Merge spetrunia@bk-internal.mysql.com:/home/bk/mysql-4.1

into mysql.com:/home/psergey/mysql-4.1-nulls-stats-r2

include/my_base.h

@@ -319,6 +319,8 @@ enum ha_base_keytype {
 #define SEARCH_NULL_ARE_EQUAL 32768	/* NULL in keys are equal */
 #define SEARCH_NULL_ARE_NOT_EQUAL 65536	/* NULL in keys are not equal */
 
+#define SEARCH_RETURN_B_POS (65536*2)   /* see ha_key_cmp for description */
+
 	/* bits in opt_flag */
 #define QUICK_USED	1
 #define READ_CACHE_USED	2

include/my_handler.h

@@ -63,4 +63,6 @@ extern int ha_key_cmp(register HA_KEYSEG *keyseg, register uchar *a,
 		      register uchar *b, uint key_length, uint nextflag,
 		      uint *diff_pos);
 
+extern HA_KEYSEG *ha_find_null(HA_KEYSEG *keyseg, uchar *a);
+
 #endif /* _my_handler_h */

include/myisam.h

@@ -322,7 +322,9 @@ typedef enum
   /* Treat NULLs as inequal when collecting statistics (default for 4.1/5.0) */
   MI_STATS_METHOD_NULLS_NOT_EQUAL,
   /* Treat NULLs as equal when collecting statistics (like 4.0 did) */
-  MI_STATS_METHOD_NULLS_EQUAL
+  MI_STATS_METHOD_NULLS_EQUAL,
+  /* Ignore NULLs - count only tuples without NULLs in the index components */
+  MI_STATS_METHOD_IGNORE_NULLS
 } enum_mi_stats_method;
 
 typedef struct st_mi_check_param
@@ -349,7 +351,14 @@ typedef struct st_mi_check_param
   int tmpfile_createflag;
   myf myf_rw;
   IO_CACHE read_cache;
+  
+  /* 
+    The next two are used to collect statistics, see update_key_parts for
+    description.
+  */
   ulonglong unique_count[MI_MAX_KEY_SEG+1];
+  ulonglong notnull_count[MI_MAX_KEY_SEG+1];
+  
   ha_checksum key_crc[MI_MAX_POSSIBLE_KEY];
   ulong rec_per_key_part[MI_MAX_KEY_SEG*MI_MAX_POSSIBLE_KEY];
   void *thd;
@@ -409,7 +418,8 @@ void update_auto_increment_key(MI_CHECK *param, MI_INFO *info,
 			       my_bool repair);
 int update_state_info(MI_CHECK *param, MI_INFO *info,uint update);
 void update_key_parts(MI_KEYDEF *keyinfo, ulong *rec_per_key_part,
-			     ulonglong *unique, ulonglong records);
+                      ulonglong *unique, ulonglong *notnull, 
+                      ulonglong records);
 int filecopy(MI_CHECK *param, File to,File from,my_off_t start,
 	     my_off_t length, const char *type);
 int movepoint(MI_INFO *info,byte *record,my_off_t oldpos,

myisam/myisamchk.c

@@ -339,7 +339,8 @@ static struct my_option my_long_options[] =
     REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
   {"stats_method", OPT_STATS_METHOD,
    "Specifies how index statistics collection code should threat NULLs. "
-   "Possible values of name are \"nulls_unequal\" (default behavior for 4.1/5.0), and \"nulls_equal\" (emulate 4.0 behavior).",
+   "Possible values of name are \"nulls_unequal\" (default behavior for 4.1/5.0), "
+   "\"nulls_equal\" (emulate 4.0 behavior), and \"nulls_ignored\".",
    (gptr*) &myisam_stats_method_str, (gptr*) &myisam_stats_method_str, 0,
     GET_STR, REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
   { 0, 0, 0, 0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0}
@@ -451,6 +452,10 @@ static void usage(void)
   -a, --analyze	      Analyze distribution of keys. Will make some joins in\n\
 		      MySQL faster.  You can check the calculated distribution\n\
 		      by using '--description --verbose table_name'.\n\
+  --stats_method=name Specifies how index statistics collection code should\n\
+                      threat NULLs. Possible values of name are \"nulls_unequal\"\n\
+                      (default for 4.1/5.0), \"nulls_equal\" (emulate 4.0), and \n\
+                      \"nulls_ignored\".\n\
   -d, --description   Prints some information about table.\n\
   -A, --set-auto-increment[=value]\n\
 		      Force auto_increment to start at this or higher value\n\
@@ -472,7 +477,7 @@ static void usage(void)
 #include <help_end.h>
 
 const char *myisam_stats_method_names[] = {"nulls_unequal", "nulls_equal",
-                                           NullS};
+                                           "nulls_ignored", NullS};
 TYPELIB myisam_stats_method_typelib= {
   array_elements(myisam_stats_method_names) - 1, "",
   myisam_stats_method_names, NULL};
@@ -699,14 +704,25 @@ get_one_option(int optid,
   case OPT_STATS_METHOD:
   {
     int method;
+    enum_mi_stats_method method_conv;
     myisam_stats_method_str= argument;
     if ((method=find_type(argument, &myisam_stats_method_typelib, 2)) <= 0)
     {
       fprintf(stderr, "Invalid value of stats_method: %s.\n", argument);
       exit(1);
     }
-    check_param.stats_method= test(method-1)? MI_STATS_METHOD_NULLS_EQUAL :
-                                              MI_STATS_METHOD_NULLS_NOT_EQUAL;
+    switch (method-1) {
+    case 0: 
+      method_conv= MI_STATS_METHOD_NULLS_EQUAL;
+      break;
+    case 1:
+      method_conv= MI_STATS_METHOD_NULLS_NOT_EQUAL;
+      break;
+    case 2:
+      method_conv= MI_STATS_METHOD_IGNORE_NULLS;
+      break;
+    }
+    check_param.stats_method= method_conv;
     break;
   }
 #ifdef DEBUG					/* Only useful if debugging */

myisam/myisamdef.h

@@ -297,7 +297,14 @@ typedef struct st_mi_sort_param
   pthread_t  thr;
   IO_CACHE read_cache, tempfile, tempfile_for_exceptions;
   DYNAMIC_ARRAY buffpek;
+  
+  /* 
+    The next two are used to collect statistics, see update_key_parts for
+    description.
+  */
   ulonglong unique[MI_MAX_KEY_SEG+1];
+  ulonglong notnull[MI_MAX_KEY_SEG+1];
+
   my_off_t pos,max_pos,filepos,start_recpos;
   uint key, key_length,real_key_length,sortbuff_size;
   uint maxbuffers, keys, find_length, sort_keys_length;

myisam/sort.c

@@ -481,8 +481,12 @@ int thr_write_keys(MI_SORT_PARAM *sort_param)
     {
       share->state.key_map|=(ulonglong) 1 << sinfo->key;
       if (param->testflag & T_STATISTICS)
-        update_key_parts(sinfo->keyinfo, rec_per_key_part,
-                         sinfo->unique, (ulonglong) info->state->records);
+        update_key_parts(sinfo->keyinfo, rec_per_key_part, sinfo->unique,
+                         param->stats_method == MI_STATS_METHOD_IGNORE_NULLS?
+                         sinfo->notnull: NULL,
+                         (ulonglong) info->state->records);
+
+
       if (!sinfo->buffpek.elements)
       {
         if (param->testflag & T_VERBOSE)

mysql-test/r/myisam.result

@@ -670,3 +670,35 @@ show index from t1;
 Table	Non_unique	Key_name	Seq_in_index	Column_name	Collation	Cardinality	Sub_part	Packed	Null	Index_type	Comment
 t1	1	a	1	a	A	10	NULL	NULL	YES	BTREE	
 drop table t1;
+set myisam_stats_method=nulls_ignored;
+show variables like 'myisam_stats_method';
+Variable_name	Value
+myisam_stats_method	nulls_ignored
+create table t1 (
+a char(3), b char(4), c char(5), d char(6),
+key(a,b,c,d)
+);
+insert into t1 values ('bcd','def1', NULL, 'zz');
+insert into t1 values ('bcd','def2', NULL, 'zz');
+insert into t1 values ('bce','def1', 'yuu', NULL);
+insert into t1 values ('bce','def2', NULL, 'quux');
+analyze table t1;
+Table	Op	Msg_type	Msg_text
+test.t1	analyze	status	OK
+show index from t1;
+Table	Non_unique	Key_name	Seq_in_index	Column_name	Collation	Cardinality	Sub_part	Packed	Null	Index_type	Comment
+t1	1	a	1	a	A	2	NULL	NULL	YES	BTREE	
+t1	1	a	2	b	A	4	NULL	NULL	YES	BTREE	
+t1	1	a	3	c	A	4	NULL	NULL	YES	BTREE	
+t1	1	a	4	d	A	4	NULL	NULL	YES	BTREE	
+delete from t1;
+analyze table t1;
+Table	Op	Msg_type	Msg_text
+test.t1	analyze	status	OK
+show index from t1;
+Table	Non_unique	Key_name	Seq_in_index	Column_name	Collation	Cardinality	Sub_part	Packed	Null	Index_type	Comment
+t1	1	a	1	a	A	0	NULL	NULL	YES	BTREE	
+t1	1	a	2	b	A	0	NULL	NULL	YES	BTREE	
+t1	1	a	3	c	A	0	NULL	NULL	YES	BTREE	
+t1	1	a	4	d	A	0	NULL	NULL	YES	BTREE	
+set myisam_stats_method=DEFAULT;

mysql-test/t/myisam.test

@@ -637,4 +637,23 @@ show index from t1;
 
 drop table t1;
 
+# WL#2609, CSC#XXXX: MyISAM 
+set myisam_stats_method=nulls_ignored;
+show variables like 'myisam_stats_method';
+
+create table t1 (
+  a char(3), b char(4), c char(5), d char(6),
+  key(a,b,c,d)
+);
+insert into t1 values ('bcd','def1', NULL, 'zz');
+insert into t1 values ('bcd','def2', NULL, 'zz');
+insert into t1 values ('bce','def1', 'yuu', NULL);
+insert into t1 values ('bce','def2', NULL, 'quux');
+analyze table t1;
+show index from t1;
+delete from t1;
+analyze table t1;
+show index from t1;
+
+set myisam_stats_method=DEFAULT;
 # End of 4.1 tests

mysys/my_handler.c

@@ -75,7 +75,7 @@ static int compare_bin(uchar *a, uint a_length, uchar *b, uint b_length,
 
   SYNOPSIS
     ha_key_cmp()
-    keyseg	Key segments of key to compare
+    keyseg	Array of key segments of key to compare
     a		First key to compare, in format from _mi_pack_key()
 		This is normally key specified by user
     b		Second key to compare.  This is always from a row
@@ -84,6 +84,28 @@ static int compare_bin(uchar *a, uint a_length, uchar *b, uint b_length,
     next_flag	How keys should be compared
 		If bit SEARCH_FIND is not set the keys includes the row
 		position and this should also be compared
+    diff_pos    OUT Number of first keypart where values differ, counting 
+                from one.
+  
+  DESCRIPTION
+    If SEARCH_RETURN_B_POS flag is set, diff_pos must point to array of 2
+    values, first value has the meaning as described in parameter
+    description above, the second value is:
+
+    diff_pos[1]  OUT  (b + diff_pos[1]) points to first value in tuple b
+                      that is different from corresponding value in tuple a.
+
+  EXAMPLES 
+   Example1: if the function is called for tuples
+     ('aaa','bbb') and ('eee','fff'), then
+     diff_pos[0] = 1 (as 'aaa' != 'eee')
+     diff_pos[1] = 0 (offset from beggining of tuple b to 'eee' keypart).
+
+   Example2: if the index function is called for tuples
+     ('aaa','bbb') and ('aaa','fff'),
+     diff_pos[0] = 2 (as 'aaa' != 'eee')
+     diff_pos[1] = 3 (offset from beggining of tuple b to 'fff' keypart,
+                      here we assume that first key part is CHAR(3) NOT NULL)
 
   NOTES
     Number-keys can't be splited
@@ -107,6 +129,7 @@ int ha_key_cmp(register HA_KEYSEG *keyseg, register uchar *a,
   float f_1,f_2;
   double d_1,d_2;
   uint next_key_length;
+  uchar *orig_b= b;
 
   *diff_pos=0;
   for ( ; (int) key_length >0 ; key_length=next_key_length, keyseg++)
@@ -115,6 +138,9 @@ int ha_key_cmp(register HA_KEYSEG *keyseg, register uchar *a,
     uint piks=! (keyseg->flag & HA_NO_SORT);
     (*diff_pos)++;
 
+    if (nextflag & SEARCH_RETURN_B_POS)
+      diff_pos[1]= (uint)(b - orig_b);
+
     /* Handle NULL part */
     if (keyseg->null_bit)
     {
@@ -448,3 +474,84 @@ int ha_key_cmp(register HA_KEYSEG *keyseg, register uchar *a,
   }
   return 0;
 } /* ha_key_cmp */
+
+
+/*
+  Find the first NULL value in index-suffix values tuple
+
+  SYNOPSIS
+    ha_find_null()
+      keyseg     Array of keyparts for key suffix
+      a          Key suffix value tuple
+
+  DESCRIPTION
+    Find the first NULL value in index-suffix values tuple.
+    TODO Consider optimizing this fuction or its use so we don't search for
+         NULL values in completely NOT NULL index suffixes.
+
+  RETURN
+    First key part that has NULL as value in values tuple, or the last key part 
+    (with keyseg->type==HA_TYPE_END) if values tuple doesn't contain NULLs.
+*/
+
+HA_KEYSEG *ha_find_null(HA_KEYSEG *keyseg, uchar *a)
+{
+  for (; (enum ha_base_keytype) keyseg->type != HA_KEYTYPE_END; keyseg++)
+  {
+    uchar *end;
+    if (keyseg->null_bit)
+    {
+      if (!*a++)
+        return keyseg;
+    }
+    end= a+ keyseg->length;
+
+    switch ((enum ha_base_keytype) keyseg->type) {
+    case HA_KEYTYPE_TEXT:
+    case HA_KEYTYPE_BINARY:
+      if (keyseg->flag & HA_SPACE_PACK)
+      {
+        int a_length;
+        get_key_length(a_length, a);
+        a += a_length;
+        break;
+      }
+      else
+        a= end;
+      break;
+    case HA_KEYTYPE_VARTEXT:
+    case HA_KEYTYPE_VARBINARY:
+      {
+        int a_length;
+        get_key_length(a_length, a);
+        a+= a_length;
+        break;
+      }
+    case HA_KEYTYPE_NUM:
+      if (keyseg->flag & HA_SPACE_PACK)
+      {
+        int alength= *a++;
+        end= a+alength;
+      }
+      a= end;
+      break;
+    case HA_KEYTYPE_INT8:
+    case HA_KEYTYPE_SHORT_INT:
+    case HA_KEYTYPE_USHORT_INT:
+    case HA_KEYTYPE_LONG_INT:
+    case HA_KEYTYPE_ULONG_INT:
+    case HA_KEYTYPE_INT24:
+    case HA_KEYTYPE_UINT24:
+#ifdef HAVE_LONG_LONG
+    case HA_KEYTYPE_LONGLONG:
+    case HA_KEYTYPE_ULONGLONG:
+#endif
+    case HA_KEYTYPE_FLOAT:
+    case HA_KEYTYPE_DOUBLE:
+      a= end;
+      break;
+    }
+  }
+  return keyseg;
+}
+

sql/ha_myisam.cc

@@ -40,7 +40,7 @@ TYPELIB myisam_recover_typelib= {array_elements(myisam_recover_names)-1,"",
 				 myisam_recover_names, NULL};
 
 const char *myisam_stats_method_names[] = {"nulls_unequal", "nulls_equal",
-                                           NullS};
+                                           "nulls_ignored", NullS};
 TYPELIB myisam_stats_method_typelib= {
   array_elements(myisam_stats_method_names) - 1, "",
   myisam_stats_method_names, NULL};

sql/mysqld.cc

@@ -5281,7 +5281,8 @@ The minimum value for this variable is 4096.",
    GET_ULONG, REQUIRED_ARG, 8192*1024, 4, ~0L, 0, 1, 0},
   {"myisam_stats_method", OPT_MYISAM_STATS_METHOD,
    "Specifies how MyISAM index statistics collection code should threat NULLs. "
-   "Possible values of name are \"nulls_unequal\" (default behavior for 4.1/5.0), and \"nulls_equal\" (emulate 4.0 behavior).",
+   "Possible values of name are \"nulls_unequal\" (default behavior for 4.1/5.0), "
+   "\"nulls_equal\" (emulate 4.0 behavior), and \"nulls_ignored\".",
    (gptr*) &myisam_stats_method_str, (gptr*) &myisam_stats_method_str, 0,
     GET_STR, REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
   {"net_buffer_length", OPT_NET_BUFFER_LENGTH,
@@ -6475,16 +6476,26 @@ get_one_option(int optid, const struct my_option *opt __attribute__((unused)),
   }
   case OPT_MYISAM_STATS_METHOD:
   {
-    myisam_stats_method_str= argument;
     int method;
+    ulong method_conv;
+    myisam_stats_method_str= argument;
     if ((method=find_type(argument, &myisam_stats_method_typelib, 2)) <= 0)
     {
       fprintf(stderr, "Invalid value of myisam_stats_method: %s.\n", argument);
       exit(1);
     }
-    global_system_variables.myisam_stats_method= 
-      test(method-1)? MI_STATS_METHOD_NULLS_EQUAL : 
-                      MI_STATS_METHOD_NULLS_NOT_EQUAL;
+    switch (method-1) {
+    case 0: 
+      method_conv= MI_STATS_METHOD_NULLS_EQUAL;
+      break;
+    case 1:
+      method_conv= MI_STATS_METHOD_NULLS_NOT_EQUAL;
+      break;
+    case 2:
+      method_conv= MI_STATS_METHOD_IGNORE_NULLS;
+      break;
+    }
+    global_system_variables.myisam_stats_method= method_conv;
     break;
   }
   case OPT_SQL_MODE: