Greg Hennessy <greg(dot)hennessy(at)gmail(dot)com> writes:
> On Thu, May 26, 2022 at 3:10 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Attaching a support function to a SQL-language function seems pretty
>> weird to me.
> Is there a way to set the selectivity of a SQL-language function?
I think it'd work if you prevented inlining, but doing so would defeat
most of the value of writing it as a SQL function as opposed to (say)
plpgsql. Can you do anything useful with attaching selectivity estimates
to the functions it references, instead?
regards, tom lane