Am Di., 8. Sept. 2020 um 17:40 Uhr schrieb John Cowan <xxxxxx@ccil.org>:

> Ranges would be mutable if:
>
> 1) they might be the same type as some mutable type (in which case they could potentially, though not portably, be mutated using the procedures of that type), or
>
> 2) actual mutation procedures such as range-set! or range-map! were provided.
>
> But since neither of these things are true, ranges are immutable (as I keep saying) by construction.  And indeed, the only place they are called immutable is in the abstract, which is non-normative.
>
> (Again, frame-breaking operations like record and procedure inspection may allow direct mutation of a range.)

I think we agree 99%. The remaining 1% is that I don't make a
difference between 1) and 2). For me providing a (non-portable)
extended set of procedures (like range-set!) would be effectively the
same as extending (non-portably) existing procedures (like extending
vector-set! to ranges under the assumption that ranges may not
distinguishable from vector literals).

:)