Re: Refactoring dynamic roots

[Kevin Ryde]

Andy Wingo <address@hidden> writes:
>
> Have you seen http://srfi.schemers.org/srfi-39/srfi-39.html -- Parameter
> objects? As long as we're redesigning, no harm in using a standard
> interface to the problem.

I added some ugly code to make srfi-39.scm do what the spec says for
current-input-port etc.  (Actually, maybe current-load-port should be
added there too.)

Making them fluids would be easier.  Users of the ports in libguile
could either call scm_fluid_ref, or have a guarantee there's always 4
entries in the root->fluids vector and fetch directly.