On reflection I agree, so I withdraw the suggestion.

On Mon, Apr 13, 2026 at 5:19 PM Peter McGoron <xxxxxx@mcgoron.com> wrote:
>
> On 4/13/26 16:07, John Cowan wrote:
> > I think I brought this up before, a long time ago, but what about
> > allowing a hash table as a storage class, for highly sparse arrays?
>
> Is there anything in the definition of make-storage-class that forbids
> this? Otherwise the end user could configure it themselves. That would
> solve the "number of flavors of hash tables" problem.
>
> I think such a situation would be somewhat niche, and would require
> design tradeoffs that a default design wouldn't be able to accommodate.
> For instance, if the default hash tables are OK with the copy you
> described, or if a specialized ordered hash table would be better.
>
> -- Peter McGoron
>