Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> So should `cl-remove-duplicates' be a 2-in-1 (lists for small input,
>
> Don't know/don't care about cl-remove-duplicates, but for delete-dups, I'd
> welcome a patch which switches to a hash-table after the Nth element.

I attach the patch. I did a bunch of `benchmark-run' and it seems that
100 elements is the breaking point.

> PS: BTW, helm-fast-remove-dups's docstring says:
>
>     This is same as `remove-duplicates' but with memoisation.
>
>   but it doesn't actually use memoization, AFAICT.

It does in a way, for an optimized-out "is-element-in-collection"
function.

Oleh