Post-merge and post-review fixes for the patch for

Bug#23631 "Events: SHOW VARIABLES doesn't work when mysql.event 
is damaged:


mysql-test/r/events.result:
  Update results (a post-merge fix)
mysql-test/r/events_bugs.result:
  Update results (a post-merge fix)
mysql-test/r/events_scheduling.result:
  Update results (a post-merge fix)
mysql-test/t/events_scheduling.test:
  Make sure this test has no races.
sql/event_data_objects.cc:
  Manual post-merge fix for the events replication patch.
sql/event_data_objects.h:
  A post-merge fix.
sql/event_db_repository.cc:
  A post-merge fix.
sql/event_scheduler.cc:
  We should drop the event inside ::execute since there we have
  the right credentials to do so (otherwise Events::drop_event
  returns "access denied" error).
sql/events.cc:
  A post-review fix for: rename start_or_stop_event_scheduler
  to switch_event_scheduler_state.
sql/events.h:
  A post-review fix for: rename start_or_stop_event_scheduler
  to switch_event_scheduler_state.
sql/set_var.cc:
  A post-review fix for: rename start_or_stop_event_scheduler
  to switch_event_scheduler_state.
sql/sql_yacc.yy:
  Remove unused declaratoins.

1 files changed, 1 insertions, 22 deletions

diff --git a/sql/event_scheduler.cc b/sql/event_scheduler.cc
index a9a93cbc74b..fa5cde9a43a 100644
--- a/sql/event_scheduler.cc
+++ b/sql/event_scheduler.cc
@@ -309,7 +309,7 @@ Event_worker_thread::run(THD *thd, Event_queue_element_for_exec *event)
 
   thd->enable_slow_log= TRUE;
 
-  ret= job_data->execute(thd);
+  ret= job_data->execute(thd, event->dropped);
 
   print_warnings(thd, job_data);
 
@@ -338,27 +338,6 @@ Event_worker_thread::run(THD *thd, Event_queue_element_for_exec *event)
 end:
   delete job_data;
 
-  if (event->dropped)
-  {
-    sql_print_information("Event Scheduler: Dropping %s.%s",
-                          event->dbname.str, event->name.str);
-    /*
-      Using db_repository can lead to a race condition because we access
-      the table without holding LOCK_metadata.
-      Scenario:
-      1. CREATE EVENT xyz AT ...    (conn thread)
-      2. execute xyz                (worker)
-      3. CREATE EVENT XYZ EVERY ... (conn thread)
-      4. drop xyz                   (worker)
-      5. XYZ was just created on disk but `drop xyz` of the worker dropped it.
-         A consequent load to create Event_queue_element will fail.
-
-      If all operations are performed under LOCK_metadata there is no such
-      problem. However, this comes at the price of introduction bi-directional
-      association between class Events and class Event_worker_thread.
-    */
-    Events::drop_event(thd, event->dbname, event->name, FALSE);
-  }
   DBUG_PRINT("info", ("Done with Event %s.%s", event->dbname.str,
              event->name.str));