I think that allowing verbatim method names and before/after/around
methods in mutex sets will aloow for a quite fine level
of concurrency control (together with the method managers).

Regarding the P =? NP problem:
what is the complexity of deciding whether a conjunctive normal form
of the propositional calculus with at most two literals per clause
is satisfiable?

-- Karl

From ghulten@ccs.neu.edu Wed Jun  4 12:43:49 1997
Subject: Re: coordination objects 
From: Geoff Hulten <ghulten@ccs.neu.edu>

>I'm not clear on what you mean by: 
>> It looks like that synchronizing on traversals would lock too much.
>> We would like to refer to before/after/around methods in
>> mutex sets.
>
From Johan:
>But my brain is pretty much in P =? NP mode right now. I'll think some
>more about it tonight. In the meanwhile, do you have a concrete
>example where locking on traversals is no good, but verbatim methods
>is better? I'll then try to come up wiht counter examples, and so
>forth.

Hrm...

I think that syncing on the combination of a visitor and traversal is more
powerful than syncing on a traversal alone.  Doug has pointed out a couple good
justifications for this (reuse and the fact that traversals don't actually do
anything by themselves) and I agree with him.

One of the reasons we decided to sync on verbatim methods is that visitors and
traversals are combined within them, making them a kind of clumsy way to talk
about this combination.  I think that this solution will probably make it
difficult to write adaptive programs that take maximum advantage of
concurrency, but think it's a better first step than just syncing on
traversals.

-Geoff