+1 to exploring this idea. A couple additional thoughts...
I can imagine real world use cases that would benefit from this beyond the
super pathological N-1 case. With dense terms, I can believe that the
disjunction would start to accumulate "blocks" of documents that all match
as the bitset gets
It sounds like a lot of complexity to handle an unusual edge case, but
... I guess this actually happened? Can you give any sense of the
end-user behavior that caused it?
On Fri, Nov 4, 2022 at 2:26 AM Patrick Zhai wrote:
>
> Hi Froh,
>
> The idea sounds reasonable to me, altho I wonder whether u