Hi Dow,
I don't think this one [locator] is quite there yet since it doesn't directly
address the question of how (a) namespace semantics and (b) the
associated routing system are involved in characterizing the name of an
attachment point as a locator (or not). I think some of this relation
is implied in the language above, but it is still fuzzy...
The definition of locator isn't supposed to involve the specific routing
system and it's not meant to specify or constrain the path in the
general case. Locators are supposed to define 'where' something is, not
'how to get there'.
Using the postal example,
123 High Street,
Bigtown,
Green County,
Outer Luvania
We would agree (I think) that if an object / person moved next door, and
still wanted to be reachable, it would need to use a different locator
(say, 125 High Street). However, what if larger sections of the
geography itself move - perhaps today Green County is beside Blue
County, but tomorrow it is next to Red County.
Well, it's then the job of routing to compute that path. Certainly, you
wouldn't want a situation where every node on the network would have to
recompute its locator every time there was a topology change.
Tony
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
