perf: use one null-aware mark join for hashable IN subqueries in projections - #25338

Open

adriangb wants to merge 8 commits into

apache:mainfrom

pydantic:in-exists-subquery-projection

Commits on Sep 18, 2026

test: keep an un-decorrelated IN subquery in the extracted-alias regression test

adriangb
and
claude
committed
95c1181
perf: use one null-aware mark join for hashable IN subqueries in projections

committed
7aae42b
test: cover mark-join plan shapes and NULL semantics for projected IN subqueries

adriangb
and
claude
committed
75e22e9
fix: derive mark-join null-awareness from the join key expressions

adriangb
and
claude
committed
0f4fa7e
fix: keep the column null test for a correlated NOT IN with two keys

adriangb
and
claude
committed
a30d591
fix: give `join_keys_may_be_null` its new argument list at the constant `IN` call

adriangb
and
claude
committed
7679b23
test: regenerate the stale plan of a projected `IN` with a non-equality correlation

adriangb
and
claude
committed
e501c78
fix: keep the column null test for a `NOT IN` anti join with a residual filter

adriangb
and
claude
committed
3af8737