merged

2 files changed, 43 insertions, 14 deletions

diff --git a/myisam/mi_check.c b/myisam/mi_check.c
index ee64f9b9979..1f453278eea 100644
--- a/myisam/mi_check.c
+++ b/myisam/mi_check.c
@@ -80,6 +80,7 @@ void myisamchk_init(MI_CHECK *param)
   param->start_check_pos=0;
   param->max_record_length= LONGLONG_MAX;
   param->key_cache_block_size= KEY_CACHE_BLOCK_SIZE;
+  param->stats_method= MI_STATS_METHOD_NULLS_NOT_EQUAL;
 }
 
 	/* Check the status flags for the table */
@@ -559,10 +560,11 @@ static int chk_index(MI_CHECK *param, MI_INFO *info, MI_KEYDEF *keyinfo,
 		     ha_checksum *key_checksum, uint level)
 {
   int flag;
-  uint used_length,comp_flag,nod_flag,key_length=0,not_used;
+  uint used_length,comp_flag,nod_flag,key_length=0;
   uchar key[MI_MAX_POSSIBLE_KEY_BUFF],*temp_buff,*keypos,*old_keypos,*endpos;
   my_off_t next_page,record;
   char llbuff[22];
+  uint diff_pos;
   DBUG_ENTER("chk_index");
   DBUG_DUMP("buff",(byte*) buff,mi_getint(buff));
 
@@ -620,7 +622,7 @@ static int chk_index(MI_CHECK *param, MI_INFO *info, MI_KEYDEF *keyinfo,
     }
     if ((*keys)++ &&
 	(flag=ha_key_cmp(keyinfo->seg,info->lastkey,key,key_length,
-			 comp_flag, &not_used)) >=0)
+			 comp_flag, &diff_pos)) >=0)
     {
       DBUG_DUMP("old",(byte*) info->lastkey, info->lastkey_length);
       DBUG_DUMP("new",(byte*) key, key_length);
@@ -636,11 +638,11 @@ static int chk_index(MI_CHECK *param, MI_INFO *info, MI_KEYDEF *keyinfo,
     {
       if (*keys != 1L)				/* not first_key */
       {
-	uint diff;
-	ha_key_cmp(keyinfo->seg,info->lastkey,key,USE_WHOLE_KEY,
-		   SEARCH_FIND | SEARCH_NULL_ARE_NOT_EQUAL,
-		   &diff);
-	param->unique_count[diff-1]++;
+        if (param->stats_method == MI_STATS_METHOD_NULLS_NOT_EQUAL)
+          ha_key_cmp(keyinfo->seg,info->lastkey,key,USE_WHOLE_KEY,
+                     SEARCH_FIND | SEARCH_NULL_ARE_NOT_EQUAL,
+                     &diff_pos);
+	param->unique_count[diff_pos-1]++;
       }
     }
     (*key_checksum)+= mi_byte_checksum((byte*) key,
@@ -2014,7 +2016,7 @@ int mi_repair_by_sort(MI_CHECK *param, register MI_INFO *info,
   sort_param.sort_info=&sort_info;
   sort_param.fix_datafile= (my_bool) (! rep_quick);
   sort_param.master =1;
-
+  
   del=info->state->del;
   param->glob_crc=0;
   if (param->testflag & T_CALC_CHECKSUM)
@@ -3250,9 +3252,10 @@ static int sort_key_write(MI_SORT_PARAM *sort_param, const void *a)
     cmp=ha_key_cmp(sort_param->seg,sort_info->key_block->lastkey,
 		   (uchar*) a, USE_WHOLE_KEY,SEARCH_FIND | SEARCH_UPDATE,
 		   &diff_pos);
-    ha_key_cmp(sort_param->seg,sort_info->key_block->lastkey,
-               (uchar*) a, USE_WHOLE_KEY,SEARCH_FIND | SEARCH_NULL_ARE_NOT_EQUAL,
-               &diff_pos);
+    if (param->stats_method == MI_STATS_METHOD_NULLS_NOT_EQUAL)
+      ha_key_cmp(sort_param->seg,sort_info->key_block->lastkey,
+                 (uchar*) a, USE_WHOLE_KEY, 
+                 SEARCH_FIND | SEARCH_NULL_ARE_NOT_EQUAL, &diff_pos);
     sort_param->unique[diff_pos-1]++;
   }
   else
@@ -3989,9 +3992,10 @@ void update_auto_increment_key(MI_CHECK *param, MI_INFO *info,
     unique[0]= (#different values of {keypart1}) - 1
     unique[1]= (#different values of {keypart2,keypart1} tuple) - unique[0] - 1
     ...
-    Here we assume that NULL != NULL (see SEARCH_NULL_ARE_NOT_EQUAL). The
-    'unique' array is collected in one sequential scan through the entire
+    The 'unique' array is collected in one sequential scan through the entire
     index. This is done in two places: in chk_index() and in sort_key_write().
+    Statistics collection may consider NULLs as either equal or inequal (see
+    SEARCH_NULL_ARE_NOT_EQUAL, MI_STATS_METHOD_*).
 
     Output is an array:
     rec_per_key_part[k] = 
diff --git a/myisam/myisamchk.c b/myisam/myisamchk.c
index 78ae756bf7c..cfc3930779f 100644
--- a/myisam/myisamchk.c
+++ b/myisam/myisamchk.c
@@ -67,6 +67,7 @@ static const char *field_pack[]=
  "no zeros", "blob", "constant", "table-lockup",
  "always zero","varchar","unique-hash","?","?"};
 
+static const char *myisam_stats_method_str="nulls_inequal";
 
 static void get_options(int *argc,char * * *argv);
 static void print_version(void);
@@ -155,7 +156,7 @@ enum options_mc {
   OPT_READ_BUFFER_SIZE, OPT_WRITE_BUFFER_SIZE, OPT_SORT_BUFFER_SIZE,
   OPT_SORT_KEY_BLOCKS, OPT_DECODE_BITS, OPT_FT_MIN_WORD_LEN,
   OPT_FT_MAX_WORD_LEN, OPT_FT_STOPWORD_FILE,
-  OPT_MAX_RECORD_LENGTH, OPT_AUTO_CLOSE
+  OPT_MAX_RECORD_LENGTH, OPT_AUTO_CLOSE, OPT_STATS_METHOD
 };
 
 static struct my_option my_long_options[] =
@@ -336,6 +337,11 @@ static struct my_option my_long_options[] =
     "Use stopwords from this file instead of built-in list.",
     (gptr*) &ft_stopword_file, (gptr*) &ft_stopword_file, 0, GET_STR,
     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_inequal\" (default behavior for 4.1/5.0), and \"nulls_equal\" (emulate 4.0 behavior).",
+   (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}
 };
 
@@ -465,6 +471,12 @@ static void usage(void)
 
 #include <help_end.h>
 
+const char *myisam_stats_method_names[] = {"nulls_inequal", "nulls_equal",
+                                           NullS};
+TYPELIB myisam_stats_method_typelib= {
+  array_elements(myisam_stats_method_names) - 1, "",
+  myisam_stats_method_names, NULL};
+
 	 /* Read options */
 
 static my_bool
@@ -684,6 +696,19 @@ get_one_option(int optid,
     else
       check_param.testflag|= T_CALC_CHECKSUM;
     break;
+  case OPT_STATS_METHOD:
+  {
+    myisam_stats_method_str= argument;
+    int method;
+    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;
+    break;
+  }
 #ifdef DEBUG					/* Only useful if debugging */
   case OPT_START_CHECK_POS:
     check_param.start_check_pos= strtoull(argument, NULL, 0);