Prevent improper reordering of antijoins vs. outer joins.

An outer join appearing within the RHS of an antijoin can't commute with
the antijoin, but somehow I missed teaching make_outerjoininfo() about
that.  In Teodor Sigaev's recent trouble report, this manifests as a
"could not find RelOptInfo for given relids" error within eqjoinsel();
but I think silently wrong query results are possible too, if the planner
misorders the joins and doesn't happen to trigger any internal consistency
checks.  It's broken as far back as we had antijoins, so back-patch to all
supported branches.

diff --git a/src/backend/optimizer/plan/initsplan.c b/src/backend/optimizer/plan/initsplan.c
index 26441fc2d964b7dfca134485728cd64825532b6d..4a41e5746bb478b1712fccbf480cb82dcccc545c 100644 (file)
--- a/src/backend/optimizer/plan/initsplan.c
+++ b/src/backend/optimizer/plan/initsplan.c
@@ -694,9 +694,9 @@ make_outerjoininfo(PlannerInfo *root,
                 * For a lower OJ in our RHS, if our join condition does not use the
                 * lower join's RHS and the lower OJ's join condition is strict, we
                 * can interchange the ordering of the two OJs; otherwise we must add
-                * lower OJ's full syntactic relset to min_righthand.  Here, we must
-                * preserve ordering anyway if either the current join is a semijoin,
-                * or the lower OJ is either a semijoin or an antijoin.
+                * the lower OJ's full syntactic relset to min_righthand.  Also, we
+                * must preserve ordering anyway if either the current join or the
+                * lower OJ is either a semijoin or an antijoin.
                 *
                 * Here, we have to consider that "our join condition" includes any
                 * clauses that syntactically appeared above the lower OJ and below
@@ -713,6 +713,7 @@ make_outerjoininfo(PlannerInfo *root,
                {
                        if (bms_overlap(clause_relids, otherinfo->syn_righthand) ||
                                jointype == JOIN_SEMI ||
+                               jointype == JOIN_ANTI ||
                                otherinfo->jointype == JOIN_SEMI ||
                                otherinfo->jointype == JOIN_ANTI ||
                                !otherinfo->lhs_strict || otherinfo->delay_upper_joins)

diff --git a/src/test/regress/expected/join.out b/src/test/regress/expected/join.out
index b484620dba11ba0183708236f9a61c4e176e4477..97a6a6101530ea51540fa2e018ab8a6122e3dfd3 100644 (file)
--- a/src/test/regress/expected/join.out
+++ b/src/test/regress/expected/join.out
@@ -2277,6 +2277,48 @@ WHERE d.f1 IS NULL;
  9999
 (3 rows)
 
+--
+-- regression test for proper handling of outer joins within antijoins
+--
+create temp table tt4x(c1 int, c2 int, c3 int);
+explain (costs off)
+select * from tt4x t1
+where not exists (
+  select 1 from tt4x t2
+    left join tt4x t3 on t2.c3 = t3.c1
+    left join ( select t5.c1 as c1
+                from tt4x t4 left join tt4x t5 on t4.c2 = t5.c1
+              ) a1 on t3.c2 = a1.c1
+  where t1.c1 = t2.c2
+);
+                       QUERY PLAN                        
+---------------------------------------------------------
+ Hash Anti Join
+   Hash Cond: (t1.c1 = t2.c2)
+   ->  Seq Scan on tt4x t1
+   ->  Hash
+         ->  Merge Right Join
+               Merge Cond: (t5.c1 = t3.c2)
+               ->  Merge Join
+                     Merge Cond: (t4.c2 = t5.c1)
+                     ->  Sort
+                           Sort Key: t4.c2
+                           ->  Seq Scan on tt4x t4
+                     ->  Sort
+                           Sort Key: t5.c1
+                           ->  Seq Scan on tt4x t5
+               ->  Sort
+                     Sort Key: t3.c2
+                     ->  Merge Left Join
+                           Merge Cond: (t2.c3 = t3.c1)
+                           ->  Sort
+                                 Sort Key: t2.c3
+                                 ->  Seq Scan on tt4x t2
+                           ->  Sort
+                                 Sort Key: t3.c1
+                                 ->  Seq Scan on tt4x t3
+(24 rows)
+
 --
 -- regression test for problems of the sort depicted in bug #3494
 --

diff --git a/src/test/regress/sql/join.sql b/src/test/regress/sql/join.sql
index 39a701a9e02d013dff85c4a3b947bbb13a681556..fb6981f54d3c4cf9ee459e4c3bd47e82c0b75531 100644 (file)
--- a/src/test/regress/sql/join.sql
+++ b/src/test/regress/sql/join.sql
@@ -443,6 +443,23 @@ LEFT JOIN (
 ) AS d ON (a.f1 = d.f1)
 WHERE d.f1 IS NULL;
 
+--
+-- regression test for proper handling of outer joins within antijoins
+--
+
+create temp table tt4x(c1 int, c2 int, c3 int);
+
+explain (costs off)
+select * from tt4x t1
+where not exists (
+  select 1 from tt4x t2
+    left join tt4x t3 on t2.c3 = t3.c1
+    left join ( select t5.c1 as c1
+                from tt4x t4 left join tt4x t5 on t4.c2 = t5.c1
+              ) a1 on t3.c2 = a1.c1
+  where t1.c1 = t2.c2
+);
+
 --
 -- regression test for problems of the sort depicted in bug #3494
 --