Fix another semijoin-ordering bug. We already knew that we couldn't

tglsfdc · tglsfdc · commit b2c51e6eba1d · 2009-07-21T02:02:44.000Z
reorder a semijoin into or out of the righthand side of another semijoin,
but actually it doesn't work to reorder it into or out of the righthand
side of a left or antijoin, either.  Per bug #4906 from Mathieu Fenniak.

This was sloppy thinking on my part.  This identity does work:

	( A left join B on (Pab) ) semijoin C on (Pac)
==
	( A semijoin C on (Pac) ) left join B on (Pab)

but I failed to see that that doesn't mean this does:

	( A left join B on (Pab) ) semijoin C on (Pbc)
!=
	A left join ( B semijoin C on (Pbc) ) on (Pab)
diff --git a/src/backend/optimizer/README b/src/backend/optimizer/README
@@ -1,4 +1,4 @@
-$PostgreSQL: pgsql/src/backend/optimizer/README,v 1.49 2009/02/27 22:41:37 tgl Exp $
+$PostgreSQL: pgsql/src/backend/optimizer/README,v 1.50 2009/07/21 02:02:44 tgl Exp $
 
 Optimizer
 =========
@@ -214,10 +214,10 @@ out of the nullable side of an outer join:
 	!= (A leftjoin B on (Pab)) join C on (Pbc)
 
 SEMI joins work a little bit differently.  A semijoin can be reassociated
-into or out of the lefthand side of another semijoin, but not into or out
-of the righthand side.  Likewise, an inner join, left join, or antijoin
-can be reassociated into or out of the lefthand side of a semijoin, but
-not into or out of the righthand side.
+into or out of the lefthand side of another semijoin, left join, or
+antijoin, but not into or out of the righthand side.  Likewise, an inner
+join, left join, or antijoin can be reassociated into or out of the
+lefthand side of a semijoin, but not into or out of the righthand side.
 
 ANTI joins work approximately like LEFT joins, except that identity 3
 fails if the join to C is an antijoin (even if Pbc is strict, and in
diff --git a/src/backend/optimizer/plan/initsplan.c b/src/backend/optimizer/plan/initsplan.c
@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/optimizer/plan/initsplan.c,v 1.154 2009/06/11 14:48:59 momjian Exp $
+ *	  $PostgreSQL: pgsql/src/backend/optimizer/plan/initsplan.c,v 1.155 2009/07/21 02:02:44 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -630,16 +630,16 @@ make_outerjoininfo(PlannerInfo *root,
 		 * min_lefthand + min_righthand.  This is because there might be other
 		 * OJs below this one that this one can commute with, but we cannot
 		 * commute with them if we don't with this one.)  Also, if the current
-		 * join is an antijoin, we must preserve ordering regardless of
-		 * strictness.
+		 * join is a semijoin or antijoin, we must preserve ordering
+		 * regardless of strictness.
 		 *
 		 * Note: I believe we have to insist on being strict for at least one
 		 * rel in the lower OJ's min_righthand, not its whole syn_righthand.
 		 */
 		if (bms_overlap(left_rels, otherinfo->syn_righthand))
 		{
 			if (bms_overlap(clause_relids, otherinfo->syn_righthand) &&
-				(jointype == JOIN_ANTI ||
+				(jointype == JOIN_SEMI || jointype == JOIN_ANTI ||
 				 !bms_overlap(strict_relids, otherinfo->min_righthand)))
 			{
 				min_lefthand = bms_add_members(min_lefthand,
@@ -655,7 +655,7 @@ make_outerjoininfo(PlannerInfo *root,
 		 * 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 an antijoin.
+		 * 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
@@ -672,6 +672,7 @@ make_outerjoininfo(PlannerInfo *root,
 		{
 			if (bms_overlap(clause_relids, otherinfo->syn_righthand) ||
 				jointype == JOIN_SEMI ||
+				otherinfo->jointype == JOIN_SEMI ||
 				otherinfo->jointype == JOIN_ANTI ||
 				!otherinfo->lhs_strict || otherinfo->delay_upper_joins)
 			{

Original file line number	Diff line number	Diff line change
`@@ -8,7 +8,7 @@`
`8`	`8`	`*`
`9`	`9`	`*`
`10`	`10`	`* IDENTIFICATION`
`11`		`- * $PostgreSQL: pgsql/src/backend/optimizer/plan/initsplan.c,v 1.154 2009/06/11 14:48:59 momjian Exp $`
	`11`	`+ * $PostgreSQL: pgsql/src/backend/optimizer/plan/initsplan.c,v 1.155 2009/07/21 02:02:44 tgl Exp $`
`12`	`12`	`*`
`13`	`13`	`*-------------------------------------------------------------------------`
`14`	`14`	`*/`
`@@ -630,16 +630,16 @@ make_outerjoininfo(PlannerInfo *root,`
`630`	`630`	`* min_lefthand + min_righthand. This is because there might be other`
`631`	`631`	`* OJs below this one that this one can commute with, but we cannot`
`632`	`632`	`* commute with them if we don't with this one.) Also, if the current`
`633`		`- * join is an antijoin, we must preserve ordering regardless of`
`634`		`- * strictness.`
	`633`	`+ * join is a semijoin or antijoin, we must preserve ordering`
	`634`	`+ * regardless of strictness.`
`635`	`635`	`*`
`636`	`636`	`* Note: I believe we have to insist on being strict for at least one`
`637`	`637`	`* rel in the lower OJ's min_righthand, not its whole syn_righthand.`
`638`	`638`	`*/`
`639`	`639`	`if (bms_overlap(left_rels, otherinfo->syn_righthand))`
`640`	`640`	`{`
`641`	`641`	`if (bms_overlap(clause_relids, otherinfo->syn_righthand) &&`
`642`		`- (jointype == JOIN_ANTI \|\|`
	`642`	`+ (jointype == JOIN_SEMI \|\| jointype == JOIN_ANTI \|\|`
`643`	`643`	`!bms_overlap(strict_relids, otherinfo->min_righthand)))`
`644`	`644`	`{`
`645`	`645`	`min_lefthand = bms_add_members(min_lefthand,`
`@@ -655,7 +655,7 @@ make_outerjoininfo(PlannerInfo *root,`
`655`	`655`	`* can interchange the ordering of the two OJs; otherwise we must add`
`656`	`656`	`* lower OJ's full syntactic relset to min_righthand. Here, we must`
`657`	`657`	`* preserve ordering anyway if either the current join is a semijoin,`
`658`		`- * or the lower OJ is an antijoin.`
	`658`	`+ * or the lower OJ is either a semijoin or an antijoin.`
`659`	`659`	`*`
`660`	`660`	`* Here, we have to consider that "our join condition" includes any`
`661`	`661`	`* clauses that syntactically appeared above the lower OJ and below`
`@@ -672,6 +672,7 @@ make_outerjoininfo(PlannerInfo *root,`
`672`	`672`	`{`
`673`	`673`	`if (bms_overlap(clause_relids, otherinfo->syn_righthand) \|\|`
`674`	`674`	`jointype == JOIN_SEMI \|\|`
	`675`	`+ otherinfo->jointype == JOIN_SEMI \|\|`
`675`	`676`	`otherinfo->jointype == JOIN_ANTI \|\|`
`676`	`677`	`!otherinfo->lhs_strict \|\| otherinfo->delay_upper_joins)`
`677`	`678`	`{`