Re: Surprised by index choice for count(*)

On Tue, May 1, 2018 at 8:46 AM, Rob Sargent <robjsargent(at)gmail(dot)com> wrote:

> Should I be? I would have thought the pk would have been chosen v.
> function index?
> Indexes:
> "segment_pkey" PRIMARY KEY, btree (id)
> "useg" UNIQUE, btree (probandset_id, chrom, startbase, endbase)
> "fpv" btree (pv(events_less, events_equal, events_greater, 0))
> "segment_markerset_id_probandset_id_idx" btree (markerset_id,
> probandset_id)
>

​Haven't looked at the code but among those 4 indexes the one that was
chosen is the one that comes first alphabetically.​ Ideally it would avoid
multi-column indexes since they are larger; and I believe that a float is
smaller than a uuid so not only alphabetically but the fpv index wins on
size too.

That the index is functional is immaterial here.

David J.