Re: Reduce TupleHashEntryData struct size by half

On Sun, 2025-01-12 at 14:54 +1300, David Rowley wrote:
> I wonder if there's some other better way of doing this. Would it be
> worth having some function like ExecCopySlotMinimalTuple() that
> accepts an additional parameter so that the palloc allocates N more
> bytes at the end?

Attached a new series that implements this idea.

I also added in the change to use the Bump allocator for the tablecxt.
In my simple test, the whole patch series reduces HashAgg memory usage
by more than 40% and increases speed by a few percent.

In your test, patch 0003 seemed to have a regression, but then 0007
made up for it (and maybe a hair faster). I investigated, but the
profile was a bit misleading so I'm not clear on why 0003 came out
slower. I can investigate further, but the primary purpose of this
patch series is reducing memory usage, so as long as the overall series
has no regression then I think it's fine.

Regards,
Jeff Davis