Dow Street wrote:
I think you can say there is a often a continuity convention in names for sections of geographic landscapes. That is, someplace in Green county is generally reachable from another place in Green country without having to exit Green county. This is, of course, not always the case (e.g. Alaska). There is also an implied hierarchy in the set of names for map sections in this address, e.g. Bigtown is "in" Green County. Based on these conventions, and given the name of the location, one might say you could derive a loose path from your present location in the landscape to the point named by this locator, even without an full / explicit map. "First get to Outer Luvania, then find Green County, etc.." I think this information constitutes some notion of "how to get there".
That's certainly one possible path, but it may well not be optimal. What you're really pointing out here is that in a hierarchical locator system, that following the path via aggregates itself becomes a candidate path, *if* you can tell where the aggregates are. However, to do that you again need the information held in the routing subsystem.
There are trivial examples where the candidate path implied by such routing are suboptimal, and where "cold-potato" routing is optimal. Imagine the case where all of the roads in Green County are very, very poorly maintained and full of potholes that make you drive slowly (i.e., high metric). The optimal path is to stay outside of Green County until you get to the point outside of the county that is closest to the destination.
However, we may not want to call it a "path", but more of a "plan for a path" that does not become an actual path until we add a road sign that says "Bigtown over there". If this (path) is what you are defining as information on "how to get there", then I think we are in sync conceptually.
Agreed. Tony _______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
