Re: plan shape work

On Tue, Oct 7, 2025 at 4:37 AM Robert Haas <robertmhaas@gmail.com> wrote:
> I see that Richard's PoC last patch confused cfbot. Here's a new
> version of just the patch proposed for commit for CfBot testing.

Does it make sense to explicitly initialize glob->subplanNames in
standard_planner()?  I understand this might seem pointless since
makeNode() zeroes all fields by default, but subplanNames is currently
the only field in PlannerGlobal that isn't explicitly initialized.  I
previously committed a patch (2c0ed86d3) to ensure all PlannerGlobal
fields are explicitly initialized, and I'd prefer to maintain that
consistency.

I actually suggested the same in [1] (the last paragraph), but it
seems to have been overlooked.

[1] https://postgr.es/m/CAMbWs4-ysLvZiWp=w5=+noCMdX9FHFrrc0Wuk-TcUz1RDmEbkQ@mail.gmail.com

- Richard