Changes

* new split algorithm (as proposed in http://archives.postgresql.org/pgsql-hackers/2006-06/msg00254.php)
  * possible call pickSplit() for second and below columns
  * add spl_(l|r)datum_exists to GIST_SPLITVEC -
    pickSplit should check its values to use already defined
    spl_(l|r)datum for splitting. pickSplit should set
    spl_(l|r)datum_exists to 'false' (if they was 'true') to
    signal to caller about using spl_(l|r)datum.
  * support for old pickSplit(): not very optimal
    but correct split
* remove 'bytes' field from GISTENTRY: in any case size of
  value is defined by it's type.
* split GIST_SPLITVEC to two structures: one for using in picksplit
  and second - for internal use.
* some code refactoring
* support of subsplit to rtree opclasses

TODO: add support of subsplit to contrib modules

src/include/access/gist.h

@@ -9,7 +9,7 @@
  * Portions Copyright (c) 1996-2006, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/gist.h,v 1.53 2006/06/25 01:02:12 momjian Exp $
+ * $PostgreSQL: pgsql/src/include/access/gist.h,v 1.54 2006/06/28 12:00:14 teodor Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -75,37 +75,29 @@ typedef GISTPageOpaqueData *GISTPageOpaque;
 
 /*
  * This is the Split Vector to be returned by the PickSplit method.
+ * PickSplit should check spl_(r|l)datum_exists. If it is 'true',
+ * that corresponding spl_(r|l)datum already defined and
+ * PickSplit should use that value. PickSplit should always set
+ * spl_(r|l)datum_exists to false: GiST will check value to 
+ * control supportng this feature by PickSplit...
  */
 typedef struct GIST_SPLITVEC
 {
 	OffsetNumber *spl_left;		/* array of entries that go left */
 	int			spl_nleft;		/* size of this array */
 	Datum		spl_ldatum;		/* Union of keys in spl_left */
-	Datum		spl_lattr[INDEX_MAX_KEYS];		/* Union of subkeys in
-												 * spl_left */
-	int			spl_lattrsize[INDEX_MAX_KEYS];
-	bool		spl_lisnull[INDEX_MAX_KEYS];
-	bool		spl_leftvalid;
+	bool		spl_ldatum_exists;  /* true, if spl_ldatum already exists. */
 
 	OffsetNumber *spl_right;	/* array of entries that go right */
 	int			spl_nright;		/* size of the array */
 	Datum		spl_rdatum;		/* Union of keys in spl_right */
-	Datum		spl_rattr[INDEX_MAX_KEYS];		/* Union of subkeys in
-												 * spl_right */
-	int			spl_rattrsize[INDEX_MAX_KEYS];
-	bool		spl_risnull[INDEX_MAX_KEYS];
-	bool		spl_rightvalid;
-
-	int		   *spl_idgrp;
-	int		   *spl_ngrp;		/* number in each group */
-	char	   *spl_grpflag;	/* flags of each group */
+	bool		spl_rdatum_exists;  /* true, if spl_rdatum already exists. */
 } GIST_SPLITVEC;
 
 /*
  * An entry on a GiST node.  Contains the key, as well as its own
  * location (rel,page,offset) which can supply the matching pointer.
- * The size of the key is in bytes, and leafkey is a flag to tell us
- * if the entry is in a leaf node.
+ * leafkey is a flag to tell us if the entry is in a leaf node.
  */
 typedef struct GISTENTRY
 {
@@ -113,7 +105,6 @@ typedef struct GISTENTRY
 	Relation	rel;
 	Page		page;
 	OffsetNumber offset;
-	int			bytes;
 	bool		leafkey;
 } GISTENTRY;
 
@@ -147,8 +138,8 @@ typedef struct
 /*
  * macro to initialize a GISTENTRY
  */
-#define gistentryinit(e, k, r, pg, o, b, l) \
+#define gistentryinit(e, k, r, pg, o, l) \
 	do { (e).key = (k); (e).rel = (r); (e).page = (pg); \
-		 (e).offset = (o); (e).bytes = (b); (e).leafkey = (l); } while (0)
+		 (e).offset = (o); (e).leafkey = (l); } while (0)
 
 #endif   /* GIST_H */

src/include/access/gist_private.h

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2006, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/gist_private.h,v 1.17 2006/05/29 12:50:06 teodor Exp $
+ * $PostgreSQL: pgsql/src/include/access/gist_private.h,v 1.18 2006/06/28 12:00:14 teodor Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -180,6 +180,21 @@ typedef struct GISTInsertStack
 	struct GISTInsertStack *next;
 } GISTInsertStack;
 
+typedef struct GistSplitVector {
+	GIST_SPLITVEC	splitVector;  /* to/from PickSplit method */
+
+	Datum       spl_lattr[INDEX_MAX_KEYS];      /* Union of subkeys in spl_left */
+	bool        spl_lisnull[INDEX_MAX_KEYS];
+	bool        spl_leftvalid;
+
+	Datum       spl_rattr[INDEX_MAX_KEYS];      /* Union of subkeys in spl_right */
+	bool        spl_risnull[INDEX_MAX_KEYS];
+	bool        spl_rightvalid;
+
+	bool        *spl_equiv; /* equivalent tuples which can be freely 
+							 * distributed between left and right pages */ 
+} GistSplitVector;
+
 #define XLogRecPtrIsInvalid( r )	( (r).xlogid == 0 && (r).xrecoff == 0 )
 
 typedef struct
@@ -206,12 +221,6 @@ typedef struct
 /* root page of a gist index */
 #define GIST_ROOT_BLKNO				0
 
-#define ATTSIZE(datum, tupdesc, i, isnull) \
-		( \
-				(isnull) ? 0 : \
-				   att_addlength(0, (tupdesc)->attrs[(i)-1]->attlen, (datum)) \
-		)
-
 /*
  * mark tuples on inner pages during recovery
  */
@@ -281,7 +290,7 @@ extern IndexTuple gistgetadjusted(Relation r,
 				IndexTuple addtup,
 				GISTSTATE *giststate);
 extern IndexTuple gistFormTuple(GISTSTATE *giststate,
-			  Relation r, Datum *attdata, int *datumsize, bool *isnull);
+			  Relation r, Datum *attdata, bool *isnull, bool newValues);
 
 extern OffsetNumber gistchoose(Relation r, Page p,
 		   IndexTuple it,
@@ -289,34 +298,34 @@ extern OffsetNumber gistchoose(Relation r, Page p,
 extern void gistcentryinit(GISTSTATE *giststate, int nkey,
 			   GISTENTRY *e, Datum k,
 			   Relation r, Page pg,
-			   OffsetNumber o, int b, bool l, bool isNull);
-extern void gistDeCompressAtt(GISTSTATE *giststate, Relation r,
-				  IndexTuple tuple, Page p, OffsetNumber o,
-				  GISTENTRY *attdata, bool *isnull);
-
-typedef struct {
-	int		*attrsize;
-	Datum	*attr;
-	int		len;
-	OffsetNumber *entries;
-	bool	*isnull;
-	int		*idgrp;
-} GistSplitVec;
-
-extern void gistunionsubkeyvec(GISTSTATE *giststate, 
-	IndexTuple *itvec, GistSplitVec *gsvp,  int startkey);
-extern void gistunionsubkey(GISTSTATE *giststate, IndexTuple *itvec, 
-		GIST_SPLITVEC *spl, int attno);
+			   OffsetNumber o, bool l, bool isNull);
 
 extern void GISTInitBuffer(Buffer b, uint32 f);
 extern void gistdentryinit(GISTSTATE *giststate, int nkey, GISTENTRY *e,
 			   Datum k, Relation r, Page pg, OffsetNumber o,
-			   int b, bool l, bool isNull);
-bool gistUserPicksplit(Relation r, GistEntryVector *entryvec, int attno, GIST_SPLITVEC *v,
-				  IndexTuple *itup, int len, GISTSTATE *giststate);
+			   bool l, bool isNull);
+
+extern float gistpenalty(GISTSTATE *giststate, int attno,
+            			GISTENTRY *key1, bool isNull1,
+						GISTENTRY *key2, bool isNull2);
+extern bool gistMakeUnionItVec(GISTSTATE *giststate, IndexTuple *itvec, int len, int startkey,
+                    Datum *attr, bool *isnull );
+extern bool gistKeyIsEQ(GISTSTATE *giststate, int attno, Datum a, Datum b);
+extern void gistDeCompressAtt(GISTSTATE *giststate, Relation r, IndexTuple tuple, Page p,
+				OffsetNumber o, GISTENTRY *attdata, bool *isnull);
+
+extern void gistMakeUnionKey( GISTSTATE *giststate, int attno,
+                    GISTENTRY   *entry1, bool isnull1,
+					GISTENTRY   *entry2, bool isnull2,
+					Datum   *dst, bool *dstisnull );
 
 /* gistvacuum.c */
 extern Datum gistbulkdelete(PG_FUNCTION_ARGS);
 extern Datum gistvacuumcleanup(PG_FUNCTION_ARGS);
 
+/* gistsplit.c */
+extern void gistSplitByKey(Relation r, Page page, IndexTuple *itup, 
+							int len, GISTSTATE *giststate,
+							GistSplitVector *v, GistEntryVector *entryvec, 
+							int attno);
 #endif   /* GIST_PRIVATE_H */