On Wed, 29 Jul 1998, Ken Hornstein wrote:

> >On top of that, there was a math proof that was done several years ago 
> >that showed that the remote locking semantics could not actually work
> >right (which might explain why sun has released about fifty million
> >patched versions of lockd over the years).
> 
> You know, I've heard this before, but I could never track this down.
> Does anyone have a reference for this?  (Not that I'm calling anyone
> a liar, but it would be nice to be able to refer to this proof).

I don't know that much about remote locking semantics, but I have seen
some (what I think) related results.

Maurice Heller and Sergio Rajsbaum has written a nice paper "A Primer
on Algebraic Topology and Distributed Computing" that may be found
at

   http://www.cs.brown.edu/people/mph/

>From the abstract:

   Models and techniques borrowed from classical algebraic topology
   have recently yielded a varity on new lower bounds and impossibility
   results for distributed and concurrent computation. This paper 
   explains the basic concepts underlying this approach, and shows how
   they apply to a simple distributed problem.

You can find all kinds of cool results by following up the "related 
works" references...

   /Krister