Merge

4 files changed, 109 insertions, 31 deletions

diff --git a/sql/sql_class.h b/sql/sql_class.h
index 2844cab9b4e..0cca72d7d1e 100644
--- a/sql/sql_class.h
+++ b/sql/sql_class.h
@@ -502,6 +502,7 @@ typedef struct system_variables
   ulong optimizer_use_condition_selectivity;
   ulong use_stat_tables;
   ulong histogram_size;
+  ulong histogram_type;
   ulong preload_buff_size;
   ulong profiling_history_size;
   ulong read_buff_size;
diff --git a/sql/sql_statistics.cc b/sql/sql_statistics.cc
index 0f26ca7520a..74ed90cf7a1 100644
--- a/sql/sql_statistics.cc
+++ b/sql/sql_statistics.cc
@@ -928,6 +928,9 @@ public:
         case COLUMN_STAT_HIST_SIZE:
           stat_field->store(table_field->collected_stats->histogram.get_size());
           break;
+        case COLUMN_STAT_HIST_TYPE:
+          stat_field->store(table_field->collected_stats->histogram.get_type());
+          break;
         case COLUMN_STAT_HISTOGRAM:
           const char * col_histogram=
           (const char *) (table_field->collected_stats->histogram.get_values());
@@ -971,7 +974,7 @@ public:
       char buff[MAX_FIELD_WIDTH];
       String val(buff, sizeof(buff), &my_charset_utf8_bin);
 
-      for (uint i= COLUMN_STAT_MIN_VALUE; i <= COLUMN_STAT_HIST_SIZE; i++)
+      for (uint i= COLUMN_STAT_MIN_VALUE; i <= COLUMN_STAT_HIST_TYPE; i++)
       {  
         Field *stat_field= stat_table->field[i];
 
@@ -1007,6 +1010,10 @@ public:
           case COLUMN_STAT_HIST_SIZE:
             table_field->read_stats->histogram.set_size(stat_field->val_int());
             break;            
+          case COLUMN_STAT_HIST_TYPE:
+            Histogram_type hist_type= (Histogram_type) (stat_field->val_int());
+            table_field->read_stats->histogram.set_type(hist_type);
+            break;            
           }
         }
       }
@@ -1238,7 +1245,7 @@ class Histogram_builder
   Field *min_value;
   Field *max_value;
   Histogram *histogram;
-  uint hist_size;
+  uint hist_width;
   double bucket_capacity;  
   uint curr_bucket;
   ulonglong count;
@@ -1252,8 +1259,8 @@ public:
     min_value= col_stats->min_value;
     max_value= col_stats->max_value;
     histogram= &col_stats->histogram;
-    hist_size= histogram->get_size();
-    bucket_capacity= (double) records / (hist_size + 1);
+    hist_width= histogram->get_width();
+    bucket_capacity= (double) records / (hist_width + 1);
     curr_bucket= 0;
     count= 0;
     count_distinct= 0;    
@@ -1265,7 +1272,7 @@ public:
   {
     count_distinct++;
     count+= elem_cnt;
-    if (curr_bucket == hist_size)
+    if (curr_bucket == hist_width)
       return 0;
     if (count > bucket_capacity * (curr_bucket + 1))
     {
@@ -1273,7 +1280,7 @@ public:
       histogram->set_value(curr_bucket,
                            column->middle_point_pos(min_value, max_value)); 
       curr_bucket++;
-      while (curr_bucket != hist_size &&
+      while (curr_bucket != hist_width &&
              count > bucket_capacity * (curr_bucket + 1))
       {
         histogram->set_prev_value(curr_bucket);
@@ -1794,6 +1801,7 @@ int alloc_statistics_for_table(THD* thd, TABLE *table)
       columns++;
   }
   uint hist_size= thd->variables.histogram_size;
+  Histogram_type hist_type= (Histogram_type) (thd->variables.histogram_type);
   uchar *histogram= NULL;
   if (hist_size > 0)
     histogram= (uchar *) alloc_root(&table->mem_root, hist_size * columns);
@@ -1818,6 +1826,7 @@ int alloc_statistics_for_table(THD* thd, TABLE *table)
     if (bitmap_is_set(table->read_set, (*field_ptr)->field_index))
     {
       column_stats->histogram.set_size(hist_size);
+      column_stats->histogram.set_type(hist_type);
       column_stats->histogram.set_values(histogram);
       histogram+= hist_size;
     }
@@ -2200,6 +2209,7 @@ void Column_statistics_collected::finish(ha_rows rows)
     set_not_null(COLUMN_STAT_HIST_SIZE);
     if (hist_size && distincts)
     {
+      set_not_null(COLUMN_STAT_HIST_TYPE);
       histogram.set_values(count_distinct->get_histogram());
       set_not_null(COLUMN_STAT_HISTOGRAM);
     } 
diff --git a/sql/sql_statistics.h b/sql/sql_statistics.h
index 9a2b5c2433b..b699ec7100a 100644
--- a/sql/sql_statistics.h
+++ b/sql/sql_statistics.h
@@ -24,6 +24,13 @@ enum enum_use_stat_tables_mode
   PEFERABLY,
 } Use_stat_tables_mode;
 
+typedef
+enum enum_histogram_type
+{
+  SINGLE_PREC_HB,
+  DOUBLE_PREC_HB
+} Histogram_type;
+
 enum enum_stat_tables
 {
   TABLE_STAT,
@@ -59,6 +66,7 @@ enum enum_column_stat_col
   COLUMN_STAT_AVG_LENGTH,
   COLUMN_STAT_AVG_FREQUENCY,
   COLUMN_STAT_HIST_SIZE,
+  COLUMN_STAT_HIST_TYPE,
   COLUMN_STAT_HISTOGRAM
 };
 
@@ -99,46 +107,74 @@ double get_column_range_cardinality(Field *field,
                                     key_range *min_endp,
                                     key_range *max_endp);
 
-#define HIST_FACTOR 255
-#define INV_HIST_FACTOR ((double) 1.0 / HIST_FACTOR)
-
 class Histogram
 {
+
 private:
+  Histogram_type type;
+  uint8 size;
+  uchar *values;
+
+  uint prec_factor()
+  {
+    switch (type) {
+    case SINGLE_PREC_HB:
+      return ((uint) (1 << 8) - 1);
+    case DOUBLE_PREC_HB:
+      return ((uint) (1 << 16) - 1);
+    }
+  }
+
 public:
+  uint get_width()
+  {
+    switch (type) {
+    case SINGLE_PREC_HB:
+      return size;
+    case DOUBLE_PREC_HB:
+      return size / 2;
+    }
+  }
 
 private:
-  uint8 size;
-  uint8 *values;
+  uint get_value(uint i)
+  {
+    switch (type) {
+    case SINGLE_PREC_HB:
+      return (uint) (((uint8 *) values)[i]);
+    case DOUBLE_PREC_HB:
+      return (uint) (((uint16 *) values)[i]);
+    }
+  }
 
   uint find_bucket(double pos, bool first)
   {
-    uint8 val= (uint8) (pos * HIST_FACTOR);
+    uint val= (uint) (pos * prec_factor());
     int lp= 0;
-    int rp= size - 1;
-    int i= 0;
-    for (int d= size / 2 ; d;  d= (rp - lp) / 2)
+    int rp= get_width() - 1;
+    uint i= 0;
+    for (int d= get_width() / 2 ; d;  d= (rp - lp) / 2)
     {
       i= lp + d;
-      if (val == values[i])
+      if (val == get_value(i))
 	break; 
-      if (val < values[i])
+      if (val < get_value(i))
         rp= i;
-      else if (val > values[i + 1])
+      else if (val > get_value(i + 1))
         lp= i + 1;
       else
         break;
     }
-    if (val == values[i])
+    if (val == get_value(i))
     {
       if (first)
       {
-        while(i && val == values[i - 1])
+        while(i && val == get_value(i - 1))
           i--;
       }
       else
       {
-        while(i + 1 < size && val == values[i + 1])
+        while(i + 1 < get_width() && val == get_value(i + 1))
           i++;
       }
     }
@@ -149,24 +185,44 @@ public:
 
   uint get_size() { return (uint) size; }
 
+  Histogram_type get_type() { return type; }
+
   uchar *get_values() { return (uchar *) values; }
 
   void set_size (ulonglong sz) { size= (uint8) sz; }
 
-  void set_values (uchar *vals) { values= (uint8 *) vals; }
+  void set_type (Histogram_type t) { type= t; }
+
+  void set_values (uchar *vals) { values= (uchar *) vals; }
 
   void set_value(uint i, double val)
   {
-    values[i]= (uint8) (val * HIST_FACTOR);
+    switch (type) {
+    case SINGLE_PREC_HB:   
+      ((uint8 *) values)[i]= (uint8) (val * prec_factor());
+      return;
+    case DOUBLE_PREC_HB:
+      ((uint16 *) values)[i]= (uint16) (val * prec_factor());
+      return;
+    }
   }
 
-  void set_prev_value(uint i) { values[i]= values[i-1]; }
- 
+  void set_prev_value(uint i)
+  {
+    switch (type) {
+    case SINGLE_PREC_HB:   
+      ((uint8 *) values)[i]= ((uint8 *) values)[i-1];
+      return;
+    case DOUBLE_PREC_HB:
+      ((uint16 *) values)[i]= ((uint16 *) values)[i-1];
+      return;
+    }
+  }
 
   double range_selectivity(double min_pos, double max_pos)
   {
     double sel;
-    double bucket_sel= 1.0/(size + 1);  
+    double bucket_sel= 1.0/(get_width() + 1);  
     uint min= find_bucket(min_pos, TRUE);
     uint max= find_bucket(max_pos, FALSE);
     sel= bucket_sel * (max - min + 1);
@@ -176,14 +232,14 @@ public:
   double point_selectivity(double pos, double avg_sel)
   {
     double sel;
-    double bucket_sel= 1.0/(size + 1);  
+    double bucket_sel= 1.0/(get_width() + 1);  
     uint min= find_bucket(pos, TRUE);
     uint max= min;
-    while (max + 1 < size && values[max + 1] == values[max])
+    while (max + 1 < get_width() && get_value(max + 1) == get_value(max))
       max++;
-    double width= ((max + 1 == size ? 1.0 : values[max]) -
-	           (min == 0 ? 0.0 : values[min-1])) *
-	          INV_HIST_FACTOR;  
+    double width= ((max + 1 == get_width() ? 1.0 : get_value(max)) -
+	           (min == 0 ? 0.0 : get_value(min-1))) *
+      ((double) 1.0 / prec_factor());  
     sel= avg_sel * (bucket_sel * (max + 1 - min)) / width;
     return sel;
   }
diff --git a/sql/sys_vars.cc b/sql/sys_vars.cc
index 2bf9a55f8b1..aa0c4232466 100644
--- a/sql/sys_vars.cc
+++ b/sql/sys_vars.cc
@@ -3777,6 +3777,17 @@ static Sys_var_ulong Sys_histogram_size(
        SESSION_VAR(histogram_size), CMD_LINE(REQUIRED_ARG),
        VALID_RANGE(0, 255), DEFAULT(0), BLOCK_SIZE(1));
 
+const char *histogram_types[] =
+           {"SINGLE_PREC_HB", "DOUBLE_PREC_HB", 0};
+static Sys_var_enum Sys_histogram_type(
+       "histogram_type",
+       "Specifies type of the histograms created by ANALYZE. "
+       "Possible values are: "
+       "SINGLE_PREC_HB - single precision height-balanced, "
+       "DOUBLE_PREC_HB - double precision height-balanced.",
+       SESSION_VAR(histogram_type), CMD_LINE(REQUIRED_ARG),
+       histogram_types, DEFAULT(0));
+
 static Sys_var_mybool Sys_no_thread_alarm(
        "debug_no_thread_alarm",
        "Disable system thread alarm calls. Disabling it may be useful "