version compare and partial orgls

[Mark S. Miller <mark>]

Date: Sat, 2 Sep 89 20:01:52 PDT
   From: tribble (Eric Dean Tribble)

   I strongly dislike linguistic abstractions that can't be represented or
   achieved.  Especially now that we agree that freezing doesn't affect
   sharing-informability, the notion of *true* identity is at best
   distracting.  We can't represent it, it adds no useful insights as an
   analogy, and I think it is sufficiently opposed to what is really
   possible in a distributed system that it prevents us from considering
   the issues of DataObject identity correctly in the distributed
   context.  I suppose the case isn't quite as strong as I'm making it,
   but....

As I said before, the notion of *true* identity is a useful rhetorical
device that may be more confusing than its worth.  It helps me think
about this, but I'm content to drop it.  It is sort of like the
misleading notion in mathematical economics that there is an
equilibrium state the economy is at or approaching.  In both cases one
is getting an orientation wrt the direction of a motion (towards more
identity information or towards greater economic efficiency) by
bogusly painting an end state that that motion is tending towards.
Considering the damage the equilibrium notion has had in economics, I
guess I agree it would be better to drop *true* identity.