Partial Mode in Aggregate Functions

On Wednesday, February 25, 2026, David Rowley <dgrowleyml(at)gmail(dot)com> wrote:
>
>
> As for the proposed patch, I'm strongly against it. If people are
> confused about what partial aggregation is, then let's modify the
> documentation to explain what it means.
>

Overall that helps a lot - given its application to partitioning, “parallel
mode” indeed is not a good choice here. However, I think It reads a bit too
DBA and custom function author oriented which, IMO, is too technically
precise and detailed for someone who just wants to use these functions and
understand how the “false” subset is limited. A bit shorter and more use
of colloquial terminology would be nice. I may give that a shot.

I was pondering whether Batch Mode as more commonly used term would be
sufficiently accurate here. But I’m not going to try hard to avoid Partial
Mode.

I do question whether the column is even particularly valuable to keep
though. I would rate function volatility and strictness much higher in the
list of things people want to know about our functions and we don’t
document that, at least not in table form. Unless we advise people to
write their queries differently based upon this optimization property if
they need the feature they will use the function regardless. Showing it
likely just creates noise in their mind for something they cannot control
anyway.

David J.