At the risk of making things even more confused...it seems we have at
least the following things that we are trying to talk about:
1. landscape - an object or system with a notion of distance
2. a point on a landscape
3. a name for a point on a landscape
4. an object that is attached to a point on the landscape
5. a name for an object that is attached to a point on a landscape
6. a map of a landscape describing which points are named what
7. a name for a section of a landscape (i.e., a collection of points
on a landscape)
In the current system (I think) we have a physical topology (landscape
A) and the IP topology (landscape B). There is an implicit map for
landscape B - i.e. for the names of points in landscape B. This
implicit map is captured in the proximity semantics of the IP address
- that is, by default, locations with names that are numerically close
to each other are assumed to be topologically close to each other in
the IP landscape. This same property allows us to name sections of
the IP landscape, where the section name is implicitly related to the
names of the points comprising that section.
There is an explicit map (the routing system state) that relates
locations in landscape B to locations in landscape A. This explicit
map (I think) also sometimes overrides the implicit map in landscape B.
If an object attached to landscape A with a name in landscape B
changes location in landscape A, either the explicit map relating A to
B must change, or the object's name in B must change to (or retain) a
name in B that (i) conforms to the implicit map for landscape B and
(ii) does not violate the current relation of A to B.
Ok - going to sleep...
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
