Peter,

> I think the SRFI should be split into (srfi 282 parameter) and (srfi
> 282 environment), each exporting the appropriate procedure. Each is
> optional, and the user can check if one is supported using
>
>     (cond-expand
>       ((library (srfi 282 parameter)) ...)
>       (else ...))
>
> If one is supplied, then (predicate? x) being true means that the "it
> is an error" behavior doesn't occur.

Huh, that’s a genuinely good idea! Solves the type support problem too!
I think I’ll go down that route.

Thanks,
--
Artyom Bologov
https://aartaka.me