> > Can't you simply use the reference implementation?  You can implement
> > make-mutable-parameter rather easily if you have a working
> > make-parameter.  You need to define a parameter that corresponds to
> > the dynamic environment, and then use that instead of the global
> > variable dynamic-env-local.
>
> Yes. I was thinking completely backward.
>
> A programmer could also put a mutable value inside a `make-parameter'
> to get the same expressiveness, right?

Yes.

> I find this line of reasoning so compelling that I now advocate the
> elimination of `make-mutable-parameter'.

You mean: parameters should be mutable according to the specified
semantics and "make-parameter" should create mutable parameter
objects, consequently "make-mutable-parameter" can be removed.

Since this was my original specification I'll gladly switch back to
it, unless there are strong objections.

> FWIW, I found your arguments in favor of `make-mutable-parameter'
> uncompelling: no particular use for it, but seems on the surface to be
> easier to implement, etc. These sort of arguments seem suspiciously
> like the ones that support dynamic scope over lexical scope.

No the "efficiency of implementation" (of thread creation and
continuation transfer) is only a secondary reason.  The primary reason
is that the semantics of parameter objects is closer to lexical
variables (the only difference is dynamic vs. lexical scope).  This
makes them easier to understand.  It is nice however that the
semantically "right thing" does not cause thread creation and
continuation transfer to be slow (maybe this is not a coincidence, but
I don't want to start philosophizing about programming language
design...).

Marc