Rewrite of planner statistics-gathering code. ANALYZE is now available as

a separate statement (though it can still be invoked as part of VACUUM, too).
pg_statistic redesigned to be more flexible about what statistics are
stored.  ANALYZE now collects a list of several of the most common values,
not just one, plus a histogram (not just the min and max values).  Random
sampling is used to make the process reasonably fast even on very large
tables.  The number of values and histogram bins collected is now
user-settable via an ALTER TABLE command.

There is more still to do; the new stats are not being used everywhere
they could be in the planner.  But the remaining changes for this project
should be localized, and the behavior is already better than before.

A not-very-related change is that sorting now makes use of btree comparison
routines if it can find one, rather than invoking '<' twice.

20 files changed, 367 insertions, 322 deletions

diff --git a/src/include/access/tuptoaster.h b/src/include/access/tuptoaster.h
index 759ab3d39e..6e38529204 100644
--- a/src/include/access/tuptoaster.h
+++ b/src/include/access/tuptoaster.h
@@ -6,15 +6,13 @@
  *
  * Copyright (c) 2000, PostgreSQL Development Team
  *
- * $Id: tuptoaster.h,v 1.10 2001/03/22 04:00:32 momjian Exp $
+ * $Id: tuptoaster.h,v 1.11 2001/05/07 00:43:24 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
 #ifndef TUPTOASTER_H
 #define TUPTOASTER_H
 
-#ifdef TUPLE_TOASTER_ACTIVE
-
 #include "access/heapam.h"
 #include "access/htup.h"
 #include "access/tupmacs.h"
@@ -109,7 +107,13 @@ extern varattrib *heap_tuple_untoast_attr(varattrib *attr);
  */
 extern Datum toast_compress_datum(Datum value);
 
-#endif	 /* TUPLE_TOASTER_ACTIVE */
+/* ----------
+ * toast_raw_datum_size -
+ *
+ *	Return the raw (detoasted) size of a varlena datum
+ * ----------
+ */
+extern Size toast_raw_datum_size(Datum value);
 
 
 #endif	 /* TUPTOASTER_H */
diff --git a/src/include/catalog/catversion.h b/src/include/catalog/catversion.h
index 963b11c1d3..832f91fb09 100644
--- a/src/include/catalog/catversion.h
+++ b/src/include/catalog/catversion.h
@@ -37,7 +37,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: catversion.h,v 1.70 2001/03/22 04:00:35 momjian Exp $
+ * $Id: catversion.h,v 1.71 2001/05/07 00:43:24 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -53,6 +53,6 @@
  */
 
 /*							yyyymmddN */
-#define CATALOG_VERSION_NO	200101061
+#define CATALOG_VERSION_NO	200105051
 
 #endif
diff --git a/src/include/catalog/heap.h b/src/include/catalog/heap.h
index a7248f6c6d..7ab04b05fb 100644
--- a/src/include/catalog/heap.h
+++ b/src/include/catalog/heap.h
@@ -7,13 +7,14 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: heap.h,v 1.34 2001/03/22 04:00:35 momjian Exp $
+ * $Id: heap.h,v 1.35 2001/05/07 00:43:24 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
 #ifndef HEAP_H
 #define HEAP_H
 
+#include "catalog/pg_attribute.h"
 #include "utils/rel.h"
 
 typedef struct RawColumnDefault
@@ -44,4 +45,6 @@ extern void AddRelationRawConstraints(Relation rel,
 						  List *rawColDefaults,
 						  List *rawConstraints);
 
+extern Form_pg_attribute SystemAttributeDefinition(AttrNumber attno);
+
 #endif	 /* HEAP_H */
diff --git a/src/include/catalog/index.h b/src/include/catalog/index.h
index 1dac0bb1c3..07aaad61c7 100644
--- a/src/include/catalog/index.h
+++ b/src/include/catalog/index.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: index.h,v 1.33 2001/03/22 04:00:35 momjian Exp $
+ * $Id: index.h,v 1.34 2001/05/07 00:43:24 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -46,7 +46,7 @@ extern void FormIndexDatum(IndexInfo *indexInfo,
 			   Datum *datum,
 			   char *nullv);
 
-extern void UpdateStats(Oid relid, long reltuples);
+extern void UpdateStats(Oid relid, double reltuples);
 extern bool IndexesAreActive(Oid relid, bool comfirmCommitted);
 extern void setRelhasindex(Oid relid, bool hasindex);
 
diff --git a/src/include/catalog/indexing.h b/src/include/catalog/indexing.h
index 41a580a377..cc155cf1bb 100644
--- a/src/include/catalog/indexing.h
+++ b/src/include/catalog/indexing.h
@@ -8,7 +8,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: indexing.h,v 1.48 2001/03/22 04:00:36 momjian Exp $
+ * $Id: indexing.h,v 1.49 2001/05/07 00:43:24 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -171,7 +171,7 @@ DECLARE_UNIQUE_INDEX(pg_rewrite_rulename_index on pg_rewrite using btree(rulenam
 xDECLARE_UNIQUE_INDEX(pg_shadow_name_index on pg_shadow using btree(usename name_ops));
 xDECLARE_UNIQUE_INDEX(pg_shadow_sysid_index on pg_shadow using btree(usesysid int4_ops));
 */
-DECLARE_INDEX(pg_statistic_relid_att_index on pg_statistic using btree(starelid oid_ops, staattnum int2_ops));
+DECLARE_UNIQUE_INDEX(pg_statistic_relid_att_index on pg_statistic using btree(starelid oid_ops, staattnum int2_ops));
 DECLARE_INDEX(pg_trigger_tgconstrname_index on pg_trigger using btree(tgconstrname name_ops));
 DECLARE_INDEX(pg_trigger_tgconstrrelid_index on pg_trigger using btree(tgconstrrelid oid_ops));
 DECLARE_INDEX(pg_trigger_tgrelid_index on pg_trigger using btree(tgrelid oid_ops));
diff --git a/src/include/catalog/pg_attribute.h b/src/include/catalog/pg_attribute.h
index 58724e94dc..6e11aa6d53 100644
--- a/src/include/catalog/pg_attribute.h
+++ b/src/include/catalog/pg_attribute.h
@@ -8,7 +8,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: pg_attribute.h,v 1.70 2001/03/22 04:00:37 momjian Exp $
+ * $Id: pg_attribute.h,v 1.71 2001/05/07 00:43:24 tgl Exp $
  *
  * NOTES
  *	  the genbki.sh script reads this file and generates .bki
@@ -36,15 +36,14 @@
  *		typedef struct FormData_pg_attribute
  *
  *		If you change the following, make sure you change the structs for
- *		system attributes in heap.c and index.c also.
+ *		system attributes in catalog/heap.c also.
  * ----------------
  */
 CATALOG(pg_attribute) BOOTSTRAP
 {
 	Oid			attrelid;		/* OID of relation containing this
 								 * attribute */
-	NameData	attname;
-	Oid			atttypid;
+	NameData	attname;		/* name of attribute */
 
 	/*
 	 * atttypid is the OID of the instance in Catalog Class pg_type that
@@ -53,30 +52,20 @@ CATALOG(pg_attribute) BOOTSTRAP
 	 * attalign attributes of this instance, so they had better match or
 	 * Postgres will fail.
 	 */
-
-	float4		attdispersion;
+	Oid			atttypid;
 
 	/*
-	 * attdispersion is the dispersion statistic of the column (0.0 to
-	 * 1.0), or zero if the statistic has not been calculated, or -1.0 if
-	 * VACUUM found that the column contains no duplicate entries (in
-	 * which case the dispersion should be taken as 1.0/numberOfRows for
-	 * the current table size).  The -1.0 hack is useful because the
-	 * number of rows may be updated more often than attdispersion is. We
-	 * assume that the column will retain its no-duplicate-entry property.
-	 * (Perhaps this should be driven off the existence of a UNIQUE index
-	 * for the column, instead of being a statistical guess?)
+	 * attstattarget is the target number of statistics datapoints to collect
+	 * during VACUUM ANALYZE of this column.  A zero here indicates that we
+	 * do not wish to collect any stats about this column.
 	 */
-
-	int2		attlen;
+	int4		attstattarget;
 
 	/*
 	 * attlen is a copy of the typlen field from pg_type for this
-	 * attribute.  See atttypid above.	See struct Form_pg_type for
-	 * definition.
+	 * attribute.  See atttypid comments above.
 	 */
-
-	int2		attnum;
+	int2		attlen;
 
 	/*
 	 * attnum is the "attribute number" for the attribute:	A value that
@@ -91,10 +80,13 @@ CATALOG(pg_attribute) BOOTSTRAP
 	 *
 	 * Note that (attnum - 1) is often used as the index to an array.
 	 */
+	int2		attnum;
 
-	int4		attnelems;		/* number of dimensions, if an array type */
-
-	int4		attcacheoff;
+	/*
+	 * attndims is the declared number of dimensions, if an array type,
+	 * otherwise zero.
+	 */
+	int4		attndims;
 
 	/*
 	 * fastgetattr() uses attcacheoff to cache byte offsets of attributes
@@ -103,8 +95,7 @@ CATALOG(pg_attribute) BOOTSTRAP
 	 * tuple descriptor, we may then update attcacheoff in the copies.
 	 * This speeds up the attribute walking process.
 	 */
-
-	int4		atttypmod;
+	int4		attcacheoff;
 
 	/*
 	 * atttypmod records type-specific data supplied at table creation
@@ -113,16 +104,13 @@ CATALOG(pg_attribute) BOOTSTRAP
 	 * argument. The value will generally be -1 for types that do not need
 	 * typmod.
 	 */
-
-	bool		attbyval;
+	int4		atttypmod;
 
 	/*
 	 * attbyval is a copy of the typbyval field from pg_type for this
-	 * attribute.  See atttypid above.	See struct Form_pg_type for
-	 * definition.
+	 * attribute.  See atttypid comments above.
 	 */
-
-	char		attstorage;
+	bool		attbyval;
 
 	/*----------
 	 * attstorage tells for VARLENA attributes, what the heap access
@@ -137,30 +125,31 @@ CATALOG(pg_attribute) BOOTSTRAP
 	 * but only as a last resort ('e' and 'x' fields are moved first).
 	 *----------
 	 */
+	char		attstorage;
 
+	/* This flag indicates that the attribute is really a set */
 	bool		attisset;
-	char		attalign;
 
 	/*
 	 * attalign is a copy of the typalign field from pg_type for this
-	 * attribute.  See atttypid above.	See struct Form_pg_type for
-	 * definition.
+	 * attribute.  See atttypid comments above.
 	 */
-
-	bool		attnotnull;
+	char		attalign;
 
 	/* This flag represents the "NOT NULL" constraint */
-	bool		atthasdef;
+	bool		attnotnull;
 
 	/* Has DEFAULT value or not */
+	bool		atthasdef;
 } FormData_pg_attribute;
 
 /*
  * someone should figure out how to do this properly. (The problem is
- * the size of the C struct is not the same as the size of the tuple.)
+ * the size of the C struct is not the same as the size of the tuple
+ * because of alignment padding at the end of the struct.)
  */
 #define ATTRIBUTE_TUPLE_SIZE \
-	(offsetof(FormData_pg_attribute,atthasdef) + sizeof(char))
+	(offsetof(FormData_pg_attribute,atthasdef) + sizeof(bool))
 
 /* ----------------
  *		Form_pg_attribute corresponds to a pointer to a tuple with
@@ -178,10 +167,10 @@ typedef FormData_pg_attribute *Form_pg_attribute;
 #define Anum_pg_attribute_attrelid		1
 #define Anum_pg_attribute_attname		2
 #define Anum_pg_attribute_atttypid		3
-#define Anum_pg_attribute_attdispersion 4
+#define Anum_pg_attribute_attstattarget 4
 #define Anum_pg_attribute_attlen		5
 #define Anum_pg_attribute_attnum		6
-#define Anum_pg_attribute_attnelems		7
+#define Anum_pg_attribute_attndims		7
 #define Anum_pg_attribute_attcacheoff	8
 #define Anum_pg_attribute_atttypmod		9
 #define Anum_pg_attribute_attbyval		10
@@ -206,6 +195,7 @@ typedef FormData_pg_attribute *Form_pg_attribute;
 	(attribute)->attnotnull = false; \
 	(attribute)->atthasdef = false;
 #endif	 /* _DROP_COLUMN_HACK__ */
+
 /* ----------------
  *		SCHEMA_ macros for declaring hardcoded tuple descriptors.
  *		these are used in utils/cache/relcache.c
@@ -231,25 +221,25 @@ typedef FormData_pg_attribute *Form_pg_attribute;
  * ----------------
  */
 #define Schema_pg_type \
-{ 1247, {"typname"},	   19, 0, NAMEDATALEN,	1, 0, -1, -1, '\0', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1247, {"typowner"},	   23, 0,	4,	2, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1247, {"typlen"},		   21, 0,	2,	3, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1247, {"typprtlen"},	   21, 0,	2,	4, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1247, {"typbyval"},	   16, 0,	1,	5, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1247, {"typtype"},	   18, 0,	1,	6, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1247, {"typisdefined"},  16, 0,	1,	7, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1247, {"typdelim"},	   18, 0,	1,	8, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1247, {"typrelid"},	   26, 0,	4,	9, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1247, {"typelem"},	   26, 0,	4, 10, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1247, {"typinput"},	   24, 0,	4, 11, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1247, {"typoutput"},	   24, 0,	4, 12, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1247, {"typreceive"},    24, 0,	4, 13, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1247, {"typsend"},	   24, 0,	4, 14, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1247, {"typalign"},	   18, 0,	1, 15, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1247, {"typstorage"},    18, 0,	1, 16, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1247, {"typdefault"},    25, 0,  -1, 17, 0, -1, -1, '\0'	, 'x', '\0', 'i', '\0', '\0' }
-
-DATA(insert OID = 0 ( 1247 typname			19 0 NAMEDATALEN   1 0 -1 -1 f p f i f f));
+{ 1247, {"typname"},	   19, DEFAULT_ATTSTATTARGET, NAMEDATALEN,	1, 0, -1, -1, false, 'p', false, 'i', false, false }, \
+{ 1247, {"typowner"},	   23, 0,	4,	2, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1247, {"typlen"},		   21, 0,	2,	3, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1247, {"typprtlen"},	   21, 0,	2,	4, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1247, {"typbyval"},	   16, 0,	1,	5, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1247, {"typtype"},	   18, 0,	1,	6, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1247, {"typisdefined"},  16, 0,	1,	7, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1247, {"typdelim"},	   18, 0,	1,	8, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1247, {"typrelid"},	   26, 0,	4,	9, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1247, {"typelem"},	   26, 0,	4, 10, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1247, {"typinput"},	   24, 0,	4, 11, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1247, {"typoutput"},	   24, 0,	4, 12, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1247, {"typreceive"},    24, 0,	4, 13, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1247, {"typsend"},	   24, 0,	4, 14, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1247, {"typalign"},	   18, 0,	1, 15, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1247, {"typstorage"},    18, 0,	1, 16, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1247, {"typdefault"},    25, 0,  -1, 17, 0, -1, -1, false	, 'x', false, 'i', false, false }
+
+DATA(insert OID = 0 ( 1247 typname			19 DEFAULT_ATTSTATTARGET NAMEDATALEN   1 0 -1 -1 f p f i f f));
 DATA(insert OID = 0 ( 1247 typowner			23 0  4   2 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1247 typlen			21 0  2   3 0 -1 -1 t p f s f f));
 DATA(insert OID = 0 ( 1247 typprtlen		21 0  2   4 0 -1 -1 t p f s f f));
@@ -299,25 +289,25 @@ DATA(insert OID = 0 ( 1262 tableoid			26 0  4  -7 0 -1 -1 t p f i f f));
  * ----------------
  */
 #define Schema_pg_proc \
-{ 1255, {"proname"},			19, 0, NAMEDATALEN,  1, 0, -1, -1, '\0', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"proowner"},			23, 0,	4,	2, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"prolang"},			26, 0,	4,	3, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"proisinh"},			16, 0,	1,	4, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1255, {"proistrusted"},		16, 0,	1,	5, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1255, {"proiscachable"},		16, 0,	1,	6, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1255, {"proisstrict"},		16, 0,	1,	7, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1255, {"pronargs"},			21, 0,	2,	8, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1255, {"proretset"},			16, 0,	1,	9, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1255, {"prorettype"},			26, 0,	4, 10, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"proargtypes"},		30, 0, INDEX_MAX_KEYS*4, 11, 0, -1, -1, '\0', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"probyte_pct"},		23, 0,	4, 12, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"properbyte_cpu"},		23, 0,	4, 13, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"propercall_cpu"},		23, 0,	4, 14, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"prooutin_ratio"},		23, 0,	4, 15, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"prosrc"},				25, 0, -1, 16, 0, -1, -1, '\0', 'x', '\0', 'i', '\0', '\0' }, \
-{ 1255, {"probin"},				17, 0, -1, 17, 0, -1, -1, '\0', 'x', '\0', 'i', '\0', '\0' }
-
-DATA(insert OID = 0 ( 1255 proname			19 0 NAMEDATALEN   1 0 -1 -1 f p f i f f));
+{ 1255, {"proname"},			19, DEFAULT_ATTSTATTARGET, NAMEDATALEN,  1, 0, -1, -1, false, 'p', false, 'i', false, false }, \
+{ 1255, {"proowner"},			23, 0,	4,	2, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1255, {"prolang"},			26, 0,	4,	3, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1255, {"proisinh"},			16, 0,	1,	4, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1255, {"proistrusted"},		16, 0,	1,	5, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1255, {"proiscachable"},		16, 0,	1,	6, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1255, {"proisstrict"},		16, 0,	1,	7, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1255, {"pronargs"},			21, 0,	2,	8, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1255, {"proretset"},			16, 0,	1,	9, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1255, {"prorettype"},			26, 0,	4, 10, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1255, {"proargtypes"},		30, 0, INDEX_MAX_KEYS*4, 11, 0, -1, -1, false, 'p', false, 'i', false, false }, \
+{ 1255, {"probyte_pct"},		23, 0,	4, 12, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1255, {"properbyte_cpu"},		23, 0,	4, 13, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1255, {"propercall_cpu"},		23, 0,	4, 14, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1255, {"prooutin_ratio"},		23, 0,	4, 15, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1255, {"prosrc"},				25, 0, -1, 16, 0, -1, -1, false, 'x', false, 'i', false, false }, \
+{ 1255, {"probin"},				17, 0, -1, 17, 0, -1, -1, false, 'x', false, 'i', false, false }
+
+DATA(insert OID = 0 ( 1255 proname			19 DEFAULT_ATTSTATTARGET NAMEDATALEN   1 0 -1 -1 f p f i f f));
 DATA(insert OID = 0 ( 1255 proowner			23 0  4   2 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1255 prolang			26 0  4   3 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1255 proisinh			16 0  1   4 0 -1 -1 t p f c f f));
@@ -346,8 +336,8 @@ DATA(insert OID = 0 ( 1255 tableoid			26 0  4  -7 0 -1 -1 t p f i f f));
  *		pg_shadow
  * ----------------
  */
-DATA(insert OID = 0 ( 1260 usename			19	0 NAMEDATALEN	1 0 -1 -1 f p f i f f));
-DATA(insert OID = 0 ( 1260 usesysid			23	0	4	2 0 -1 -1 t p f i f f));
+DATA(insert OID = 0 ( 1260 usename			19	DEFAULT_ATTSTATTARGET NAMEDATALEN	1 0 -1 -1 f p f i f f));
+DATA(insert OID = 0 ( 1260 usesysid			23	DEFAULT_ATTSTATTARGET	4	2 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1260 usecreatedb		16	0	1	3 0 -1 -1 t p f c f f));
 DATA(insert OID = 0 ( 1260 usetrace			16	0	1	4 0 -1 -1 t p f c f f));
 DATA(insert OID = 0 ( 1260 usesuper			16	0	1	5 0 -1 -1 t p f c f f));
@@ -366,8 +356,8 @@ DATA(insert OID = 0 ( 1260 tableoid			26 0  4  -7 0 -1 -1 t p f i f f));
  *		pg_group
  * ----------------
  */
-DATA(insert OID = 0 ( 1261 groname			19 0 NAMEDATALEN  1 0 -1 -1 f p f i f f));
-DATA(insert OID = 0 ( 1261 grosysid			23 0  4   2 0 -1 -1 t p f i f f));
+DATA(insert OID = 0 ( 1261 groname			19 DEFAULT_ATTSTATTARGET NAMEDATALEN  1 0 -1 -1 f p f i f f));
+DATA(insert OID = 0 ( 1261 grosysid			23 DEFAULT_ATTSTATTARGET  4   2 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1261 grolist		  1007 0 -1   3 0 -1 -1 f x f i f f));
 DATA(insert OID = 0 ( 1261 ctid				27 0  6  -1 0 -1 -1 f p f i f f));
 DATA(insert OID = 0 ( 1261 oid				26 0  4  -2 0 -1 -1 t p f i f f));
@@ -382,29 +372,29 @@ DATA(insert OID = 0 ( 1261 tableoid			26 0  4  -7 0 -1 -1 t p f i f f));
  * ----------------
  */
 #define Schema_pg_attribute \
-{ 1249, {"attrelid"},	  26, 0,	4,	1, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1249, {"attname"},	  19, 0, NAMEDATALEN,	2, 0, -1, -1, '\0', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1249, {"atttypid"},	  26, 0,	4,	3, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1249, {"attdispersion"}, 700, 0,	4,	4, 0, -1, -1, '\0', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1249, {"attlen"},		  21, 0,	2,	5, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1249, {"attnum"},		  21, 0,	2,	6, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1249, {"attnelems"},	  23, 0,	4,	7, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1249, {"attcacheoff"},  23, 0,	4,	8, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1249, {"atttypmod"},	  23, 0,	4,	9, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1249, {"attbyval"},	  16, 0,	1, 10, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1249, {"attstorage"},   18, 0,	1, 11, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1249, {"attisset"},	  16, 0,	1, 12, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1249, {"attalign"},	  18, 0,	1, 13, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1249, {"attnotnull"},  16, 0, 1, 14, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1249, {"atthasdef"},	 16, 0, 1, 15, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }
-
-DATA(insert OID = 0 ( 1249 attrelid			26 0  4   1 0 -1 -1 t p f i f f));
-DATA(insert OID = 0 ( 1249 attname			19 0 NAMEDATALEN  2 0 -1 -1 f p f i f f));
+{ 1249, {"attrelid"},	  26, DEFAULT_ATTSTATTARGET,	4,	1, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1249, {"attname"},	  19, DEFAULT_ATTSTATTARGET, NAMEDATALEN,	2, 0, -1, -1, false, 'p', false, 'i', false, false }, \
+{ 1249, {"atttypid"},	  26, 0,	4,	3, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1249, {"attstattarget"}, 23, 0,	4,	4, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1249, {"attlen"},		  21, 0,	2,	5, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1249, {"attnum"},		  21, 0,	2,	6, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1249, {"attndims"},	  23, 0,	4,	7, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1249, {"attcacheoff"},  23, 0,	4,	8, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1249, {"atttypmod"},	  23, 0,	4,	9, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1249, {"attbyval"},	  16, 0,	1, 10, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1249, {"attstorage"},   18, 0,	1, 11, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1249, {"attisset"},	  16, 0,	1, 12, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1249, {"attalign"},	  18, 0,	1, 13, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1249, {"attnotnull"},  16, 0, 1, 14, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1249, {"atthasdef"},	 16, 0, 1, 15, 0, -1, -1, true, 'p', false, 'c', false, false }
+
+DATA(insert OID = 0 ( 1249 attrelid			26 DEFAULT_ATTSTATTARGET  4   1 0 -1 -1 t p f i f f));
+DATA(insert OID = 0 ( 1249 attname			19 DEFAULT_ATTSTATTARGET NAMEDATALEN  2 0 -1 -1 f p f i f f));
 DATA(insert OID = 0 ( 1249 atttypid			26 0  4   3 0 -1 -1 t p f i f f));
-DATA(insert OID = 0 ( 1249 attdispersion   700 0  4   4 0 -1 -1 f p f i f f));
+DATA(insert OID = 0 ( 1249 attstattarget	23 0  4   4 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1249 attlen			21 0  2   5 0 -1 -1 t p f s f f));
 DATA(insert OID = 0 ( 1249 attnum			21 0  2   6 0 -1 -1 t p f s f f));
-DATA(insert OID = 0 ( 1249 attnelems		23 0  4   7 0 -1 -1 t p f i f f));
+DATA(insert OID = 0 ( 1249 attndims			23 0  4   7 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1249 attcacheoff		23 0  4   8 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1249 atttypmod		23 0  4   9 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1249 attbyval			16 0  1  10 0 -1 -1 t p f c f f));
@@ -426,36 +416,36 @@ DATA(insert OID = 0 ( 1249 tableoid			26 0  4  -7 0 -1 -1 t p f i f f));
  * ----------------
  */
 #define Schema_pg_class \
-{ 1259, {"relname"},	   19, 0, NAMEDATALEN,	1, 0, -1, -1, '\0', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1259, {"reltype"},	   26, 0,	4,	2, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1259, {"relowner"},	   23, 0,	4,	3, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1259, {"relam"},		   26, 0,	4,	4, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1259, {"relfilenode"},   26, 0,	4,	5, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1259, {"relpages"},	   23, 0,	4,	6, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1259, {"reltuples"},	   23, 0,	4,	7, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1259, {"reltoastrelid"}, 26, 0,	4,	8, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1259, {"reltoastidxid"}, 26, 0,	4,	9, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }, \
-{ 1259, {"relhasindex"},   16, 0,	1, 10, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1259, {"relisshared"},   16, 0,	1, 11, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1259, {"relkind"},	   18, 0,	1, 12, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1259, {"relnatts"},	   21, 0,	2, 13, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1259, {"relchecks"},	   21, 0,	2, 14, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1259, {"reltriggers"},   21, 0,	2, 15, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1259, {"relukeys"},	   21, 0,	2, 16, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1259, {"relfkeys"},	   21, 0,	2, 17, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1259, {"relrefs"},	   21, 0,	2, 18, 0, -1, -1, '\001', 'p', '\0', 's', '\0', '\0' }, \
-{ 1259, {"relhaspkey"},    16, 0,	1, 19, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1259, {"relhasrules"},   16, 0,	1, 20, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1259, {"relhassubclass"},16, 0,	1, 21, 0, -1, -1, '\001', 'p', '\0', 'c', '\0', '\0' }, \
-{ 1259, {"relacl"},		 1034, 0,  -1, 22, 0, -1, -1,	'\0', 'x', '\0', 'i', '\0', '\0' }
-
-DATA(insert OID = 0 ( 1259 relname			19 0 NAMEDATALEN   1 0 -1 -1 f p f i f f));
+{ 1259, {"relname"},	   19, DEFAULT_ATTSTATTARGET, NAMEDATALEN,	1, 0, -1, -1, false, 'p', false, 'i', false, false }, \
+{ 1259, {"reltype"},	   26, 0,	4,	2, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1259, {"relowner"},	   23, 0,	4,	3, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1259, {"relam"},		   26, 0,	4,	4, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1259, {"relfilenode"},   26, 0,	4,	5, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1259, {"relpages"},	   23, 0,	4,	6, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1259, {"reltuples"},	   700, 0,	4,	7, 0, -1, -1, false, 'p', false, 'i', false, false }, \
+{ 1259, {"reltoastrelid"}, 26, 0,	4,	8, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1259, {"reltoastidxid"}, 26, 0,	4,	9, 0, -1, -1, true, 'p', false, 'i', false, false }, \
+{ 1259, {"relhasindex"},   16, 0,	1, 10, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1259, {"relisshared"},   16, 0,	1, 11, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1259, {"relkind"},	   18, 0,	1, 12, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1259, {"relnatts"},	   21, 0,	2, 13, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1259, {"relchecks"},	   21, 0,	2, 14, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1259, {"reltriggers"},   21, 0,	2, 15, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1259, {"relukeys"},	   21, 0,	2, 16, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1259, {"relfkeys"},	   21, 0,	2, 17, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1259, {"relrefs"},	   21, 0,	2, 18, 0, -1, -1, true, 'p', false, 's', false, false }, \
+{ 1259, {"relhaspkey"},    16, 0,	1, 19, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1259, {"relhasrules"},   16, 0,	1, 20, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1259, {"relhassubclass"},16, 0,	1, 21, 0, -1, -1, true, 'p', false, 'c', false, false }, \
+{ 1259, {"relacl"},		 1034, 0,  -1, 22, 0, -1, -1,	false, 'x', false, 'i', false, false }
+
+DATA(insert OID = 0 ( 1259 relname			19 DEFAULT_ATTSTATTARGET NAMEDATALEN   1 0 -1 -1 f p f i f f));
 DATA(insert OID = 0 ( 1259 reltype			26 0  4   2 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1259 relowner			23 0  4   3 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1259 relam			26 0  4   4 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1259 relfilenode		26 0  4   5 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1259 relpages			23 0  4   6 0 -1 -1 t p f i f f));
-DATA(insert OID = 0 ( 1259 reltuples		23 0  4   7 0 -1 -1 t p f i f f));
+DATA(insert OID = 0 ( 1259 reltuples	   700 0  4   7 0 -1 -1 f p f i f f));
 DATA(insert OID = 0 ( 1259 reltoastrelid	26 0  4   8 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1259 reltoastidxid	26 0  4   9 0 -1 -1 t p f i f f));
 DATA(insert OID = 0 ( 1259 relhasindex		16 0  1  10 0 -1 -1 t p f c f f));
@@ -544,7 +534,7 @@ DATA(insert OID = 0 ( 1219 tableoid			26 0  4  -7 0 -1 -1 t p f i f f));
  * ----------------
  */
 #define Schema_pg_variable \
-{ 1264, {"varfoo"},  26, 0, 4, 1, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }
+{ 1264, {"varfoo"},  26, 0, 4, 1, 0, -1, -1, true, 'p', false, 'i', false, false }
 
 DATA(insert OID = 0 ( 1264 varfoo			26 0  4   1 0 -1 -1 t p f i f f));
 
@@ -555,7 +545,7 @@ DATA(insert OID = 0 ( 1264 varfoo			26 0  4   1 0 -1 -1 t p f i f f));
  * ----------------
  */
 #define Schema_pg_log \
-{ 1269, {"logfoo"},  26, 0, 4, 1, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }
+{ 1269, {"logfoo"},  26, 0, 4, 1, 0, -1, -1, true, 'p', false, 'i', false, false }
 
 DATA(insert OID = 0 ( 1269 logfoo			26 0  4   1 0 -1 -1 t p f i f f));
 
@@ -566,7 +556,7 @@ DATA(insert OID = 0 ( 1269 logfoo			26 0  4   1 0 -1 -1 t p f i f f));
  * ----------------
  */
 #define Schema_pg_xactlock \
-{ 376, {"xactlockfoo"},  26, 0, 4, 1, 0, -1, -1, '\001', 'p', '\0', 'i', '\0', '\0' }
+{ 376, {"xactlockfoo"},  26, 0, 4, 1, 0, -1, -1, true, 'p', false, 'i', false, false }
 
 DATA(insert OID = 0 ( 376 xactlockfoo		26 0  4   1 0 -1 -1 t p f i f f));
 
diff --git a/src/include/catalog/pg_class.h b/src/include/catalog/pg_class.h
index 81e75e14b6..86de88cc9b 100644
--- a/src/include/catalog/pg_class.h
+++ b/src/include/catalog/pg_class.h
@@ -8,7 +8,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: pg_class.h,v 1.47 2001/03/22 04:00:38 momjian Exp $
+ * $Id: pg_class.h,v 1.48 2001/05/07 00:43:25 tgl Exp $
  *
  * NOTES
  *	  the genbki.sh script reads this file and generates .bki
@@ -52,7 +52,7 @@ CATALOG(pg_class) BOOTSTRAP
 	Oid			relam;
 	Oid			relfilenode;
 	int4		relpages;
-	int4		reltuples;
+	float4		reltuples;
 	Oid			reltoastrelid;
 	Oid			reltoastidxid;
 	bool		relhasindex;
diff --git a/src/include/catalog/pg_statistic.h b/src/include/catalog/pg_statistic.h
index 2f39bea324..8d6a6b37c1 100644
--- a/src/include/catalog/pg_statistic.h
+++ b/src/include/catalog/pg_statistic.h
@@ -8,7 +8,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: pg_statistic.h,v 1.10 2001/01/24 19:43:22 momjian Exp $
+ * $Id: pg_statistic.h,v 1.11 2001/05/07 00:43:25 tgl Exp $
  *
  * NOTES
  *	  the genbki.sh script reads this file and generates .bki
@@ -36,40 +36,91 @@ CATALOG(pg_statistic)
 	/* These fields form the unique key for the entry: */
 	Oid			starelid;		/* relation containing attribute */
 	int2		staattnum;		/* attribute (column) stats are for */
-	Oid			staop;			/* '<' comparison op used for lo/hi vals */
+
+	/* the fraction of the column's entries that are NULL: */
+	float4		stanullfrac;
 
 	/*
-	 * Note: the current VACUUM code will never produce more than one
-	 * entry per column, but in theory there could be multiple entries if
-	 * a datatype has more than one useful ordering operator.  Also, the
-	 * current code will not write an entry unless it found at least one
-	 * non-NULL value in the column; so the remaining fields will never be
-	 * NULL.
+	 * stawidth is the average width in bytes of non-null entries.  For
+	 * fixed-width datatypes this is of course the same as the typlen, but
+	 * for varlena types it is more useful.  Note that this is the average
+	 * width of the data as actually stored, post-TOASTing (eg, for a
+	 * moved-out-of-line value, only the size of the pointer object is
+	 * counted).  This is the appropriate definition for the primary use of
+	 * the statistic, which is to estimate sizes of in-memory hash tables of
+	 * tuples.
+	 */
+	int4		stawidth;
+
+	/* ----------------
+	 * stadistinct indicates the (approximate) number of distinct non-null
+	 * data values in the column.  The interpretation is:
+	 *		0		unknown or not computed
+	 *		> 0		actual number of distinct values
+	 *		< 0		negative of multiplier for number of rows
+	 * The special negative case allows us to cope with columns that are
+	 * unique (stadistinct = -1) or nearly so (for example, a column in
+	 * which values appear about twice on the average could be represented
+	 * by stadistinct = -0.5).  Because the number-of-rows statistic in
+	 * pg_class may be updated more frequently than pg_statistic is, it's
+	 * important to be able to describe such situations as a multiple of
+	 * the number of rows, rather than a fixed number of distinct values.
+	 * But in other cases a fixed number is correct (eg, a boolean column).
+	 * ----------------
+	 */
+	float4		stadistinct;
+
+	/* ----------------
+	 * To allow keeping statistics on different kinds of datatypes,
+	 * we do not hard-wire any particular meaning for the remaining
+	 * statistical fields.  Instead, we provide several "slots" in which
+	 * statistical data can be placed.  Each slot includes:
+	 *		kind			integer code identifying kind of data
+	 *		op				OID of associated operator, if needed
+	 *		numbers			float4 array (for statistical values)
+	 *		values			text array (for representations of data values)
+	 * The ID and operator fields are never NULL; they are zeroes in an
+	 * unused slot.  The numbers and values fields are NULL in an unused
+	 * slot, and might also be NULL in a used slot if the slot kind has
+	 * no need for one or the other.
+	 * ----------------
 	 */
 
+	int2		stakind1;
+	int2		stakind2;
+	int2		stakind3;
+	int2		stakind4;
+
+	Oid			staop1;
+	Oid			staop2;
+	Oid			staop3;
+	Oid			staop4;
+
 	/*
-	 * These fields contain the stats about the column indicated by the
-	 * key
+	 * THE REST OF THESE ARE VARIABLE LENGTH FIELDS, and may even be absent
+	 * (NULL). They cannot be accessed as C struct entries; you have to use
+	 * the full field access machinery (heap_getattr) for them.  We declare
+	 * them here for the catalog machinery.
 	 */
-	float4		stanullfrac;	/* the fraction of the entries that are
-								 * NULL */
-	float4		stacommonfrac;	/* the fraction that are the most common
-								 * val */
+
+	float4		stanumbers1[1];
+	float4		stanumbers2[1];
+	float4		stanumbers3[1];
+	float4		stanumbers4[1];
 
 	/*
-	 * THE REST OF THESE ARE VARIABLE LENGTH FIELDS. They cannot be
-	 * accessed as C struct entries; you have to use the full field access
-	 * machinery (heap_getattr) for them.
-	 *
-	 * All three of these are text representations of data values of the
-	 * column's data type.  To re-create the actual Datum, do
-	 * datatypein(textout(givenvalue)).
+	 * Values in these text arrays are external representations of values
+	 * of the column's data type.  To re-create the actual Datum, do
+	 * datatypein(textout(arrayelement)).
 	 */
-	text		stacommonval;	/* most common non-null value in column */
-	text		staloval;		/* smallest non-null value in column */
-	text		stahival;		/* largest non-null value in column */
+	text		stavalues1[1];
+	text		stavalues2[1];
+	text		stavalues3[1];
+	text		stavalues4[1];
 } FormData_pg_statistic;
 
+#define STATISTIC_NUM_SLOTS  4
+
 /* ----------------
  *		Form_pg_statistic corresponds to a pointer to a tuple with
  *		the format of pg_statistic relation.
@@ -81,14 +132,78 @@ typedef FormData_pg_statistic *Form_pg_statistic;
  *		compiler constants for pg_statistic
  * ----------------
  */
-#define Natts_pg_statistic				8
+#define Natts_pg_statistic				21
 #define Anum_pg_statistic_starelid		1
 #define Anum_pg_statistic_staattnum		2
-#define Anum_pg_statistic_staop			3
-#define Anum_pg_statistic_stanullfrac	4
-#define Anum_pg_statistic_stacommonfrac 5
-#define Anum_pg_statistic_stacommonval	6
-#define Anum_pg_statistic_staloval		7
-#define Anum_pg_statistic_stahival		8
+#define Anum_pg_statistic_stanullfrac	3
+#define Anum_pg_statistic_stawidth		4
+#define Anum_pg_statistic_stadistinct	5
+#define Anum_pg_statistic_stakind1		6
+#define Anum_pg_statistic_stakind2		7
+#define Anum_pg_statistic_stakind3		8
+#define Anum_pg_statistic_stakind4		9
+#define Anum_pg_statistic_staop1		10
+#define Anum_pg_statistic_staop2		11
+#define Anum_pg_statistic_staop3		12
+#define Anum_pg_statistic_staop4		13
+#define Anum_pg_statistic_stanumbers1	14
+#define Anum_pg_statistic_stanumbers2	15
+#define Anum_pg_statistic_stanumbers3	16
+#define Anum_pg_statistic_stanumbers4	17
+#define Anum_pg_statistic_stavalues1	18
+#define Anum_pg_statistic_stavalues2	19
+#define Anum_pg_statistic_stavalues3	20
+#define Anum_pg_statistic_stavalues4	21
+
+/*
+ * Currently, three statistical slot "kinds" are defined: most common values,
+ * histogram, and correlation.  Additional "kinds" will probably appear in
+ * future to help cope with non-scalar datatypes.
+ *
+ * Code reading the pg_statistic relation should not assume that a particular
+ * data "kind" will appear in any particular slot.  Instead, search the
+ * stakind fields to see if the desired data is available.
+ */
+
+/*
+ * In a "most common values" slot, staop is the OID of the "=" operator
+ * used to decide whether values are the same or not.  stavalues contains
+ * the K most common non-null values appearing in the column, and stanumbers
+ * contains their frequencies (fractions of total row count).  The values
+ * shall be ordered in decreasing frequency.  Note that since the arrays are
+ * variable-size, K may be chosen by the statistics collector.  Values should
+ * not appear in MCV unless they have been observed to occur more than once;
+ * a unique column will have no MCV slot.
+ */
+#define STATISTIC_KIND_MCV  1
+
+/*
+ * A "histogram" slot describes the distribution of scalar data.  staop is
+ * the OID of the "<" operator that describes the sort ordering.  (In theory,
+ * more than one histogram could appear, if a datatype has more than one
+ * useful sort operator.)  stavalues contains M (>=2) non-null values that
+ * divide the non-null column data values into M-1 bins of approximately equal
+ * population.  The first stavalues item is the MIN and the last is the MAX.
+ * stanumbers is not used and should be NULL.  IMPORTANT POINT: if an MCV
+ * slot is also provided, then the histogram describes the data distribution
+ * *after removing the values listed in MCV* (thus, it's a "compressed
+ * histogram" in the technical parlance).  This allows a more accurate
+ * representation of the distribution of a column with some very-common
+ * values.  In a column with only a few distinct values, it's possible that
+ * the MCV list describes the entire data population; in this case the
+ * histogram reduces to empty and should be omitted.
+ */
+#define STATISTIC_KIND_HISTOGRAM  2
+
+/*
+ * A "correlation" slot describes the correlation between the physical order
+ * of table tuples and the ordering of data values of this column, as seen
+ * by the "<" operator identified by staop.  (As with the histogram, more
+ * than one entry could theoretically appear.)  stavalues is not used and
+ * should be NULL.  stanumbers contains a single entry, the correlation
+ * coefficient between the sequence of data values and the sequence of
+ * their actual tuple positions.  The coefficient ranges from +1 to -1.
+ */
+#define STATISTIC_KIND_CORRELATION  3
 
 #endif	 /* PG_STATISTIC_H */
diff --git a/src/include/commands/command.h b/src/include/commands/command.h
index 8b108451d2..7eb1a4fab8 100644
--- a/src/include/commands/command.h
+++ b/src/include/commands/command.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: command.h,v 1.26 2001/03/22 04:00:41 momjian Exp $
+ * $Id: command.h,v 1.27 2001/05/07 00:43:25 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -43,9 +43,13 @@ extern void PortalCleanup(Portal portal);
 extern void AlterTableAddColumn(const char *relationName,
 					bool inh, ColumnDef *colDef);
 
-extern void AlterTableAlterColumn(const char *relationName,
-					  bool inh, const char *colName,
-					  Node *newDefault);
+extern void AlterTableAlterColumnDefault(const char *relationName,
+										 bool inh, const char *colName,
+										 Node *newDefault);
+
+extern void AlterTableAlterColumnStatistics(const char *relationName,
+											bool inh, const char *colName,
+											Node *statsTarget);
 
 extern void AlterTableDropColumn(const char *relationName,
 					 bool inh, const char *colName,
diff --git a/src/include/commands/vacuum.h b/src/include/commands/vacuum.h
index d82d22fcdf..87bb0007aa 100644
--- a/src/include/commands/vacuum.h
+++ b/src/include/commands/vacuum.h
@@ -1,129 +1,27 @@
 /*-------------------------------------------------------------------------
  *
  * vacuum.h
- *	  header file for postgres vacuum cleaner
+ *	  header file for postgres vacuum cleaner and statistics analyzer
  *
  *
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: vacuum.h,v 1.34 2001/03/22 04:00:43 momjian Exp $
+ * $Id: vacuum.h,v 1.35 2001/05/07 00:43:25 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
 #ifndef VACUUM_H
 #define VACUUM_H
 
-#include "catalog/pg_attribute.h"
-#include "catalog/pg_index.h"
-#include "fmgr.h"
-#include "nodes/pg_list.h"
-#include "storage/itemptr.h"
+#include "nodes/parsenodes.h"
 
 
-typedef struct VAttListData
-{
-	int			val_dummy;
-	struct VAttListData *val_next;
-} VAttListData;
-
-typedef VAttListData *VAttList;
-
-typedef struct VacPageData
-{
-	BlockNumber blkno;			/* BlockNumber of this Page */
-	Size		free;			/* FreeSpace on this Page */
-	uint16		offsets_used;	/* Number of OffNums used by vacuum */
-	uint16		offsets_free;	/* Number of OffNums free or to be free */
-	OffsetNumber offsets[1];	/* Array of its OffNums */
-} VacPageData;
-
-typedef VacPageData *VacPage;
-
-typedef struct VacPageListData
-{
-	int			empty_end_pages;/* Number of "empty" end-pages */
-	int			num_pages;		/* Number of pages in pagedesc */
-	int			num_allocated_pages;	/* Number of allocated pages in
-										 * pagedesc */
-	VacPage    *pagedesc;		/* Descriptions of pages */
-} VacPageListData;
-
-typedef VacPageListData *VacPageList;
-
-typedef struct
-{
-	Form_pg_attribute attr;
-	Datum		best,
-				guess1,
-				guess2,
-				max,
-				min;
-	int			best_len,
-				guess1_len,
-				guess2_len,
-				max_len,
-				min_len;
-	long		best_cnt,
-				guess1_cnt,
-				guess1_hits,
-				guess2_hits,
-				null_cnt,
-				nonnull_cnt,
-				max_cnt,
-				min_cnt;
-	FmgrInfo	f_cmpeq,
-				f_cmplt,
-				f_cmpgt;
-	Oid			op_cmplt;
-	regproc		outfunc;
-	Oid			typelem;
-	bool		initialized;
-} VacAttrStats;
-
-typedef struct VRelListData
-{
-	Oid			vrl_relid;
-	struct VRelListData *vrl_next;
-} VRelListData;
-
-typedef VRelListData *VRelList;
-
-typedef struct VTupleLinkData
-{
-	ItemPointerData new_tid;
-	ItemPointerData this_tid;
-} VTupleLinkData;
-
-typedef VTupleLinkData *VTupleLink;
-
-typedef struct VTupleMoveData
-{
-	ItemPointerData tid;		/* tuple ID */
-	VacPage		vacpage;		/* where to move */
-	bool		cleanVpd;		/* clean vacpage before using */
-} VTupleMoveData;
-
-typedef VTupleMoveData *VTupleMove;
-
-typedef struct VRelStats
-{
-	Oid			relid;
-	int			num_tuples;
-	int			num_pages;
-	Size		min_tlen;
-	Size		max_tlen;
-	bool		hasindex;
-	int			num_vtlinks;
-	VTupleLink	vtlinks;
-} VRelStats;
-
-extern bool VacuumRunning;
-
-extern void vc_abort(void);
-extern void vacuum(char *vacrel, bool verbose, bool analyze, List *anal_cols);
-extern void analyze_rel(Oid relid, List *anal_cols2, int MESSAGE_LEVEL);
-
-#define ATTNVALS_SCALE	1000000000		/* XXX so it can act as a float4 */
+/* in commands/vacuum.c */
+extern void vacuum(VacuumStmt *vacstmt);
+extern void vac_update_relstats(Oid relid, long num_pages, double num_tuples,
+								bool hasindex);
+/* in commands/analyze.c */
+extern void analyze_rel(Oid relid, VacuumStmt *vacstmt);
 
 #endif	 /* VACUUM_H */
diff --git a/src/include/config.h.in b/src/include/config.h.in
index 0d989dbbb3..01593a4ce9 100644
--- a/src/include/config.h.in
+++ b/src/include/config.h.in
@@ -8,7 +8,7 @@
  * or in config.h afterwards.  Of course, if you edit config.h, then your
  * changes will be overwritten the next time you run configure.
  *
- * $Id: config.h.in,v 1.162 2001/04/14 22:55:02 petere Exp $
+ * $Id: config.h.in,v 1.163 2001/05/07 00:43:25 tgl Exp $
  */
 
 #ifndef CONFIG_H
@@ -157,6 +157,11 @@
 #define FUNC_MAX_ARGS		INDEX_MAX_KEYS
 
 /*
+ * System default value for pg_attribute.attstattarget
+ */
+#define DEFAULT_ATTSTATTARGET	10
+
+/*
  * Define this to make libpgtcl's "pg_result -assign" command process C-style
  * backslash sequences in returned tuple data and convert Postgres array
  * attributes into Tcl lists.  CAUTION: this conversion is *wrong* unless
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 2cf9378cf1..0967bef24b 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: execnodes.h,v 1.57 2001/03/22 04:00:50 momjian Exp $
+ * $Id: execnodes.h,v 1.58 2001/05/07 00:43:25 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -628,7 +628,6 @@ typedef struct GroupState
  *	 SortState information
  *
  *		sort_Done		indicates whether sort has been performed yet
- *		sort_Keys		scan key structures describing the sort keys
  *		tuplesortstate	private state of tuplesort.c
  * ----------------
  */
@@ -636,7 +635,6 @@ typedef struct SortState
 {
 	CommonScanState csstate;	/* its first field is NodeTag */
 	bool		sort_Done;
-	ScanKey		sort_Keys;
 	void	   *tuplesortstate;
 } SortState;
 
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 1614d787bc..63b1b1046a 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: parsenodes.h,v 1.126 2001/03/23 04:49:56 momjian Exp $
+ * $Id: parsenodes.h,v 1.127 2001/05/07 00:43:25 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -118,11 +118,12 @@ typedef struct AlterTableStmt
 	NodeTag		type;
 	char		subtype;		/*------------
 								 * 	A = add column
-								 *	T = alter column
+								 *	T = alter column default
+								 *	S = alter column statistics
 								 *	D = drop column
 								 *	C = add constraint
 								 *	X = drop constraint
-								 *	E = add toast table,
+								 *	E = create toast table
 								 *	U = change owner
 								 *------------
 								 */
@@ -690,16 +691,20 @@ typedef struct ClusterStmt
 } ClusterStmt;
 
 /* ----------------------
- *		Vacuum Statement
+ *		Vacuum and Analyze Statements
+ *
+ * Even though these are nominally two statements, it's convenient to use
+ * just one node type for both.
  * ----------------------
  */
 typedef struct VacuumStmt
 {
 	NodeTag		type;
-	bool		verbose;		/* print status info */
-	bool		analyze;		/* analyze data */
-	char	   *vacrel;			/* table to vacuum */
-	List	   *va_spec;		/* columns to analyse */
+	bool		vacuum;			/* do VACUUM step */
+	bool		analyze;		/* do ANALYZE step */
+	bool		verbose;		/* print progress info */
+	char	   *vacrel;			/* name of single table to process, or NULL */
+	List	   *va_cols;		/* list of column names, or NIL for all */
 } VacuumStmt;
 
 /* ----------------------
diff --git a/src/include/nodes/primnodes.h b/src/include/nodes/primnodes.h
index 3ae8e09f57..9e69ed6099 100644
--- a/src/include/nodes/primnodes.h
+++ b/src/include/nodes/primnodes.h
@@ -10,7 +10,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: primnodes.h,v 1.53 2001/03/22 04:00:52 momjian Exp $
+ * $Id: primnodes.h,v 1.54 2001/05/07 00:43:25 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -45,8 +45,8 @@ typedef struct FunctionCache *FunctionCachePtr;
  * reskey and reskeyop are the execution-time representation of sorting.
  * reskey must be zero in any non-sort-key item.  The reskey of sort key
  * targetlist items for a sort plan node is 1,2,...,n for the n sort keys.
- * The reskeyop of each such targetlist item is the sort operator's
- * regproc OID.  reskeyop will be zero in non-sort-key items.
+ * The reskeyop of each such targetlist item is the sort operator's OID.
+ * reskeyop will be zero in non-sort-key items.
  *
  * Both reskey and reskeyop are typically zero during parse/plan stages.
  * The executor does not pay any attention to ressortgroupref.
@@ -62,7 +62,7 @@ typedef struct Resdom
 	Index		ressortgroupref;
 	/* nonzero if referenced by a sort/group clause */
 	Index		reskey;			/* order of key in a sort (for those > 0) */
-	Oid			reskeyop;		/* sort operator's regproc Oid */
+	Oid			reskeyop;		/* sort operator's Oid */
 	bool		resjunk;		/* set to true to eliminate the attribute
 								 * from final target list */
 } Resdom;
diff --git a/src/include/nodes/relation.h b/src/include/nodes/relation.h
index f643ef8796..c76d9b4af7 100644
--- a/src/include/nodes/relation.h
+++ b/src/include/nodes/relation.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: relation.h,v 1.54 2001/03/22 04:00:53 momjian Exp $
+ * $Id: relation.h,v 1.55 2001/05/07 00:43:26 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -512,8 +512,8 @@ typedef struct RestrictInfo
 	Oid			hashjoinoperator;		/* copy of clause operator */
 
 	/* cache space for hashclause processing; -1 if not yet set */
-	Selectivity left_dispersion;/* dispersion of left side */
-	Selectivity right_dispersion;		/* dispersion of right side */
+	Selectivity left_bucketsize;		/* avg bucketsize of left side */
+	Selectivity right_bucketsize;		/* avg bucketsize of right side */
 } RestrictInfo;
 
 /*
diff --git a/src/include/optimizer/cost.h b/src/include/optimizer/cost.h
index 5caa576f0c..cbf6df063a 100644
--- a/src/include/optimizer/cost.h
+++ b/src/include/optimizer/cost.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: cost.h,v 1.38 2001/02/16 00:03:05 tgl Exp $
+ * $Id: cost.h,v 1.39 2001/05/07 00:43:26 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -64,7 +64,8 @@ extern void cost_mergejoin(Path *path, Path *outer_path, Path *inner_path,
 			   List *restrictlist,
 			   List *outersortkeys, List *innersortkeys);
 extern void cost_hashjoin(Path *path, Path *outer_path, Path *inner_path,
-			  List *restrictlist, Selectivity innerdispersion);
+			  List *restrictlist, Selectivity innerbucketsize);
+extern Selectivity estimate_hash_bucketsize(Query *root, Var *var);
 extern Cost cost_qual_eval(List *quals);
 extern void set_baserel_size_estimates(Query *root, RelOptInfo *rel);
 extern void set_joinrel_size_estimates(Query *root, RelOptInfo *rel,
diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h
index 5b71eded86..0839feb4b2 100644
--- a/src/include/optimizer/pathnode.h
+++ b/src/include/optimizer/pathnode.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: pathnode.h,v 1.35 2001/03/22 04:00:54 momjian Exp $
+ * $Id: pathnode.h,v 1.36 2001/05/07 00:43:26 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -59,7 +59,7 @@ extern HashPath *create_hashjoin_path(RelOptInfo *joinrel,
 					 Path *inner_path,
 					 List *restrict_clauses,
 					 List *hashclauses,
-					 Selectivity innerdispersion);
+					 Selectivity innerbucketsize);
 
 /*
  * prototypes for relnode.c
diff --git a/src/include/utils/lsyscache.h b/src/include/utils/lsyscache.h
index f1c4aff1c8..6b35deed28 100644
--- a/src/include/utils/lsyscache.h
+++ b/src/include/utils/lsyscache.h
@@ -6,7 +6,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: lsyscache.h,v 1.30 2001/03/22 04:01:13 momjian Exp $
+ * $Id: lsyscache.h,v 1.31 2001/05/07 00:43:26 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -21,8 +21,6 @@ extern AttrNumber get_attnum(Oid relid, char *attname);
 extern Oid	get_atttype(Oid relid, AttrNumber attnum);
 extern bool get_attisset(Oid relid, char *attname);
 extern int32 get_atttypmod(Oid relid, AttrNumber attnum);
-extern double get_attdispersion(Oid relid, AttrNumber attnum,
-				  double min_estimate);
 extern RegProcedure get_opcode(Oid opno);
 extern char *get_opname(Oid opno);
 extern bool op_mergejoinable(Oid opno, Oid ltype, Oid rtype,
@@ -41,6 +39,14 @@ extern bool get_typbyval(Oid typid);
 extern void get_typlenbyval(Oid typid, int16 *typlen, bool *typbyval);
 extern char get_typstorage(Oid typid);
 extern Datum get_typdefault(Oid typid);
+extern bool get_attstatsslot(HeapTuple statstuple,
+							 Oid atttype, int32 atttypmod,
+							 int reqkind, Oid reqop,
+							 Datum **values, int *nvalues,
+							 float4 **numbers, int *nnumbers);
+extern void free_attstatsslot(Oid atttype,
+							  Datum *values, int nvalues,
+							  float4 *numbers, int nnumbers);
 
 #define TypeIsToastable(typid)	(get_typstorage(typid) != 'p')
 
diff --git a/src/include/utils/syscache.h b/src/include/utils/syscache.h
index 8d4e2ae62c..342f7bf8a5 100644
--- a/src/include/utils/syscache.h
+++ b/src/include/utils/syscache.h
@@ -9,7 +9,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: syscache.h,v 1.29 2001/03/22 04:01:14 momjian Exp $
+ * $Id: syscache.h,v 1.30 2001/05/07 00:43:26 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -53,7 +53,7 @@
 #define RULEOID			22
 #define SHADOWNAME		23
 #define SHADOWSYSID		24
-#define STATRELID		25
+#define STATRELATT		25
 #define TYPENAME		26
 #define TYPEOID			27
 
diff --git a/src/include/utils/tuplesort.h b/src/include/utils/tuplesort.h
index 7f273776c3..001761796e 100644
--- a/src/include/utils/tuplesort.h
+++ b/src/include/utils/tuplesort.h
@@ -13,7 +13,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: tuplesort.h,v 1.6 2001/01/24 19:43:29 momjian Exp $
+ * $Id: tuplesort.h,v 1.7 2001/05/07 00:43:26 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -36,8 +36,9 @@ typedef struct Tuplesortstate Tuplesortstate;
  */
 
 extern Tuplesortstate *tuplesort_begin_heap(TupleDesc tupDesc,
-					 int nkeys, ScanKey keys,
-					 bool randomAccess);
+					  int nkeys,
+					  Oid *sortOperators, AttrNumber *attNums,
+					  bool randomAccess);
 extern Tuplesortstate *tuplesort_begin_index(Relation indexRel,
 					  bool enforceUnique,
 					  bool randomAccess);
@@ -75,4 +76,19 @@ extern void tuplesort_rescan(Tuplesortstate *state);
 extern void tuplesort_markpos(Tuplesortstate *state);
 extern void tuplesort_restorepos(Tuplesortstate *state);
 
+/*
+ * This routine selects an appropriate sorting function to implement
+ * a sort operator as efficiently as possible.
+ */
+typedef enum
+{
+	SORTFUNC_LT,				/* raw "<" operator */
+	SORTFUNC_CMP,				/* -1 / 0 / 1 three-way comparator */
+	SORTFUNC_REVCMP				/* 1 / 0 / -1 (reversed) 3-way comparator */
+} SortFunctionKind;
+
+extern void SelectSortFunction(Oid sortOperator,
+							   RegProcedure *sortFunction,
+							   SortFunctionKind *kind);
+
 #endif	 /* TUPLESORT_H */