Horton

Re: [PATCH] Push limit to sort through a subquery

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Douglas Doole <dougdoole@gmail.com>, Konstantin Knizhnik <k.knizhnik@postgrespro.ru>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2017-08-25T18:54:50Z
Lists: pgsql-hackers 
Robert Haas <robertmhaas@gmail.com> writes:
> On Fri, Aug 25, 2017 at 2:01 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> I looked through this a little, and feel uncomfortable with the division
>> of typedefs between execnodes.h and tuplesort.h.  I'm inclined to push
>> struct SortInstrumentation, and maybe also SharedSortInfo, into
>> tuplesort.h.

> I think moving SharedSortInfo into tuplesort.h would be a gross
> abstraction violation, but moving SortInstrumentation into tuplesort.h
> seems like a modest improvement.

Hmm, I'm not sure why SortInstrumentation belongs naturally to
tuplesort.h but putting an array of them there would be a "gross
abstraction violation".  Perhaps it would help to rename
struct SharedSortInfo to SortInstrumentationArray, and change its
field names to be less specific to the parallel-worker use case?

>> (BTW, would it make sense to number the workers from 1 not 0 in the
>> EXPLAIN printout?)

> ... So I'm in favor of leaving it alone; I don't think that 0-based
> indexing is such an obscure convention that it will flummox users.

OK, I'm not particularly set on that.

			regards, tom lane

Commits

  1. Propagate sort instrumentation from workers back to leader.

  2. Push tuple limits through Gather and Gather Merge.

  3. Code review for pushing LIMIT through subqueries.

  4. Make new regression test case parallel-safe, and improve its output.

  5. Push limit through subqueries to underlying sort, where possible.

  6. Move targetlist SRF handling from expression evaluation to new executor node.