From: Tom Lane <tgl@sss.pgh.pa.us>
To: Tomas Vondra <tomas.vondra@2ndquadrant.com>
cc: Surafel Temesgen <surafel3000@gmail.com>,
        David Steele <david@pgmasters.net>,
        Michael Paquier <michael@paquier.xyz>,
        Robert Haas <robertmhaas@gmail.com>, andrew@tao11.riddles.org.uk,
        PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Subject: Re: FETCH FIRST clause WITH TIES option
In-reply-to: <20190403190257.44fxttkpo4yopp3l@development>
References: 
 <CALAY4q9Zd24E8OEBxOHfqQbp6u+xePXrFsnvM5xoy1pq1APUgw@mail.gmail.com>
 <c1430163-a5b0-2adb-c81c-29d1029cb8cf@2ndquadrant.com>
 <CALAY4q-+CKiFkgCL5VSK-9P8i5Z9uoQQh6ZEZLkjzsUT2mySsw@mail.gmail.com>
 <20190204052857.GP29064@paquier.xyz>
 <CALAY4q8vDyK4_jhFkXyWnJRcuWGgkwOzJTw-j2Dw0ddyTiZM4g@mail.gmail.com>
 <cd96a537-a302-48ef-b41b-2e247e156745@pgmasters.net>
 <CALAY4q8X-17+AFaAKZ+YX781b+pqjdU4S=z=ZUGi3QjyHpZ6hQ@mail.gmail.com>
 <20190329005648.GA1136@development> <20190331001446.GA10804@development>
 <CALAY4q86X1rCKyNE040tx1nXgH5BGeH4nxdtVbEdUuvVxEBdRw@mail.gmail.com>
 <20190403190257.44fxttkpo4yopp3l@development>
Comments: In-reply-to Tomas Vondra <tomas.vondra@2ndquadrant.com>
	message dated "Wed, 03 Apr 2019 21:02:57 +0200"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <32412.1554318485.1@sss.pgh.pa.us>
Date: Wed, 03 Apr 2019 15:08:05 -0400
Message-ID: <32413.1554318485@sss.pgh.pa.us>
Precedence: bulk

Tomas Vondra <tomas.vondra@2ndquadrant.com> writes:
> I've tried to fix the merge conflict (essentially by moving some of the
> code to adjust_limit_rows_costs(), but I'm wondering if the code added to
> create_limit_path is actually correct
> ...
> Firstly, this seriously needs some comment explaining why we do this.

I've not looked at this patch, but TBH I wonder why it is touching
planner rowcount estimation at all.  I find it doubtful either that
a correction for WITH TIES would be significant in most use-cases,
or that we could estimate it accurately if it was significant.
It certainly doesn't seem like something that needs to be messed
with in v1 of the feature.

			regards, tom lane