Horton

Re: Index-only scans vs. partially-retrievable indexes

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

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: pgsql-hackers@lists.postgresql.org
Date: 2022-01-02T23:22:23Z
Lists: pgsql-hackers

Attachments

I wrote:
> Unless somebody's got a better idea, I'll push forward with making
> this happen.

Here's a proposed patch for that.  I ended up reverting the code
changes of 4ace45677 in favor of an alternative I'd considered
previously, which is to mark the indextlist elements as resjunk
if they're non-retrievable, and then make setrefs.c deal with
not relying on those elements.  This avoids the EXPLAIN breakage
I showed, since now we still have the indextlist elements needed
to interpret the indexqual and indexorderby expressions.

0001 is what I propose to back-patch (modulo putting the new
IndexOnlyScan.recheckqual field at the end, in the back branches).

In addition, it seems to me that we can simplify check_index_only()
by reverting b5febc1d1's changes, because we've now been forced to
put in a non-half-baked solution for the problem it addressed.
That's 0002 attached.  I'd be inclined not to back-patch that though.

			regards, tom lane

Commits

  1. Handle mixed returnable and non-returnable columns better in IOS.

  2. Fix index-only scan plans, take 2.