count(distinct) cleanup

mysql-test/t/count_distinct2.test

@@ -45,7 +45,7 @@ select count(distinct n2), n1 from t1 group by n1;
 drop table t1;
 
 # test the converstion from tree to MyISAM
-create table t1 (n int);
+create table t1 (n int default NULL);
 let $1=5000;
 while ($1)
 {

sql/item_sum.cc

@@ -839,7 +839,8 @@ int dump_leaf(byte* key, uint32 count __attribute__((unused)),
 {
   char* buf = item->table->record[0];
   int error;
-  memset(buf, 0xff, item->rec_offset); // make up for cheating in the tree
+  // the first item->rec_offset bytes are taken care of with
+  // restore_record(table,2) in setup()
   memcpy(buf + item->rec_offset, key, item->tree.size_of_element);
   if ((error = item->table->file->write_row(buf)))
   {
@@ -888,12 +889,17 @@ bool Item_sum_count_distinct::setup(THD *thd)
   table->file->extra(HA_EXTRA_NO_ROWS);		// Don't update rows
   table->no_rows=1;
 
+  
   if(table->db_type == DB_TYPE_HEAP) // no blobs, otherwise it would be
     // MyISAM
     {
       qsort_cmp2 compare_key;
       void* cmp_arg;
       int key_len;
+
+      // to make things easier for dump_leaf if we ever have to dump to
+      // MyISAM
+      restore_record(table,2);
       
       if(table->fields == 1) // if we have only one field, which is
 	// the most common use of count(distinct), it is much faster
@@ -941,7 +947,7 @@ bool Item_sum_count_distinct::setup(THD *thd)
       // but this has to be handled - otherwise someone can crash
       // the server with a DoS attack
       max_elements_in_tree = (key_len) ? max_heap_table_size/key_len :
-	max_heap_table_size;
+	1;
     }
   
   return 0;

sql/item_sum.h

@@ -148,19 +148,20 @@ class Item_sum_count_distinct :public Item_sum_int
   bool fix_fields(THD *thd,TABLE_LIST *tables);
   TMP_TABLE_PARAM *tmp_table_param;
   TREE tree;
-  uint max_elements_in_tree;
+  
   // calculated based on max_heap_table_size. If reached,
   // walk the tree and dump it into MyISAM table
+  uint max_elements_in_tree;
   
-  bool use_tree;
   // If there are no blobs, we can use a tree, which
   // is faster than heap table. In that case, we still use the table
   // to help get things set up, but we insert nothing in it
+  bool use_tree;
   
-  int rec_offset;
   // the first few bytes of record ( at least one)
   // are just markers for deleted and NULLs. We want to skip them since
   // they will just bloat the tree without providing any valuable info
+  int rec_offset;
 
   int tree_to_myisam();