> From: Keith Moore <moore@xxxxxxxxxx>
> what I mean is that a locator that means the same thing (refers to the
> same destination) no matter where you are in the net, is a lot easier
> to deal with than a locator with a meaning that changes (refers to
> different destinations) depending on where you are in the net.
Oh, I absolutely, completely, totally agree with you on that one!
Names which look like they are global, but sometimes aren't, are just asking
for trouble. This is doubly true when they don't inherently i) identify when
they aren't, and ii) include identification of the binding environment in
which they are valid, when they aren't.
The complexity and potential confusion inevitably associated with such names
means there must be a very good reason for using them in a system design -
and offhand, I can't think of a good enough reason to do so. (Before someone
says something, NAT is different - they weren't designed in, there.)
>>> locators are a lot easier to deal with if they're location-independent
>> Huh? Did you mean "identifiers are a lot easier to deal with if they're
>> location-independent"?
> I really was talking about locators, not identifiers.
Now that I understand what you actually meant, I'm not freaked out! However,
you phrased your point in a way that almost guaranteed confusion!
You didn't mean "locators are a lot easier to deal with if the name has
nothing to do with where the thing it names is", you meant "locators are a
lot easier to deal with if their meaning (i.e. the thing they are bound to)
is the same no matter where you are when you evaluate them".
Noel
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf