bk commit into 5.1 tree (sergefp:1.2580) BUG#35850

[Sergey Petrunia]

Below is the list of changes that have just been committed into a local
5.1 repository of sergefp.  When sergefp does a push these changes
will be propagated to the main repository and, within 24 hours after the
push, to the public repository.
For information on how to access the public repository
see http://dev.mysql.com/doc/mysql/en/installing-source-tree.html

ChangeSet@stripped, 2008-04-23 08:35:21+04:00, sergefp@stripped +1 -0
  BUG#35850: Subquery (actually table access) performance regression in 5.1.23/5.1.24:
  - Disable the rule introduced by BUG#35850.
  This is an experimental patch

  sql/sql_select.cc@stripped, 2008-04-23 08:35:11+04:00, sergefp@stripped +6 -0
    BUG#35850: Subquery (actually table access) performance regression in 5.1.23/5.1.24:
    - Disable the rule introduced by BUG#35850

diff -Nrup a/sql/sql_select.cc b/sql/sql_select.cc
--- a/sql/sql_select.cc	2008-03-31 11:40:36 +04:00
+++ b/sql/sql_select.cc	2008-04-23 08:35:11 +04:00
@@ -6504,6 +6504,11 @@ make_join_readinfo(JOIN *join, ulonglong
 	  else if (!table->covering_keys.is_clear_all() &&
 		   !(tab->select && tab->select->quick))
 	  {					// Only read index tree
+            /*
+              The below turned out to be not true in practice (see
+              BUG#35850), so temporarily disabling this rule:
+            */
+#if 0
 	    /*
 	      See bug #26447: "Using the clustered index for a table scan
 	      is always faster than using a secondary index".
@@ -6512,6 +6517,7 @@ make_join_readinfo(JOIN *join, ulonglong
                 table->file->primary_key_is_clustered())
               tab->index= table->s->primary_key;
             else
+#endif              
               tab->index=find_shortest_key(table, & table->covering_keys);
 	    tab->read_first_record= join_read_first;
 	    tab->type=JT_NEXT;		// Read with index_first / index_next