A fix and a test case for Bug#16248 "WHERE (col1,col2) IN ((?,?))

gives wrong results". Implement previously missing 
Item_row::cleanup. The bug is not repeatable in 5.0, probably 
due to a coincidence: the problem is present in 5.0 as well.


mysql-test/r/ps.result:
  Update the result file (Bug#16248)
mysql-test/t/ps.test:
  Add a test case for Bug#16248 "WHERE (col1,col2) IN ((?,?)) gives 
  wrong results"
sql/item_row.cc:
  Implement Item_row::cleanup(): we should reset used_tables_cache
  before reexecution of a prepared statement. In case ROW
  arguments contain a placeholder, used_tables_cache has PARAM_TABLE
  bit set in statement prepare. As a result, when executing a statement,
  the condition push down algorithm (make_cond_for_table) would think
  that the WHERE clause belongs to the non-existent PARAM_TABLE and
  wouldn't attach the WHERE clause to any of the real tables, 
  effectively optimizing the clause away.
sql/item_row.h:
  Remove a never used member 'array_holder'. Add declaration for
  Item_row::cleanup.

1 files changed, 15 insertions, 1 deletions

diff --git a/sql/item_row.cc b/sql/item_row.cc
index 12d202a1699..493eefc9ff0 100644
--- a/sql/item_row.cc
+++ b/sql/item_row.cc
@@ -26,7 +26,7 @@
 */
 
 Item_row::Item_row(List<Item> &arg):
-  Item(), used_tables_cache(0), array_holder(1), const_item_cache(1), with_null(0)
+  Item(), used_tables_cache(0), const_item_cache(1), with_null(0)
 {
 
   //TODO: think placing 2-3 component items in item (as it done for function)
@@ -85,6 +85,20 @@ bool Item_row::fix_fields(THD *thd, TABLE_LIST *tabl, Item **ref)
 }
 
 
+void Item_row::cleanup()
+{
+  DBUG_ENTER("Item_row::cleanup");
+
+  Item::cleanup();
+  /* Reset to the original values */
+  used_tables_cache= 0;
+  const_item_cache= 1;
+  with_null= 0;
+
+  DBUG_VOID_RETURN;
+}
+
+
 void Item_row::split_sum_func(THD *thd, Item **ref_pointer_array,
                               List<Item> &fields)
 {