MWL#90: Address review feedback part #5

sql/item_subselect.cc

@@ -33,11 +33,11 @@
 
 
 Item_subselect::Item_subselect():
-  Item_result_field(), value_assigned(0), thd(0), substitution(0),
-  expr_cache(0), engine(0), old_engine(0), used_tables_cache(0),
-  have_to_be_excluded(0), const_item_cache(1), inside_first_fix_fields(0),
-  done_first_fix_fields(FALSE), eliminated(FALSE), engine_changed(0),
-  changed(0), is_correlated(FALSE)
+  Item_result_field(), value_assigned(0), thd(0), old_engine(0), 
+  used_tables_cache(0), have_to_be_excluded(0), const_item_cache(1),
+  inside_first_fix_fields(0), done_first_fix_fields(FALSE), 
+  substitution(0), expr_cache(0), engine(0), eliminated(FALSE),
+  engine_changed(0), changed(0), is_correlated(FALSE)
 {
   with_subselect= 1;
   reset();

sql/item_subselect.h

@@ -36,6 +36,22 @@ class Item_subselect :public Item_result_field
 protected:
   /* thread handler, will be assigned in fix_fields only */
   THD *thd;
+  /* old engine if engine was changed */
+  subselect_engine *old_engine;
+  /* cache of used external tables */
+  table_map used_tables_cache;
+  /* allowed number of columns (1 for single value subqueries) */
+  uint max_columns;
+  /* where subquery is placed */
+  enum_parsing_place parsing_place;
+  /* work with 'substitution' */
+  bool have_to_be_excluded;
+  /* cache of constant state */
+  bool const_item_cache;
+  
+  bool inside_first_fix_fields;
+  bool done_first_fix_fields;
+public:
   /* 
     Used inside Item_subselect::fix_fields() according to this scenario:
       > Item_subselect::fix_fields
@@ -49,30 +65,12 @@ protected:
         substitution= NULL;
       < Item_subselect::fix_fields
   */
-public:
   Item *substitution;
   /* unit of subquery */
   st_select_lex_unit *unit;
   Item *expr_cache;
   /* engine that perform execution of subselect (single select or union) */
   subselect_engine *engine;
-protected:
-  /* old engine if engine was changed */
-  subselect_engine *old_engine;
-  /* cache of used external tables */
-  table_map used_tables_cache;
-  /* allowed number of columns (1 for single value subqueries) */
-  uint max_columns;
-  /* where subquery is placed */
-  enum_parsing_place parsing_place;
-  /* work with 'substitution' */
-  bool have_to_be_excluded;
-  /* cache of constant state */
-  bool const_item_cache;
-  
-  bool inside_first_fix_fields;
-  bool done_first_fix_fields;
-public:
   /* A reference from inside subquery predicate to somewhere outside of it */
   class Ref_to_outside : public Sql_alloc
   {
@@ -774,61 +772,26 @@ public:
     of subselect_single_select_engine::[prepare | cols].
   */
   subselect_single_select_engine *materialize_engine;
-protected:
-  /* The engine used to compute the IN predicate. */
-  subselect_engine *lookup_engine;
   /*
     QEP to execute the subquery and materialize its result into a
     temporary table. Created during the first call to exec().
   */
-public:
   JOIN *materialize_join;
-protected:
-  /* Keyparts of the only non-NULL composite index in a rowid merge. */
-  MY_BITMAP non_null_key_parts;
-  /* Keyparts of the single column indexes with NULL, one keypart per index. */
-  MY_BITMAP partial_match_key_parts;
-  uint count_partial_match_columns;
-  uint count_null_only_columns;
   /*
     A conjunction of all the equality condtions between all pairs of expressions
     that are arguments of an IN predicate. We need these to post-filter some
     IN results because index lookups sometimes match values that are actually
     not equal to the search key in SQL terms.
- */
-public:
+  */
   Item_cond_and *semi_join_conds;
-protected:
-  /* Possible execution strategies that can be used to compute hash semi-join.*/
-  enum exec_strategy {
-    UNDEFINED,
-    COMPLETE_MATCH, /* Use regular index lookups. */
-    PARTIAL_MATCH,  /* Use some partial matching strategy. */
-    PARTIAL_MATCH_MERGE, /* Use partial matching through index merging. */
-    PARTIAL_MATCH_SCAN,  /* Use partial matching through table scan. */
-    IMPOSSIBLE      /* Subquery materialization is not applicable. */
-  };
-  /* The chosen execution strategy. Computed after materialization. */
-  exec_strategy strategy;
-protected:
-  exec_strategy get_strategy_using_schema();
-  exec_strategy get_strategy_using_data();
-  ulonglong rowid_merge_buff_size(bool has_non_null_key,
-                                  bool has_covering_null_row,
-                                  MY_BITMAP *partial_match_key_parts);
-  void choose_partial_match_strategy(bool has_non_null_key,
-                                     bool has_covering_null_row,
-                                     MY_BITMAP *partial_match_key_parts);
-  bool make_semi_join_conds();
-  subselect_uniquesubquery_engine* make_unique_engine();
 
-public:
   subselect_hash_sj_engine(THD *thd, Item_subselect *in_predicate,
                            subselect_single_select_engine *old_engine)
-    :subselect_engine(thd, in_predicate, NULL), tmp_table(NULL),
-    is_materialized(FALSE), materialize_engine(old_engine), lookup_engine(NULL),
-    materialize_join(NULL), count_partial_match_columns(0),
-    count_null_only_columns(0), semi_join_conds(NULL), strategy(UNDEFINED)
+    : subselect_engine(thd, in_predicate, NULL), 
+      tmp_table(NULL), is_materialized(FALSE), materialize_engine(old_engine),
+      materialize_join(NULL),  semi_join_conds(NULL), lookup_engine(NULL),
+      count_partial_match_columns(0), count_null_only_columns(0),
+      strategy(UNDEFINED)
   {}
   ~subselect_hash_sj_engine();
 
@@ -856,6 +819,38 @@ public:
   //=>base class
   bool change_result(Item_subselect *si, select_result_interceptor *result);
   bool no_tables();//=>base class
+
+protected:
+  /* The engine used to compute the IN predicate. */
+  subselect_engine *lookup_engine;
+  /* Keyparts of the only non-NULL composite index in a rowid merge. */
+  MY_BITMAP non_null_key_parts;
+  /* Keyparts of the single column indexes with NULL, one keypart per index. */
+  MY_BITMAP partial_match_key_parts;
+  uint count_partial_match_columns;
+  uint count_null_only_columns;
+  /* Possible execution strategies that can be used to compute hash semi-join.*/
+  enum exec_strategy {
+    UNDEFINED,
+    COMPLETE_MATCH, /* Use regular index lookups. */
+    PARTIAL_MATCH,  /* Use some partial matching strategy. */
+    PARTIAL_MATCH_MERGE, /* Use partial matching through index merging. */
+    PARTIAL_MATCH_SCAN,  /* Use partial matching through table scan. */
+    IMPOSSIBLE      /* Subquery materialization is not applicable. */
+  };
+  /* The chosen execution strategy. Computed after materialization. */
+  exec_strategy strategy;
+  exec_strategy get_strategy_using_schema();
+  exec_strategy get_strategy_using_data();
+  ulonglong rowid_merge_buff_size(bool has_non_null_key,
+                                  bool has_covering_null_row,
+                                  MY_BITMAP *partial_match_key_parts);
+  void choose_partial_match_strategy(bool has_non_null_key,
+                                     bool has_covering_null_row,
+                                     MY_BITMAP *partial_match_key_parts);
+  bool make_semi_join_conds();
+  subselect_uniquesubquery_engine* make_unique_engine();
+
 };
 
 

sql/opt_subselect.cc

@@ -826,15 +826,12 @@ void get_delayed_table_estimates(TABLE *table,
     ((subselect_hash_sj_engine*)item->engine);
   JOIN *join= hash_sj_engine->materialize_join;
 
-  double rows= 1;
-  double read_time= 0.0;
+  double rows;
+  double read_time;
 
   /* Calculate #rows and cost of join execution */
-  for (uint i= join->const_tables; i < join->tables; i++)
-  {
-    rows      *= join->best_positions[i].records_read;
-    read_time += join->best_positions[i].read_time;
-  }
+  get_partial_join_cost(join, join->tables, &read_time, &rows);
+
   *out_rows= (ha_rows)rows;
   *startup_cost= read_time;
   /* Calculate cost of scanning the temptable */

sql/sql_join_cache.cc

@@ -185,6 +185,7 @@ void JOIN_CACHE::calc_record_fields()
   start_tab= tab;
   if (start_tab->bush_children)
     start_tab= start_tab->bush_children->start;
+  DBUG_ASSERT(!start_tab->bush_children);
 
   tab= start_tab;
 
@@ -508,7 +509,6 @@ void JOIN_CACHE::create_key_arg_fields()
   /* Now create local fields that are used to build ref for this key access */
   copy= field_descr+flag_fields;
   for (tab= start_tab; tab != join_tab; tab= next_linear_tab(join, tab, FALSE))
-  //for (tab= join_tab-tables; tab; tab= get_next_table(tab))
   {
     length+= add_table_data_fields_to_join_cache(tab, &tab->table->tmp_set,
                                                  &data_field_count, &copy,
@@ -721,8 +721,11 @@ ulong JOIN_CACHE::get_min_join_buffer_size()
   if (!min_buff_size)
   {
     size_t len= 0;
-    for (JOIN_TAB *tab= start_tab; tab != join_tab; tab= next_linear_tab(join, tab, FALSE))
+    for (JOIN_TAB *tab= start_tab; tab != join_tab; 
+         tab= next_linear_tab(join, tab, FALSE))
+    {
       len+= tab->get_max_used_fieldlength();
+    }
     len+= get_record_max_affix_length() + get_max_key_addon_space_per_record();  
     size_t min_sz= len*min_records;
     size_t add_sz= 0;
@@ -774,8 +777,11 @@ ulong JOIN_CACHE::get_max_join_buffer_size(bool optimize_buff_size)
     size_t max_sz;
     size_t min_sz= get_min_join_buffer_size(); 
     size_t len= 0;
-    for (JOIN_TAB *tab= start_tab; tab != join_tab; tab= next_linear_tab(join, tab, FALSE))
+    for (JOIN_TAB *tab= start_tab; tab != join_tab;
+         tab= next_linear_tab(join, tab, FALSE))
+    {
       len+= tab->get_used_fieldlength();
+    }
     len+= get_record_max_affix_length();
     avg_record_length= len;
     len+= get_max_key_addon_space_per_record() + avg_aux_buffer_incr;

sql/sql_select.cc

@@ -1035,31 +1035,15 @@ JOIN::optimize()
     Perform the optimization on fields evaluation mentioned above
     for all on expressions.
   */
-
+  for (JOIN_TAB *tab= first_linear_tab(this, TRUE); tab;
+       tab= next_linear_tab(this, tab, TRUE))
   {
-    List_iterator<JOIN_TAB_RANGE> it(join_tab_ranges);
-    JOIN_TAB_RANGE *jt_range;
-    /* For upper level JOIN_TABs, we need to skip the const tables: */
-    uint first_tab_offs= const_tables;
-    while ((jt_range= it++))
+    if (*tab->on_expr_ref)
     {
-      for (JOIN_TAB *tab= jt_range->start + first_tab_offs;
-           tab < jt_range->end; tab++)
-      {
-        if (*tab->on_expr_ref)
-        {
-          *tab->on_expr_ref= substitute_for_best_equal_field(*tab->on_expr_ref,
-                                                             tab->cond_equal,
-                                                             map2table);
-          (*tab->on_expr_ref)->update_used_tables();
-        }
-      }
-      /*
-        Next jt_range will refer to SJM nest (and not the top-level range). 
-        Inside SJM nests, we dont have const tables, so should start from the
-        first table:
-      */
-      first_tab_offs= 0;
+      *tab->on_expr_ref= substitute_for_best_equal_field(*tab->on_expr_ref,
+                                                         tab->cond_equal,
+                                                         map2table);
+      (*tab->on_expr_ref)->update_used_tables();
     }
   }
 
@@ -1067,47 +1051,34 @@ JOIN::optimize()
     Perform the optimization on fields evaliation mentioned above
     for all used ref items.
   */
-  //for (JOIN_TAB *tab= join_tab + const_tables; tab < join_tab + tables; tab++)
-  //{
+  for (JOIN_TAB *tab= first_linear_tab(this, TRUE); tab;
+       tab= next_linear_tab(this, tab, TRUE))
   {
-    List_iterator<JOIN_TAB_RANGE> it(join_tab_ranges);
-    JOIN_TAB_RANGE *jt_range;
-    uint first_tab_offs= const_tables;
-    while ((jt_range= it++))
-    {
-      for (JOIN_TAB *tab= jt_range->start + first_tab_offs;
-           tab < jt_range->end; tab++)
-      {
     uint key_copy_index=0;
-        for (uint i=0; i < tab->ref.key_parts; i++)
-        {
-          
-          Item **ref_item_ptr= tab->ref.items+i;
-          Item *ref_item= *ref_item_ptr;
+    for (uint i=0; i < tab->ref.key_parts; i++)
+    {
+      Item **ref_item_ptr= tab->ref.items+i;
+      Item *ref_item= *ref_item_ptr;
       if (!ref_item->used_tables() && !(select_options & SELECT_DESCRIBE))
         continue;
-          COND_EQUAL *equals= tab->first_inner ? tab->first_inner->cond_equal : 
-                                                 cond_equal;
-          ref_item= substitute_for_best_equal_field(ref_item, equals, map2table);
-          ref_item->update_used_tables();
-          if (*ref_item_ptr != ref_item)
-          {
-            *ref_item_ptr= ref_item;
-            Item *item= ref_item->real_item();
+      COND_EQUAL *equals= tab->first_inner ? tab->first_inner->cond_equal : 
+                                             cond_equal;
+      ref_item= substitute_for_best_equal_field(ref_item, equals, map2table);
+      ref_item->update_used_tables();
+      if (*ref_item_ptr != ref_item)
+      {
+        *ref_item_ptr= ref_item;
+        Item *item= ref_item->real_item();
         store_key *key_copy= tab->ref.key_copy[key_copy_index];
-            if (key_copy->type() == store_key::FIELD_STORE_KEY)
-            {
-              store_key_field *field_copy= ((store_key_field *)key_copy);
-              field_copy->change_source_field((Item_field *) item);
-            }
-          }
-      key_copy_index++;
+        if (key_copy->type() == store_key::FIELD_STORE_KEY)
+        {
+          store_key_field *field_copy= ((store_key_field *)key_copy);
+          field_copy->change_source_field((Item_field *) item);
         }
       }
-      first_tab_offs= 0;
+      key_copy_index++;
     }
   }
-  //}
 
   if (conds && const_table_map != found_const_table_map &&
       (select_options & SELECT_DESCRIBE))
@@ -1638,9 +1609,6 @@ bool JOIN::setup_subquery_caches()
     for (JOIN_TAB *tab= first_linear_tab(this, TRUE); 
          tab; 
          tab= next_linear_tab(this, tab, TRUE))
-    //for (JOIN_TAB *tab= join_tab + const_tables;
-    //     tab < join_tab + tables ;
-    //     tab++)
     {
       if (tab->select_cond)
         tab->select_cond=