Re: Optimization idea: merging multiple EXISTS(...) with constraint-based join removal

    SELECT * FROM t1

    WHERE EXISTS(

        SELECT 1 FROM t2, t3, t4

        WHERE t2.id = t1.t2_id

        AND t3.id = t2.t3_id

        AND t4.id = t3.t4_id

        AND t4.val = 'XYZ'

    ) AND EXISTS(

        SELECT 1 FROM t2, t3, t5

        WHERE t2.id = t1.t2_id

        AND t3.id = t2.t3_id

        AND t5.id = t3.t5_id

        AND t5.val = 'Blablabla'

    ) AND EXISTS(

        SELECT 1 FROM t6

        WHERE t6.id = t1.t6_id

        AND t6.val = 'Hello'

    )

The resulting query is:

    SELECT * FROM t1

        WHERE EXISTS(

            SELECT 1 FROM t2 t2_a, t3 t3_a, t4 t4_a, t2 t2_b, t3 t3_b, t5, t6

            WHERE t2_a.id = t1.t2_id

            AND t3_a.id = t2_a.t3_id

            AND t4_a.id = t3_a.t4_id

            AND t4_a.val = 'XYZ'

            AND t2_b.id = t1.t2_id

            AND t3_b.id = t2_b.t3_id

            AND t5.id = t3_b.t5_id

            AND t5.val = 'Blablabla'

            AND t6.id = t1.t6_id

            AND t6.val = 'Hello'

        )

My questions are:

- Does PostgreSQL already supports this optimization ? Maybe it's not enabled in my case only?

- If yes, is my reasoning incorrect ? Can you point me my mistake ?

- Otherwise is there any plan to add this optimization to PostgreSQL ?