Can the "LIMIT" statement be included as one of arguments when doing scan operation?

[nkg114mc]

No description provided.

[ankane]

Hi @nkg114mc, index scans don't know the limit, as they don't know which index tuples are visible to the current transaction or will be filtered by non-index conditions in the query (which needs to happen before the limit is applied).

I did some experiments a while back with setting a bound on the tuple sort (bound branch) as well as using a pairing heap (pairingheap and stages branches), but from what I recall, the performance difference didn't justify the complexity. However, someone is welcome to submit a PR (with benchmarks) with a better approach.