dynamic binding

[Joachim Parrow <joachim@xxxxxxxxx>]

There appears to be some confusion of terminology concerning dynamic
binding/re-binding/scoping etc. In the original LISP-papers such terms
were synonomous and I don't think there is a consistent current usage. I
have no strong opinion but to focus the discussion  let me describe the
phenomenon I want to model.

Suppose we have a system of processes S = (P|Q|R) where a particular
name b is local. This could be because it accesses a local resource in
one of P,Q,R, or because the interpretation of something like "hostname"
is local to S. When either of these use b it means this local b. Now
this system is itself a subsystem of a larger context with more
susbsystems,  (S|T|U). Whenever something in T or U refers to b it will
*not* reach the local b in S. In the pi-calculus we would write
S = (nu b)(P|Q|R)
and this works fine as long as S is static in the sense that it cannot
gain new components. But in a higher-order calculus for example P can
input a process, call it V, sent from T. In the pi-calculus, V can then
not access b, unless b is first transmitted to V explicitly. What I am
loking for is an abstract way of describing that V here immediately
gains access to local names. This has some dramatic consequences, for
example these local names will no longer be alpha-convertible.

I agree that all this can be encoded in pi if we superimpose a protocol
where all local names are first transmitted to incoming processes. But
this is not a very abstract way to look at it (and also does not really
conform to what happens computationally in migrating code, where for
example V would not necessarily be trusted to conform to any protocol).
Certainly Luca's ambients is a more abstract solution but this issue
seems not much discussed.

This LISP-paradigm is a rather special case (where S is the caller and V
the called function).

Joachim


  
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The "models for mobility" mailing list     mailto:moca@xxxxxxxxxxxxxxx
 http://www-sop.inria.fr/mimosa/personnel/Davide.Sangiorgi/moca.html