Clean up bogosities in pg_opclass, pg_amop, pg_amproc. There are amproc

entries now for int8 and network hash indexes.  int24_ops and int42_ops
are gone.  pg_opclass no longer contains multiple entries claiming to be
the default opclass for the same datatype.  opr_sanity regress test
extended to catch errors like these in the future.

doc/src/sgml/indices.sgml

@@ -130,17 +130,6 @@
       <literal>bigbox_ops</literal>.
      </para>
     </listitem>
-
-    <listitem>
-     <para>
-      The <literal>int24_ops</literal>
-      operator class is useful for constructing indices on int2 data, and
-      doing comparisons against int4 data in query qualifications.
-      Similarly, <literal>int42_ops</literal>
-      support indices on int4 data that is to be compared against int2 data
-      in queries.
-     </para>
-    </listitem>
    </itemizedlist>
   </para>
 

doc/src/sgml/ref/create_index.sgml

 <!--
-$Header: /cvsroot/pgsql/doc/src/sgml/ref/create_index.sgml,v 1.13 2000/05/02 20:02:03 thomas Exp $
+$Header: /cvsroot/pgsql/doc/src/sgml/ref/create_index.sgml,v 1.14 2000/06/19 03:54:15 tgl Exp $
 Postgres documentation
 -->
 
@@ -312,17 +312,6 @@ ERROR: Cannot create index: 'index_name' already exists.
       <literal>bigbox_ops</literal>.
      </para>
     </listitem>
-
-    <listitem>
-     <para>
-      The <literal>int24_ops</literal>
-      operator class is useful for constructing indices on int2 data, and
-      doing comparisons against int4 data in query qualifications.
-      Similarly, <literal>int42_ops</literal>
-      support indices on int4 data that is to be compared against int2 data
-      in queries.
-     </para>
-    </listitem>
    </itemizedlist>
   </para>
 

src/backend/access/hash/hashfunc.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/access/hash/hashfunc.c,v 1.26 2000/06/05 07:28:35 tgl Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/access/hash/hashfunc.c,v 1.27 2000/06/19 03:54:17 tgl Exp $
  *
  * NOTES
  *	  These functions are stored in pg_amproc.	For each operator class
@@ -21,10 +21,17 @@
 
 #include "access/hash.h"
 
+
+Datum
+hashchar(PG_FUNCTION_ARGS)
+{
+	PG_RETURN_UINT32(~ ((uint32) PG_GETARG_CHAR(0)));
+}
+
 Datum
 hashint2(PG_FUNCTION_ARGS)
 {
-	PG_RETURN_UINT32((uint32) ~ PG_GETARG_INT16(0));
+	PG_RETURN_UINT32(~ ((uint32) PG_GETARG_INT16(0)));
 }
 
 Datum
@@ -37,116 +44,37 @@ Datum
 hashint8(PG_FUNCTION_ARGS)
 {
 	/* we just use the low 32 bits... */
-	PG_RETURN_UINT32(~((uint32) PG_GETARG_INT64(0)));
+	PG_RETURN_UINT32(~ ((uint32) PG_GETARG_INT64(0)));
+}
+
+Datum
+hashoid(PG_FUNCTION_ARGS)
+{
+	PG_RETURN_UINT32(~ ((uint32) PG_GETARG_OID(0)));
 }
 
-/* Hash function from Chris Torek. */
 Datum
 hashfloat4(PG_FUNCTION_ARGS)
 {
 	float4		key = PG_GETARG_FLOAT4(0);
-	char	   *kp = (char *) &key;
-	int			len = sizeof(key);
-	int			loop;
-	uint32		h;
-
-#define HASH4a	 h = (h << 5) - h + *kp++;
-#define HASH4b	 h = (h << 5) + h + *kp++;
-#define HASH4 HASH4b
-
-	h = 0;
-	/*
-	 * This is a tad silly, given that we expect len = 4, but a smart
-	 * compiler should be able to eliminate the redundant code...
-	 */
-	loop = (len + 8 - 1) >> 3;
 
-	switch (len & (8 - 1))
-	{
-		case 0:
-			do
-			{					/* All fall throughs */
-				HASH4;
-		case 7:
-				HASH4;
-		case 6:
-				HASH4;
-		case 5:
-				HASH4;
-		case 4:
-				HASH4;
-		case 3:
-				HASH4;
-		case 2:
-				HASH4;
-		case 1:
-				HASH4;
-			} while (--loop);
-	}
-	PG_RETURN_UINT32(h);
+	return hash_any((char *) &key, sizeof(key));
 }
 
 Datum
 hashfloat8(PG_FUNCTION_ARGS)
 {
 	float8		key = PG_GETARG_FLOAT8(0);
-	char	   *kp = (char *) &key;
-	int			len = sizeof(key);
-	int			loop;
-	uint32		h;
 
-#define HASH4a	 h = (h << 5) - h + *kp++;
-#define HASH4b	 h = (h << 5) + h + *kp++;
-#define HASH4 HASH4b
-
-	h = 0;
-	/*
-	 * This is a tad silly, given that we expect len = 8, but a smart
-	 * compiler should be able to eliminate the redundant code...
-	 */
-	loop = (len + 8 - 1) >> 3;
-
-	switch (len & (8 - 1))
-	{
-		case 0:
-			do
-			{					/* All fall throughs */
-				HASH4;
-		case 7:
-				HASH4;
-		case 6:
-				HASH4;
-		case 5:
-				HASH4;
-		case 4:
-				HASH4;
-		case 3:
-				HASH4;
-		case 2:
-				HASH4;
-		case 1:
-				HASH4;
-			} while (--loop);
-	}
-	PG_RETURN_UINT32(h);
-}
-
-Datum
-hashoid(PG_FUNCTION_ARGS)
-{
-	PG_RETURN_UINT32(~(uint32) PG_GETARG_OID(0));
+	return hash_any((char *) &key, sizeof(key));
 }
 
 Datum
 hashoidvector(PG_FUNCTION_ARGS)
 {
 	Oid		   *key = (Oid *) PG_GETARG_POINTER(0);
-	int			i;
-	uint32		result = 0;
 
-	for (i = INDEX_MAX_KEYS; --i >= 0;)
-		result = (result << 1) ^ (~(uint32) key[i]);
-	PG_RETURN_UINT32(result);
+	return hash_any((char *) key, INDEX_MAX_KEYS * sizeof(Oid));
 }
 
 /*
@@ -158,69 +86,53 @@ Datum
 hashint2vector(PG_FUNCTION_ARGS)
 {
 	int16	   *key = (int16 *) PG_GETARG_POINTER(0);
-	int			i;
-	uint32		result = 0;
 
-	for (i = INDEX_MAX_KEYS; --i >= 0;)
-		result = (result << 1) ^ (~(uint32) key[i]);
-	PG_RETURN_UINT32(result);
+	return hash_any((char *) key, INDEX_MAX_KEYS * sizeof(int16));
 }
 
-
-#define PRIME1			37
-#define PRIME2			1048583
-
 Datum
-hashchar(PG_FUNCTION_ARGS)
+hashname(PG_FUNCTION_ARGS)
 {
-	uint32		h;
-
-	/* Convert char to integer */
-	h = (PG_GETARG_CHAR(0) - ' ');
-	h %= PRIME2;
+	char	   *key = NameStr(* PG_GETARG_NAME(0));
 
-	PG_RETURN_UINT32(h);
+	return hash_any((char *) key, NAMEDATALEN);
 }
 
+/*
+ * hashvarlena() can be used for any varlena datatype in which there are
+ * no non-significant bits, ie, distinct bitpatterns never compare as equal.
+ */
 Datum
-hashname(PG_FUNCTION_ARGS)
+hashvarlena(PG_FUNCTION_ARGS)
 {
-	char	   *key = NameStr(* PG_GETARG_NAME(0));
-	int			len = NAMEDATALEN;
-	uint32		h;
+	struct varlena *key = PG_GETARG_VARLENA_P(0);
 
-	h = 0;
-	/* Convert string to integer */
-	while (len--)
-		h = h * PRIME1 ^ (*key++ - ' ');
-	h %= PRIME2;
-
-	PG_RETURN_UINT32(h);
+	return hash_any(VARDATA(key), VARSIZE(key) - VARHDRSZ);
 }
 
+
 /*
+ * hash_any --- compute a hash function for any specified chunk of memory
+ *
+ * This can be used as the underlying hash function for any pass-by-reference
+ * data type in which there are no non-significant bits.
+ *
  * (Comment from the original db3 hashing code: )
  *
  * "This is INCREDIBLY ugly, but fast.  We break the string up into 8 byte
  * units.  On the first time through the loop we get the 'leftover bytes'
- * (strlen % 8).  On every other iteration, we perform 8 HASHC's so we handle
+ * (strlen % 8).  On every later iteration, we perform 8 HASHC's so we handle
  * all 8 bytes.  Essentially, this saves us 7 cmp & branch instructions.  If
  * this routine is heavily used enough, it's worth the ugly coding.
  *
  * "OZ's original sdbm hash"
  */
 Datum
-hashtext(PG_FUNCTION_ARGS)
+hash_any(char *keydata, int keylen)
 {
-	text	   *key = PG_GETARG_TEXT_P(0);
-	int			keylen;
-	char	   *keydata;
 	uint32		n;
 	int			loop;
 
-	keydata = VARDATA(key);
-	keylen = VARSIZE(key) - VARHDRSZ;
-
 #define HASHC	n = *keydata++ + 65599 * n
 
 	n = 0;
@@ -251,5 +163,8 @@ hashtext(PG_FUNCTION_ARGS)
 				} while (--loop);
 		}
 	}
+
+#undef HASHC
+
 	PG_RETURN_UINT32(n);
 }

src/backend/access/nbtree/nbtcompare.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/access/nbtree/nbtcompare.c,v 1.37 2000/06/15 03:31:54 momjian Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/access/nbtree/nbtcompare.c,v 1.38 2000/06/19 03:54:22 tgl Exp $
  *
  * NOTES
  *
@@ -81,34 +81,6 @@ btint8cmp(PG_FUNCTION_ARGS)
 		PG_RETURN_INT32(-1);
 }
 
-Datum
-btint24cmp(PG_FUNCTION_ARGS)
-{
-	int16		a = PG_GETARG_INT16(0);
-	int32		b = PG_GETARG_INT32(1);
-
-	if (a > b)
-		PG_RETURN_INT32(1);
-	else if (a == b)
-		PG_RETURN_INT32(0);
-	else
-		PG_RETURN_INT32(-1);
-}
-
-Datum
-btint42cmp(PG_FUNCTION_ARGS)
-{
-	int32		a = PG_GETARG_INT32(0);
-	int16		b = PG_GETARG_INT16(1);
-
-	if (a > b)
-		PG_RETURN_INT32(1);
-	else if (a == b)
-		PG_RETURN_INT32(0);
-	else
-		PG_RETURN_INT32(-1);
-}
-
 Datum
 btfloat4cmp(PG_FUNCTION_ARGS)
 {

src/backend/utils/adt/date.c

@@ -8,23 +8,27 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/utils/adt/date.c,v 1.47 2000/06/15 04:10:23 momjian Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/utils/adt/date.c,v 1.48 2000/06/19 03:54:27 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
-#include <limits.h>
-#include <time.h>
 
 #include "postgres.h"
+
+#include <limits.h>
+#include <time.h>
 #ifdef HAVE_FLOAT_H
 #include <float.h>
 #endif
+
+#include "access/hash.h"
 #include "miscadmin.h"
 #include "utils/date.h"
 #include "utils/nabstime.h"
 
-static int
-	date2tm(DateADT dateVal, int *tzp, struct tm * tm, double *fsec, char **tzn);
+
+static int date2tm(DateADT dateVal, int *tzp, struct tm * tm,
+				   double *fsec, char **tzn);
 
 
 /*****************************************************************************
@@ -762,6 +766,22 @@ timetz_cmp(PG_FUNCTION_ARGS)
 	PG_RETURN_INT32(0);
 }
 
+/*
+ * timetz, being an unusual size, needs a specialized hash function.
+ */
+Datum
+timetz_hash(PG_FUNCTION_ARGS)
+{
+	TimeTzADT   *key = PG_GETARG_TIMETZADT_P(0);
+
+	/*
+	 * Specify hash length as sizeof(double) + sizeof(int4), not as
+	 * sizeof(TimeTzADT), so that any garbage pad bytes in the structure
+	 * won't be included in the hash!
+	 */
+	return hash_any((char *) key, sizeof(double) + sizeof(int4));
+}
+
 Datum
 timetz_larger(PG_FUNCTION_ARGS)
 {

src/backend/utils/adt/timestamp.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/utils/adt/timestamp.c,v 1.29 2000/06/09 01:11:09 tgl Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/utils/adt/timestamp.c,v 1.30 2000/06/19 03:54:27 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -29,6 +29,7 @@
 #include <sys/timeb.h>
 #endif
 
+#include "access/hash.h"
 #include "access/xact.h"
 #include "miscadmin.h"
 #include "utils/builtins.h"
@@ -816,6 +817,22 @@ interval_cmp(PG_FUNCTION_ARGS)
 	PG_RETURN_INT32((span1 < span2) ? -1 : (span1 > span2) ? 1 : 0);
 }
 
+/*
+ * interval, being an unusual size, needs a specialized hash function.
+ */
+Datum
+interval_hash(PG_FUNCTION_ARGS)
+{
+	Interval   *key = PG_GETARG_INTERVAL_P(0);
+
+	/*
+	 * Specify hash length as sizeof(double) + sizeof(int4), not as
+	 * sizeof(Interval), so that any garbage pad bytes in the structure
+	 * won't be included in the hash!
+	 */
+	return hash_any((char *) key, sizeof(double) + sizeof(int4));
+}
+
 /* overlaps_timestamp()
  * Implements the SQL92 OVERLAPS operator.
  * Algorithm from Date and Darwen, 1997

src/backend/utils/adt/varchar.c

@@ -8,13 +8,14 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/utils/adt/varchar.c,v 1.65 2000/06/15 03:32:29 momjian Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/utils/adt/varchar.c,v 1.66 2000/06/19 03:54:28 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
 
 #include "postgres.h"
 
+#include "access/hash.h"
 #include "catalog/pg_type.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
@@ -797,94 +798,19 @@ varcharcmp(char *arg1, char *arg2)
 		return (int32) (cmp);
 }
 
-/*****************************************************************************
- * Hash functions (modified from hashtext in access/hash/hashfunc.c)
- *****************************************************************************/
-
-uint32
-hashbpchar(struct varlena * key)
+/*
+ * bpchar needs a specialized hash function because we want to ignore
+ * trailing blanks in comparisons.  (varchar can use plain hashvarlena.)
+ */
+Datum
+hashbpchar(PG_FUNCTION_ARGS)
 {
-	int			keylen;
+	BpChar	   *key = PG_GETARG_BPCHAR_P(0);
 	char	   *keydata;
-	uint32		n;
-	int			loop;
-
-	keydata = VARDATA(key);
-	keylen = bcTruelen((char *) key);
-
-#define HASHC	n = *keydata++ + 65599 * n
-
-	n = 0;
-	if (keylen > 0)
-	{
-		loop = (keylen + 8 - 1) >> 3;
-
-		switch (keylen & (8 - 1))
-		{
-			case 0:
-				do
-				{				/* All fall throughs */
-					HASHC;
-			case 7:
-					HASHC;
-			case 6:
-					HASHC;
-			case 5:
-					HASHC;
-			case 4:
-					HASHC;
-			case 3:
-					HASHC;
-			case 2:
-					HASHC;
-			case 1:
-					HASHC;
-				} while (--loop);
-		}
-	}
-	return n;
-}
-
-uint32
-hashvarchar(struct varlena * key)
-{
 	int			keylen;
-	char	   *keydata;
-	uint32		n;
-	int			loop;
 
 	keydata = VARDATA(key);
-	keylen = VARSIZE(key) - VARHDRSZ;
-
-#define HASHC	n = *keydata++ + 65599 * n
+	keylen = bcTruelen((char *) key);
 
-	n = 0;
-	if (keylen > 0)
-	{
-		loop = (keylen + 8 - 1) >> 3;
-
-		switch (keylen & (8 - 1))
-		{
-			case 0:
-				do
-				{				/* All fall throughs */
-					HASHC;
-			case 7:
-					HASHC;
-			case 6:
-					HASHC;
-			case 5:
-					HASHC;
-			case 4:
-					HASHC;
-			case 3:
-					HASHC;
-			case 2:
-					HASHC;
-			case 1:
-					HASHC;
-				} while (--loop);
-		}
-	}
-	return n;
+	return hash_any(keydata, keylen);
 }

src/backend/utils/cache/catcache.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/utils/cache/catcache.c,v 1.66 2000/06/17 04:56:32 tgl Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/utils/cache/catcache.c,v 1.67 2000/06/19 03:54:31 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -104,7 +104,7 @@ GetCCHashFunc(Oid keytype)
 		case INT4OID:
 			return hashint4;
 		case TEXTOID:
-			return hashtext;
+			return hashvarlena;
 		case REGPROCOID:
 		case OIDOID:
 			return hashoid;

src/include/access/hash.h

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: hash.h,v 1.33 2000/06/13 07:35:17 tgl Exp $
+ * $Id: hash.h,v 1.34 2000/06/19 03:54:35 tgl Exp $
  *
  * NOTES
  *		modeled after Margo Seltzer's hash implementation for unix.
@@ -264,17 +264,18 @@ extern Datum hashdelete(PG_FUNCTION_ARGS);
  * NOTE: some of these are also used by catcache operations, without
  * any direct connection to hash indexes.
  */
+extern Datum hashchar(PG_FUNCTION_ARGS);
 extern Datum hashint2(PG_FUNCTION_ARGS);
 extern Datum hashint4(PG_FUNCTION_ARGS);
 extern Datum hashint8(PG_FUNCTION_ARGS);
+extern Datum hashoid(PG_FUNCTION_ARGS);
 extern Datum hashfloat4(PG_FUNCTION_ARGS);
 extern Datum hashfloat8(PG_FUNCTION_ARGS);
-extern Datum hashoid(PG_FUNCTION_ARGS);
 extern Datum hashoidvector(PG_FUNCTION_ARGS);
 extern Datum hashint2vector(PG_FUNCTION_ARGS);
-extern Datum hashchar(PG_FUNCTION_ARGS);
-extern Datum hashtext(PG_FUNCTION_ARGS);
 extern Datum hashname(PG_FUNCTION_ARGS);
+extern Datum hashvarlena(PG_FUNCTION_ARGS);
+extern Datum hash_any(char *keydata, int keylen);
 
 
 /* private routines */

src/include/catalog/catversion.h

@@ -37,7 +37,7 @@
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: catversion.h,v 1.32 2000/06/16 05:27:01 tgl Exp $
+ * $Id: catversion.h,v 1.33 2000/06/19 03:54:45 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -53,6 +53,6 @@
  */
 
 /*							yyyymmddN */
-#define CATALOG_VERSION_NO	200006161
+#define CATALOG_VERSION_NO	200006181
 
 #endif

src/include/catalog/pg_amop.h

@@ -8,7 +8,7 @@
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: pg_amop.h,v 1.33 2000/04/12 17:16:28 momjian Exp $
+ * $Id: pg_amop.h,v 1.34 2000/06/19 03:54:45 tgl Exp $
  *
  * NOTES
  *	 the genbki.sh script reads this file and generates .bki
@@ -33,10 +33,11 @@
  */
 CATALOG(pg_amop)
 {
-	Oid			amopid;
-	Oid			amopclaid;
-	Oid			amopopr;
-	int2		amopstrategy;
+	Oid			amopid;			/* an index access method */
+	Oid			amopclaid;		/* an index opclass */
+	Oid			amopopr;		/* the operator OID to use */
+	int2		amopstrategy;	/* one of the strategy numbers defined
+								 * by the AM */
 } FormData_pg_amop;
 
 /* ----------------
@@ -102,16 +103,21 @@ DATA(insert OID = 0 (  402 434 489 8 ));
 
 /*
  *	rtree circle_ops (supports circles)
+ *
+ * XXX Diked out 2000-06-18 by tgl.  Since we have no rtree support functions
+ * (union, intersection, size) for circles, we can't actually build rtree
+ * indexes on circles.  These can be put back in someday if anyone ever
+ * writes such functions.
  */
 
-DATA(insert OID = 0 (  402 714 1506 1 ));
-DATA(insert OID = 0 (  402 714 1507 2 ));
-DATA(insert OID = 0 (  402 714 1513 3 ));
-DATA(insert OID = 0 (  402 714 1508 4 ));
-DATA(insert OID = 0 (  402 714 1509 5 ));
-DATA(insert OID = 0 (  402 714 1512 6 ));
-DATA(insert OID = 0 (  402 714 1511 7 ));
-DATA(insert OID = 0 (  402 714 1510 8 ));
+/* DATA(insert OID = 0 (  402 714 1506 1 )); */
+/* DATA(insert OID = 0 (  402 714 1507 2 )); */
+/* DATA(insert OID = 0 (  402 714 1513 3 )); */
+/* DATA(insert OID = 0 (  402 714 1508 4 )); */
+/* DATA(insert OID = 0 (  402 714 1509 5 )); */
+/* DATA(insert OID = 0 (  402 714 1512 6 )); */
+/* DATA(insert OID = 0 (  402 714 1511 7 )); */
+/* DATA(insert OID = 0 (  402 714 1510 8 )); */
 
 /*
  *	nbtree int2_ops
@@ -133,26 +139,6 @@ DATA(insert OID = 0 (  403 423 670 3 ));
 DATA(insert OID = 0 (  403 423 675 4 ));
 DATA(insert OID = 0 (  403 423 674 5 ));
 
-/*
- *	nbtree int24_ops
- */
-
-DATA(insert OID = 0 (  403 424 534 1 ));
-DATA(insert OID = 0 (  403 424 540 2 ));
-DATA(insert OID = 0 (  403 424 532 3 ));
-DATA(insert OID = 0 (  403 424 542 4 ));
-DATA(insert OID = 0 (  403 424 536 5 ));
-
-/*
- *	nbtree int42_ops
- */
-
-DATA(insert OID = 0 (  403 425 535 1 ));
-DATA(insert OID = 0 (  403 425 541 2 ));
-DATA(insert OID = 0 (  403 425 533 3 ));
-DATA(insert OID = 0 (  403 425 543 4 ));
-DATA(insert OID = 0 (  403 425 537 5 ));
-
 /*
  *	nbtree int4_ops
  */

src/include/catalog/pg_amproc.h

 /*-------------------------------------------------------------------------
  *
  * pg_amproc.h
- *	  definition of the system "amproc" relation (pg_amproce)
+ *	  definition of the system "amproc" relation (pg_amproc)
  *	  along with the relation's initial contents.  The amproc
- *	  catalog is used to store procedures used by indexed access
+ *	  catalog is used to store procedures used by index access
  *	  methods that aren't associated with operators.
  *
  *
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: pg_amproc.h,v 1.22 2000/03/14 23:06:43 thomas Exp $
+ * $Id: pg_amproc.h,v 1.23 2000/06/19 03:54:45 tgl Exp $
  *
  * NOTES
  *	  the genbki.sh script reads this file and generates .bki
@@ -35,10 +35,10 @@
  */
 CATALOG(pg_amproc)
 {
-	Oid			amid;
-	Oid			amopclaid;
-	Oid			amproc;
-	int2		amprocnum;
+	Oid			amid;			/* the access method this proc is for */
+	Oid			amopclaid;		/* the opclass this proc is for */
+	Oid			amproc;			/* OID of the proc */
+	int2		amprocnum;		/* support procedure index */
 } FormData_pg_amproc;
 
 /* ----------------
@@ -78,8 +78,6 @@ DATA(insert OID = 0 (402  434  199 3));
 /* btree */
 DATA(insert OID = 0 (403  421  350 1));
 DATA(insert OID = 0 (403  423  355 1));
-DATA(insert OID = 0 (403  424  353 1));
-DATA(insert OID = 0 (403  425  352 1));
 DATA(insert OID = 0 (403  426  351 1));
 DATA(insert OID = 0 (403  427  356 1));
 DATA(insert OID = 0 (403  428  354 1));
@@ -114,13 +112,17 @@ DATA(insert OID = 0 (405  428  451 1));
 DATA(insert OID = 0 (405  429  454 1));
 DATA(insert OID = 0 (405  431  456 1));
 DATA(insert OID = 0 (405  435  457 1));
+DATA(insert OID = 0 (405  652  456 1));
+DATA(insert OID = 0 (405  754  949 1));
+DATA(insert OID = 0 (405  810  456 1));
+DATA(insert OID = 0 (405  935  456 1));
 DATA(insert OID = 0 (405 1076 1080 1));
-DATA(insert OID = 0 (405 1077 1081 1));
+DATA(insert OID = 0 (405 1077  456 1));
 DATA(insert OID = 0 (405 1114  450 1));
 DATA(insert OID = 0 (405 1115  452 1));
 DATA(insert OID = 0 (405 1181  455 1));
 DATA(insert OID = 0 (405 1312  452 1));
-DATA(insert OID = 0 (405 1313  452 1));
-DATA(insert OID = 0 (405 1399  452 1));
+DATA(insert OID = 0 (405 1313 1697 1));
+DATA(insert OID = 0 (405 1399 1696 1));
 
 #endif	 /* PG_AMPROC_H */

src/include/catalog/pg_opclass.h

@@ -8,7 +8,7 @@
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: pg_opclass.h,v 1.32 2000/04/12 17:16:28 momjian Exp $
+ * $Id: pg_opclass.h,v 1.33 2000/06/19 03:54:45 tgl Exp $
  *
  * NOTES
  *	  the genbki.sh script reads this file and generates .bki
@@ -34,8 +34,8 @@
 
 CATALOG(pg_opclass)
 {
-	NameData	opcname;
-	Oid			opcdeftype;
+	NameData	opcname;		/* name of opclass defined by this row */
+	Oid			opcdeftype;		/* type that opclass is default for, or 0 */
 } FormData_pg_opclass;
 
 /* ----------------
@@ -69,10 +69,6 @@ DATA(insert OID =  422 (	box_ops			603   ));
 DESCR("");
 DATA(insert OID =  423 (	float8_ops		701   ));
 DESCR("");
-DATA(insert OID =  424 (	int24_ops		  0   ));
-DESCR("");
-DATA(insert OID =  425 (	int42_ops		  0   ));
-DESCR("");
 DATA(insert OID =  426 (	int4_ops		 23   ));
 DESCR("");
 #define INT4_OPS_OID 426
@@ -86,7 +82,7 @@ DATA(insert OID =  431 (	text_ops		 25   ));
 DESCR("");
 DATA(insert OID =  432 (	abstime_ops		702   ));
 DESCR("");
-DATA(insert OID =  433 (	bigbox_ops		603   ));
+DATA(insert OID =  433 (	bigbox_ops		  0   ));
 DESCR("");
 DATA(insert OID =  434 (	poly_ops		604   ));
 DESCR("");

src/include/catalog/pg_proc.h

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: pg_proc.h,v 1.140 2000/06/17 23:41:51 tgl Exp $
+ * $Id: pg_proc.h,v 1.141 2000/06/19 03:54:45 tgl Exp $
  *
  * NOTES
  *	  The script catalog/genbki.sh reads this file and generates .bki
@@ -736,10 +736,6 @@ DATA(insert OID = 351 (  btint4cmp		   PGUID 12 f t t t 2 f 23 "23 23" 100 0 0 1
 DESCR("btree less-equal-greater");
 DATA(insert OID = 842 (  btint8cmp		   PGUID 12 f t t t 2 f 23 "20 20" 100 0 0 100  btint8cmp - ));
 DESCR("btree less-equal-greater");
-DATA(insert OID = 352 (  btint42cmp		   PGUID 12 f t t t 2 f 23 "23 21" 100 0 0 100  btint42cmp - ));
-DESCR("btree less-equal-greater");
-DATA(insert OID = 353 (  btint24cmp		   PGUID 12 f t t t 2 f 23 "21 23" 100 0 0 100  btint24cmp - ));
-DESCR("btree less-equal-greater");
 DATA(insert OID = 354 (  btfloat4cmp	   PGUID 12 f t t t 2 f 23 "700 700" 100 0 0 100	btfloat4cmp - ));
 DESCR("btree less-equal-greater");
 DATA(insert OID = 355 (  btfloat8cmp	   PGUID 12 f t t t 2 f 23 "701 701" 100 0 0 100	btfloat8cmp - ));
@@ -832,8 +828,8 @@ DATA(insert OID = 454 (  hashchar		   PGUID 12 f t t t 1 f 23 "18" 100 0 0 100
 DESCR("hash");
 DATA(insert OID = 455 (  hashname		   PGUID 12 f t t t 1 f 23 "19" 100 0 0 100  hashname - ));
 DESCR("hash");
-DATA(insert OID = 456 (  hashtext		   PGUID 12 f t t t 1 f 23 "25" 100 0 0 100  hashtext - ));
-DESCR("hash");
+DATA(insert OID = 456 (  hashvarlena	   PGUID 12 f t t t 1 f 23 "0" 100 0 0 100  hashvarlena - ));
+DESCR("hash any varlena type");
 DATA(insert OID = 457 (  hashoidvector	   PGUID 12 f t t t 1 f 23 "30" 100 0 0 100  hashoidvector - ));
 DESCR("hash");
 DATA(insert OID = 458 (  text_larger	   PGUID 11 f t t t 2 f 25 "25 25" 100 0 0 100  text_larger - ));
@@ -1332,9 +1328,7 @@ DATA(insert OID = 1078 (  bpcharcmp		   PGUID 11 f t t t 2 f 23 "1042 1042" 100
 DESCR("less-equal-greater");
 DATA(insert OID = 1079 (  varcharcmp	   PGUID 11 f t t t 2 f 23 "1043 1043" 100 0 0 100  varcharcmp - ));
 DESCR("less-equal-greater");
-DATA(insert OID = 1080 (  hashbpchar	   PGUID 11 f t t t 1 f 23 "1042" 100 0 0 100  hashbpchar - ));
-DESCR("hash");
-DATA(insert OID = 1081 (  hashvarchar	   PGUID 11 f t t t 1 f 23 "1043" 100 0 0 100  hashvarchar - ));
+DATA(insert OID = 1080 (  hashbpchar	   PGUID 12 f t t t 1 f 23 "1042" 100 0 0 100  hashbpchar - ));
 DESCR("hash");
 
 DATA(insert OID = 1084 (  date_in		   PGUID 12 f t f t 1 f 1082 "0" 100 0 0 100	date_in - ));
@@ -2312,6 +2306,12 @@ DESCR("greater-than-or-equal");
 DATA(insert OID = 1693 (  btboolcmp			   PGUID 12 f t t t 2 f 23 "16 16" 100 0 0 100  btboolcmp - ));
 DESCR("btree less-equal-greater");
 
+DATA(insert OID = 1696 (  timetz_hash		   PGUID 12 f t t t 1 f 23 "1266" 100 0 0 100  timetz_hash - ));
+DESCR("hash");
+DATA(insert OID = 1697 (  interval_hash		   PGUID 12 f t t t 1 f 23 "1186" 100 0 0 100  interval_hash - ));
+DESCR("hash");
+
+
 /* OID's 1700 - 1799 NUMERIC data type */
 DATA(insert OID = 1701 ( numeric_in				PGUID 12 f t t t 3 f 1700 "0 26 23" 100 0 0 100  numeric_in - ));
 DESCR("(internal)");

src/include/utils/builtins.h

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: builtins.h,v 1.117 2000/06/15 03:33:10 momjian Exp $
+ * $Id: builtins.h,v 1.118 2000/06/19 03:54:48 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -160,14 +160,12 @@ extern void ltoa(int32 l, char *a);
 
 /*
  *		Per-opclass comparison functions for new btrees.  These are
- *		stored in pg_amproc and defined in nbtree/
+ *		stored in pg_amproc and defined in access/nbtree/nbtcompare.c
  */
 extern Datum btboolcmp(PG_FUNCTION_ARGS);
 extern Datum btint2cmp(PG_FUNCTION_ARGS);
 extern Datum btint4cmp(PG_FUNCTION_ARGS);
 extern Datum btint8cmp(PG_FUNCTION_ARGS);
-extern Datum btint24cmp(PG_FUNCTION_ARGS);
-extern Datum btint42cmp(PG_FUNCTION_ARGS);
 extern Datum btfloat4cmp(PG_FUNCTION_ARGS);
 extern Datum btfloat8cmp(PG_FUNCTION_ARGS);
 extern Datum btoidcmp(PG_FUNCTION_ARGS);
@@ -421,7 +419,7 @@ extern bool bpcharge(char *arg1, char *arg2);
 extern int32 bpcharcmp(char *arg1, char *arg2);
 extern int32 bpcharlen(char *arg);
 extern int32 bpcharoctetlen(char *arg);
-extern uint32 hashbpchar(struct varlena * key);
+extern Datum hashbpchar(PG_FUNCTION_ARGS);
 
 extern Datum varcharin(PG_FUNCTION_ARGS);
 extern Datum varcharout(PG_FUNCTION_ARGS);
@@ -436,7 +434,6 @@ extern bool varcharge(char *arg1, char *arg2);
 extern int32 varcharcmp(char *arg1, char *arg2);
 extern int32 varcharlen(char *arg);
 extern int32 varcharoctetlen(char *arg);
-extern uint32 hashvarchar(struct varlena * key);
 
 /* varlena.c */
 extern text *textin(char *inputText);

src/include/utils/date.h

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: date.h,v 1.4 2000/06/09 01:11:15 tgl Exp $
+ * $Id: date.h,v 1.5 2000/06/19 03:54:48 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -95,6 +95,7 @@ extern Datum timetz_le(PG_FUNCTION_ARGS);
 extern Datum timetz_gt(PG_FUNCTION_ARGS);
 extern Datum timetz_ge(PG_FUNCTION_ARGS);
 extern Datum timetz_cmp(PG_FUNCTION_ARGS);
+extern Datum timetz_hash(PG_FUNCTION_ARGS);
 extern Datum overlaps_timetz(PG_FUNCTION_ARGS);
 extern Datum timetz_larger(PG_FUNCTION_ARGS);
 extern Datum timetz_smaller(PG_FUNCTION_ARGS);

src/include/utils/timestamp.h

@@ -6,7 +6,7 @@
  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: timestamp.h,v 1.7 2000/06/15 00:52:26 momjian Exp $
+ * $Id: timestamp.h,v 1.8 2000/06/19 03:54:48 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -151,6 +151,7 @@ extern Datum interval_ge(PG_FUNCTION_ARGS);
 extern Datum interval_gt(PG_FUNCTION_ARGS);
 extern Datum interval_finite(PG_FUNCTION_ARGS);
 extern Datum interval_cmp(PG_FUNCTION_ARGS);
+extern Datum interval_hash(PG_FUNCTION_ARGS);
 extern Datum interval_smaller(PG_FUNCTION_ARGS);
 extern Datum interval_larger(PG_FUNCTION_ARGS);
 

src/test/regress/expected/opr_sanity.out

@@ -36,6 +36,8 @@ WHERE (p1.prolang = 0 OR p1.prorettype = 0 OR
 (0 rows)
 
 -- Look for conflicting proc definitions (same names and input datatypes).
+-- (This test should be dead code now that we have the unique index
+-- pg_proc_proname_narg_type_index, but I'll leave it in anyway.)
 SELECT p1.oid, p1.proname, p2.oid, p2.proname
 FROM pg_proc AS p1, pg_proc AS p2
 WHERE p1.oid != p2.oid AND
@@ -57,7 +59,8 @@ WHERE p1.oid != p2.oid AND
     p1.prosrc = p2.prosrc AND
     (p1.prolang = 11 OR p1.prolang = 12) AND
     (p2.prolang = 11 OR p2.prolang = 12) AND
-    (p1.proisinh != p2.proisinh OR
+    (p1.prolang != p2.prolang OR
+     p1.proisinh != p2.proisinh OR
      p1.proistrusted != p2.proistrusted OR
      p1.proiscachable != p2.proiscachable OR
      p1.pronargs != p2.pronargs OR
@@ -522,6 +525,19 @@ WHERE p1.aggfinalfn = p2.oid AND
 -----+---------+-----+---------
 (0 rows)
 
+-- **************** pg_opclass ****************
+-- There should not be multiple entries in pg_opclass with the same
+-- nonzero opcdeftype value, because there can be only one default opclass
+-- for a datatype.  (But multiple entries with zero opcdeftype are OK.)
+SELECT p1.oid, p2.oid
+FROM pg_opclass AS p1, pg_opclass AS p2
+WHERE p1.oid != p2.oid AND
+    p1.opcdeftype = p2.opcdeftype AND
+    p1.opcdeftype != 0;
+ oid | oid 
+-----+-----
+(0 rows)
+
 -- **************** pg_amop ****************
 -- Look for illegal values in pg_amop fields
 SELECT p1.oid
@@ -564,11 +580,14 @@ WHERE p1.amstrategies != (SELECT count(*) FROM pg_amop AS p3
 -----+--------+-----+---------
 (0 rows)
 
--- Check that amopopr points at a reasonable-looking operator
+-- Check that amopopr points at a reasonable-looking operator, ie a binary
+-- operator yielding boolean.
+-- NOTE: for 7.1, add restriction that operator inputs are of same type.
+-- We used to have opclasses like "int24_ops" but these were broken.
 SELECT p1.oid, p2.oid, p2.oprname
 FROM pg_amop AS p1, pg_operator AS p2
 WHERE p1.amopopr = p2.oid AND
-    (p2.oprkind != 'b' OR p2.oprresult != 16);
+    (p2.oprkind != 'b' OR p2.oprresult != 16 OR p2.oprleft != p2.oprright);
  oid | oid | oprname 
 -----+-----+---------
 (0 rows)
@@ -640,3 +659,37 @@ WHERE p1.oid != p3.oid AND
 -----+-----+---------+-----+-----+---------
 (0 rows)
 
+-- Cross-check that each opclass that has any entries for a given AM
+-- has all the entries that any other opclass does.  This catches cases
+-- where an opclass has pg_amop but not pg_amproc entries or vice versa.
+-- (The above tests for missing pg_amop or pg_amproc entries are redundant
+-- with this, but I'll leave them in place anyway.)
+-- All the strategy index numbers used for each AM
+CREATE TEMP TABLE amopstrategies AS
+    SELECT DISTINCT amopid, amopstrategy FROM pg_amop;
+-- All the support proc numbers used for each AM
+CREATE TEMP TABLE amprocnums AS
+    SELECT DISTINCT amid, amprocnum FROM pg_amproc;
+-- All the opclasses that claim to have support for each AM in either table.
+-- UNION implies DISTINCT, so we do not need DISTINCT in the sub-selects.
+CREATE TEMP TABLE amopclassids AS
+    SELECT amid, amopclaid FROM pg_amproc UNION
+    SELECT amopid, amopclaid FROM pg_amop;
+-- Look for AMs that are missing one or more strategy operators
+SELECT * FROM amopclassids c, amopstrategies s
+WHERE c.amid = s.amopid AND NOT EXISTS
+      (SELECT 1 FROM pg_amop a WHERE a.amopid = c.amid AND
+       a.amopclaid = c.amopclaid AND a.amopstrategy = s.amopstrategy);
+ amid | amopclaid | amopid | amopstrategy 
+------+-----------+--------+--------------
+(0 rows)
+
+-- Look for AMs that are missing one or more support procs
+SELECT * FROM amopclassids c, amprocnums p
+WHERE c.amid = p.amid AND NOT EXISTS
+      (SELECT 1 FROM pg_amproc a WHERE a.amid = c.amid AND
+       a.amopclaid = c.amopclaid AND a.amprocnum = p.amprocnum);
+ amid | amopclaid | amid | amprocnum 
+------+-----------+------+-----------
+(0 rows)
+

src/test/regress/sql/opr_sanity.sql

@@ -36,6 +36,8 @@ WHERE (p1.prolang = 0 OR p1.prorettype = 0 OR
 	AND p1.proname != 'update_pg_pwd';
 
 -- Look for conflicting proc definitions (same names and input datatypes).
+-- (This test should be dead code now that we have the unique index
+-- pg_proc_proname_narg_type_index, but I'll leave it in anyway.)
 
 SELECT p1.oid, p1.proname, p2.oid, p2.proname
 FROM pg_proc AS p1, pg_proc AS p2
@@ -56,7 +58,8 @@ WHERE p1.oid != p2.oid AND
     p1.prosrc = p2.prosrc AND
     (p1.prolang = 11 OR p1.prolang = 12) AND
     (p2.prolang = 11 OR p2.prolang = 12) AND
-    (p1.proisinh != p2.proisinh OR
+    (p1.prolang != p2.prolang OR
+     p1.proisinh != p2.proisinh OR
      p1.proistrusted != p2.proistrusted OR
      p1.proiscachable != p2.proiscachable OR
      p1.pronargs != p2.pronargs OR
@@ -432,6 +435,18 @@ WHERE p1.aggfinalfn = p2.oid AND
      p1.aggtranstype1 != p2.proargtypes[0] OR
      p1.aggtranstype2 != p2.proargtypes[1]);
 
+-- **************** pg_opclass ****************
+
+-- There should not be multiple entries in pg_opclass with the same
+-- nonzero opcdeftype value, because there can be only one default opclass
+-- for a datatype.  (But multiple entries with zero opcdeftype are OK.)
+
+SELECT p1.oid, p2.oid
+FROM pg_opclass AS p1, pg_opclass AS p2
+WHERE p1.oid != p2.oid AND
+    p1.opcdeftype = p2.opcdeftype AND
+    p1.opcdeftype != 0;
+
 -- **************** pg_amop ****************
 
 -- Look for illegal values in pg_amop fields
@@ -467,12 +482,15 @@ WHERE p1.amstrategies != (SELECT count(*) FROM pg_amop AS p3
               AND EXISTS (SELECT * FROM pg_amop AS p3
                           WHERE p3.amopid = p1.oid AND p3.amopclaid = p2.oid);
 
--- Check that amopopr points at a reasonable-looking operator
+-- Check that amopopr points at a reasonable-looking operator, ie a binary
+-- operator yielding boolean.
+-- NOTE: for 7.1, add restriction that operator inputs are of same type.
+-- We used to have opclasses like "int24_ops" but these were broken.
 
 SELECT p1.oid, p2.oid, p2.oprname
 FROM pg_amop AS p1, pg_operator AS p2
 WHERE p1.amopopr = p2.oid AND
-    (p2.oprkind != 'b' OR p2.oprresult != 16);
+    (p2.oprkind != 'b' OR p2.oprresult != 16 OR p2.oprleft != p2.oprright);
 
 -- If opclass is for a specific type, operator inputs should be of that type
 
@@ -529,3 +547,35 @@ WHERE p1.oid != p3.oid AND
     p1.amid = p3.amid AND p1.amprocnum = p3.amprocnum AND
     p1.amproc = p2.oid AND p3.amproc = p4.oid AND
     (p2.proretset OR p4.proretset OR p2.pronargs != p4.pronargs);
+
+-- Cross-check that each opclass that has any entries for a given AM
+-- has all the entries that any other opclass does.  This catches cases
+-- where an opclass has pg_amop but not pg_amproc entries or vice versa.
+-- (The above tests for missing pg_amop or pg_amproc entries are redundant
+-- with this, but I'll leave them in place anyway.)
+
+-- All the strategy index numbers used for each AM
+CREATE TEMP TABLE amopstrategies AS
+    SELECT DISTINCT amopid, amopstrategy FROM pg_amop;
+
+-- All the support proc numbers used for each AM
+CREATE TEMP TABLE amprocnums AS
+    SELECT DISTINCT amid, amprocnum FROM pg_amproc;
+
+-- All the opclasses that claim to have support for each AM in either table.
+-- UNION implies DISTINCT, so we do not need DISTINCT in the sub-selects.
+CREATE TEMP TABLE amopclassids AS
+    SELECT amid, amopclaid FROM pg_amproc UNION
+    SELECT amopid, amopclaid FROM pg_amop;
+
+-- Look for AMs that are missing one or more strategy operators
+SELECT * FROM amopclassids c, amopstrategies s
+WHERE c.amid = s.amopid AND NOT EXISTS
+      (SELECT 1 FROM pg_amop a WHERE a.amopid = c.amid AND
+       a.amopclaid = c.amopclaid AND a.amopstrategy = s.amopstrategy);
+
+-- Look for AMs that are missing one or more support procs
+SELECT * FROM amopclassids c, amprocnums p
+WHERE c.amid = p.amid AND NOT EXISTS
+      (SELECT 1 FROM pg_amproc a WHERE a.amid = c.amid AND
+       a.amopclaid = c.amopclaid AND a.amprocnum = p.amprocnum);