Create a new 'MultiExecProcNode' call API for plan nodes that don't

return just a single tuple at a time.  Currently the only such node
type is Hash, but I expect we will soon have indexscans that can return
tuple bitmaps.  A side benefit is that EXPLAIN ANALYZE now shows the
correct tuple count for a Hash node.

4 files changed, 105 insertions, 27 deletions

diff --git a/src/backend/executor/execProcnode.c b/src/backend/executor/execProcnode.c
index ff8caf16f01..555668e7799 100644
--- a/src/backend/executor/execProcnode.c
+++ b/src/backend/executor/execProcnode.c
@@ -12,7 +12,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/executor/execProcnode.c,v 1.48 2005/04/06 20:13:49 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/executor/execProcnode.c,v 1.49 2005/04/16 20:07:35 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -375,6 +375,50 @@ ExecProcNode(PlanState *node)
 	return result;
 }
 
+
+/* ----------------------------------------------------------------
+ *		MultiExecProcNode
+ *
+ *		Execute a node that doesn't return individual tuples
+ *		(it might return a hashtable, bitmap, etc).  Caller should
+ *		check it got back the expected kind of Node.
+ *
+ * This has essentially the same responsibilities as ExecProcNode,
+ * but it does not do InstrStartNode/InstrStopNode (mainly because
+ * it can't tell how many returned tuples to count).  Each per-node
+ * function must provide its own instrumentation support.
+ * ----------------------------------------------------------------
+ */
+Node *
+MultiExecProcNode(PlanState *node)
+{
+	Node *result;
+
+	CHECK_FOR_INTERRUPTS();
+
+	if (node->chgParam != NULL) /* something changed */
+		ExecReScan(node, NULL); /* let ReScan handle this */
+
+	switch (nodeTag(node))
+	{
+		/*
+		 * Only node types that actually support multiexec will be listed
+		 */
+
+		case T_HashState:
+			result = MultiExecHash((HashState *) node);
+			break;
+
+		default:
+			elog(ERROR, "unrecognized node type: %d", (int) nodeTag(node));
+			result = NULL;
+			break;
+	}
+
+	return result;
+}
+
+
 /*
  * ExecCountSlotsNode - count up the number of tuple table slots needed
  *
diff --git a/src/backend/executor/instrument.c b/src/backend/executor/instrument.c
index 52d9de4f4ac..c5b4a252d61 100644
--- a/src/backend/executor/instrument.c
+++ b/src/backend/executor/instrument.c
@@ -7,7 +7,7 @@
  * Copyright (c) 2001-2005, PostgreSQL Global Development Group
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/executor/instrument.c,v 1.11 2005/03/25 21:57:58 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/executor/instrument.c,v 1.12 2005/04/16 20:07:35 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -33,13 +33,10 @@ InstrAlloc(int n)
 void
 InstrStartNode(Instrumentation *instr)
 {
-	if (!instr)
-		return;
-
-	if (!INSTR_TIME_IS_ZERO(instr->starttime))
-		elog(DEBUG2, "InstrStartNode called twice in a row");
-	else
+	if (INSTR_TIME_IS_ZERO(instr->starttime))
 		INSTR_TIME_SET_CURRENT(instr->starttime);
+	else
+		elog(DEBUG2, "InstrStartNode called twice in a row");
 }
 
 /* Exit from a plan node */
@@ -48,12 +45,13 @@ InstrStopNode(Instrumentation *instr, bool returnedTuple)
 {
 	instr_time endtime;
 
-	if (!instr)
-		return;
+	/* count the returned tuples */
+	if (returnedTuple)
+		instr->tuplecount += 1;
 
 	if (INSTR_TIME_IS_ZERO(instr->starttime))
 	{
-		elog(DEBUG2, "InstrStopNode without start");
+		elog(DEBUG2, "InstrStopNode called without start");
 		return;
 	}
 
@@ -86,9 +84,17 @@ InstrStopNode(Instrumentation *instr, bool returnedTuple)
 		instr->running = true;
 		instr->firsttuple = INSTR_TIME_GET_DOUBLE(instr->counter);
 	}
+}
 
-	if (returnedTuple)
-		instr->tuplecount += 1;
+/* As above, but count multiple tuples returned at once */
+void
+InstrStopNodeMulti(Instrumentation *instr, double nTuples)
+{
+	/* count the returned tuples */
+	instr->tuplecount += nTuples;
+
+	/* delegate the rest */
+	InstrStopNode(instr, false);
 }
 
 /* Finish a run cycle for a plan node */
@@ -97,14 +103,14 @@ InstrEndLoop(Instrumentation *instr)
 {
 	double		totaltime;
 
-	if (!instr)
-		return;
-
 	/* Skip if nothing has happened, or already shut down */
 	if (!instr->running)
 		return;
 
-	/* Accumulate statistics */
+	if (!INSTR_TIME_IS_ZERO(instr->starttime))
+		elog(DEBUG2, "InstrEndLoop called on running node");
+
+	/* Accumulate per-cycle statistics into totals */
 	totaltime = INSTR_TIME_GET_DOUBLE(instr->counter);
 
 	instr->startup += instr->firsttuple;
diff --git a/src/backend/executor/nodeHash.c b/src/backend/executor/nodeHash.c
index 97e6738bd32..c304d930a28 100644
--- a/src/backend/executor/nodeHash.c
+++ b/src/backend/executor/nodeHash.c
@@ -8,13 +8,13 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/executor/nodeHash.c,v 1.92 2005/03/31 02:02:52 neilc Exp $
+ *	  $PostgreSQL: pgsql/src/backend/executor/nodeHash.c,v 1.93 2005/04/16 20:07:35 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
 /*
  * INTERFACE ROUTINES
- *		ExecHash		- generate an in-memory hash table of the relation
+ *		MultiExecHash	- generate an in-memory hash table of the relation
  *		ExecInitHash	- initialize node and subnodes
  *		ExecEndHash		- shutdown node and subnodes
  */
@@ -22,6 +22,7 @@
 
 #include "executor/execdebug.h"
 #include "executor/hashjoin.h"
+#include "executor/instrument.h"
 #include "executor/nodeHash.h"
 #include "executor/nodeHashjoin.h"
 #include "miscadmin.h"
@@ -36,13 +37,26 @@ static void ExecHashIncreaseNumBatches(HashJoinTable hashtable);
 /* ----------------------------------------------------------------
  *		ExecHash
  *
- *		build hash table for hashjoin, doing partitioning if more
- *		than one batch is required.
+ *		stub for pro forma compliance
  * ----------------------------------------------------------------
  */
 TupleTableSlot *
 ExecHash(HashState *node)
 {
+	elog(ERROR, "Hash node does not support ExecProcNode call convention");
+	return NULL;
+}
+
+/* ----------------------------------------------------------------
+ *		MultiExecHash
+ *
+ *		build hash table for hashjoin, doing partitioning if more
+ *		than one batch is required.
+ * ----------------------------------------------------------------
+ */
+Node *
+MultiExecHash(HashState *node)
+{
 	PlanState  *outerNode;
 	List	   *hashkeys;
 	HashJoinTable hashtable;
@@ -50,6 +64,10 @@ ExecHash(HashState *node)
 	ExprContext *econtext;
 	uint32		hashvalue;
 
+	/* must provide our own instrumentation support */
+	if (node->ps.instrument)
+		InstrStartNode(node->ps.instrument);
+
 	/*
 	 * get state info from node
 	 */
@@ -70,14 +88,24 @@ ExecHash(HashState *node)
 		slot = ExecProcNode(outerNode);
 		if (TupIsNull(slot))
 			break;
-		hashtable->hashNonEmpty = true;
+		hashtable->totalTuples += 1;
 		/* We have to compute the hash value */
 		econtext->ecxt_innertuple = slot;
 		hashvalue = ExecHashGetHashValue(hashtable, econtext, hashkeys);
 		ExecHashTableInsert(hashtable, ExecFetchSlotTuple(slot), hashvalue);
 	}
 
-	/* We needn't return a tuple slot or anything else */
+	/* must provide our own instrumentation support */
+	if (node->ps.instrument)
+		InstrStopNodeMulti(node->ps.instrument, hashtable->totalTuples);
+
+	/*
+	 * We do not return the hash table directly because it's not a subtype
+	 * of Node, and so would violate the MultiExecProcNode API.  Instead,
+	 * our parent Hashjoin node is expected to know how to fish it out
+	 * of our node state.  Ugly but not really worth cleaning up, since
+	 * Hashjoin knows quite a bit more about Hash besides that.
+	 */
 	return NULL;
 }
 
@@ -220,7 +248,7 @@ ExecHashTableCreate(Hash *node, List *hashOperators)
 	hashtable->nbatch_original = nbatch;
 	hashtable->nbatch_outstart = nbatch;
 	hashtable->growEnabled = true;
-	hashtable->hashNonEmpty = false;
+	hashtable->totalTuples = 0;
 	hashtable->innerBatchFile = NULL;
 	hashtable->outerBatchFile = NULL;
 	hashtable->spaceUsed = 0;
diff --git a/src/backend/executor/nodeHashjoin.c b/src/backend/executor/nodeHashjoin.c
index 4811b7068eb..38e48cd6dce 100644
--- a/src/backend/executor/nodeHashjoin.c
+++ b/src/backend/executor/nodeHashjoin.c
@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/executor/nodeHashjoin.c,v 1.70 2005/03/31 02:02:52 neilc Exp $
+ *	  $PostgreSQL: pgsql/src/backend/executor/nodeHashjoin.c,v 1.71 2005/04/16 20:07:35 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -123,13 +123,13 @@ ExecHashJoin(HashJoinState *node)
 		 * execute the Hash node, to build the hash table
 		 */
 		hashNode->hashtable = hashtable;
-		(void) ExecProcNode((PlanState *) hashNode);
+		(void) MultiExecProcNode((PlanState *) hashNode);
 
 		/*
 		 * If the inner relation is completely empty, and we're not doing
 		 * an outer join, we can quit without scanning the outer relation.
 		 */
-		if (!hashtable->hashNonEmpty && node->js.jointype != JOIN_LEFT)
+		if (hashtable->totalTuples == 0 && node->js.jointype != JOIN_LEFT)
 		{
 			ExecHashTableDestroy(hashtable);
 			node->hj_HashTable = NULL;