Horton

Re: Optimization issue of branching UNION ALL

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

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andrey Lepikhov <a.lepikhov@postgrespro.ru>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2022-12-22T16:05:36Z
Lists: pgsql-hackers 
Andrey Lepikhov <a.lepikhov@postgrespro.ru> writes:
> Thanks, I have written the letter because of some doubts too. But only 
> one weak point I could imagine - if someday sql standard will be changed.

Yeah, if they ever decide that LATERAL should be allowed to reference a
previous sub-query of UNION ALL, that'd probably break this.  But it'd
break a lot of other code too, so I'm not going to worry about it.

I pushed the main fix to HEAD only, and the recursion checks to
all branches.

			regards, tom lane

Commits

  1. Avoid O(N^2) cost when pulling up lots of UNION ALL subqueries.

  2. Add some recursion and looping defenses in prepjointree.c.