Re: Polyphase merge is obsolete

On Mon, Feb 27, 2017 at 2:45 PM, Peter Geoghegan <pg(at)bowt(dot)ie> wrote:
> Since we have an awful lot of stuff in the last CF, and this patch
> doesn't seem particularly strategic, I've marked it "Returned with
> Feedback".

I noticed that this is in the upcoming CF 1 for v11. I'm signed up to review.

I'd like to point out that replacement selection is also obsolete,
which is something I brought up recently [1]. I don't actually have
any feature-driven reason to want to kill replacement selection - it's
just an annoyance at this point. I do think that RS is more deserving
of being killed than Polyphase merge, because it actually costs users
something to continue to support it. The replacement_sort_tuples GUC
particularly deserves to be removed.

It would be nice if killing RS was put in scope here. I'd appreciate
it, at least, since it would simplify the heap routines noticeably.
The original analysis that led to adding replacement_sort_tuples was
based on certain performance characteristics of merging that have
since changed by quite a bit, due to our work for v10.

[1] postgr.es/m/CAH2-WzmmNjG_K0R9nqYwMq3zjyJJK+hCbiZYNGhAy-Zyjs64GQ@mail.gmail.com
--
Peter Geoghegan