Re: [lisp] Adoption of draft-chiappa-lisp-architecture-01 and draft-chiappa-lisp-introduction-01

[[e-mail] (Noel Chiappa)]

{Sorry I've been silent for a while - been taking a break.}

    > From: "Templin, Fred L" <[e-mail]>

    > I have long maintained that what LISP is calling "EID" is not really an
    > identifier

We've been around and around on this many times, and while the first few did I
think introduce some useful light, I think we're probably past the diminishing
returns at this point?

Believe it or not, this is a point on which I do have some sympathy: some
people may recognize one of my favourite quotations, which I have used in a
number of places:

  "I am far from thinking that nomenclature is a remedy for every defect in
  art or science: still I cannot but feel that confusion of terms generally
  springs from, and always leads to, confusion of ideas."

	-- John Louis Petit, "Architectural Studies in France", 1854

Was 'EID' the best term to use? Perhaps not (although the difficulty in
introducing a new terms should not be ignored).


I am moved to mark the irony that while people continually complain over this
re-use of the term 'EID', I have yet to hear almost _anyone_ (other than me)
complain about the re-use of the term 'locator', which was clearly defined to
mean 'location sensitive name _which does not necessarily appear in every
packet_' (see Nimrod WG archives:

  http://mercury.lcs.mit.edu/~jnc/tech/nimrod/1993Sep-Dec.txt

where we found that we needed a new word as we found that people couldn't
free their minds from the assumption that an 'address' was something which
_had_ to appear in every packet - we had at that point yet to grasp that
their was an equal difficulty in people being sensitive to the fact that
'addresses' a la IPvN embodied both location _and_ identity - which is yet
_another_ example of terminology being warped).

So the term 'locator' is continually mis-used in the IETF (see, e.g. RFC-2373
"IPv6 Addressing Architecture", RFC-2956 "Overview of 1999 IAB Network Layer
Workshop"), but somehow nobody has a similar-sized problem with that. Why is
that?

Which is not to say, of course, that one wrong excuses another; multiple
definitions can indeed, as Petit observes, cause confusion.  But I am somewhat
peeved at what I see as disparate treatment in the two cases.


    > if the node has multiple independent (virtual) interfaces to which LISP
    > EIDs must be assigned, it is not possible to say that only one of them
    > is the "identity" of the endpoint

Suppose I had an architecture with 'true' EIDS. Suppose further that I
assigned a single node multiple 'true' EIDs. Which one is the 'identity'
of the node?

	Noel