sense, storing the graph
> > > URL or its node (which can be null) would be better than storing the view
> > > start node.
> >
> > trying to keep up with this too but not comprehending yet..
>
>
> When we create new nodes, most URIs will be our URNs. But if we want to 
> assert something about the graph as a whole, the subject URI is the URI of 
> the graph itself, and we need to know it somehow. 
>
> But there isn't necessarily such node in the graph.

This is wrong. There may not be assertions about the node in the graph, 
but the graph is a set of *connections between* nodes, not a set of 
nodes. A graph's URI is always a node in the graph (as are all other URIs).

> Neither is that node 
> necessarily the one we want to start our views at.

Definitely not. That's the reason it *refers to* the home node, instead 
of *being* the home node.

Currently, we pick the starting point randomly, as Toni rightly 
remembered. A better way if a graph doesn't have a home node would be to 
pick something that looks like a "root," algorithmically. Just have to 
figure out the algorithm :-)

- Benja