On 2/14/17 1:18 PM, Tom Lane wrote:
> I think this might be code-complete, modulo the question of whether we
> want an enabling GUC for it. I'm still concerned about the question
> of whether it adds more planning time than it's worth for most users.
> (Obviously it needs some regression test cases too, and a lot more
> real-world testing than I've given it.)
Don't we normally provide a GUC for this level of query manipulation? We
can always remove it later if evidence shows there's not sufficient
downside (perhaps after gaining the ability for the planner to do extra
work on queries that appear to be large enough to warrant it).
> One point that could use further review is whether the de-duplication
> algorithm is actually correct. I'm only about 95% convinced by the
> argument I wrote in planunionor.c's header comment.
Another reason for a GUC...
I'll put some thought into it and see if I can find any holes. Are you
only worried about the removal of "useless" rels or is there more?
--
Jim Nasby, Data Architect, Blue Treble Consulting, Austin TX
Experts in Analytics, Data Architecture and PostgreSQL
Data in Trouble? Get it in Treble! http://BlueTreble.com
855-TREBLE2 (855-873-2532)