> But that was exactly the point of the original example: T1 was NOT in
> an explicit critical section, when it was killed. As I understand the
> example uses mutexes specifically for the purpose _other_ than to
> introduce a critical section.

Actually I came up with the mailbox example to show how mutexes could
be used to implement other things than critical sections.  As you
point out, this is a questionable example because it doesn't support
thread termination very well (but in a context that does not forcibly
terminate threads it would be a fine use of mutexes).  The use of
mutex-lock-anonymously! in my code was to prevent "abandoned mutex"
exceptions upon **normal** thread termination followed by a put! or
get!.

> It appears that you have already given a solution to the problem.
> In a message
> 	http://srfi.schemers.org/srfi-18/mail-archive/msg00038.html
> you showed an example that uses a mutex just to guard a critical
> section. If that example is modified as

I agree that the condition-variable approach is viable, and the way to
go to implement the mailbox in the presence of forced thread
termination.  My comment was simply that I did not see how the problem
could be solved with mutexes alone given the current semantics.

Marc