WL#2971 "change log-bin-trust-routine-creators=0 to apply only to functions".

Indeed now that stored procedures CALL is not binlogged, but instead the invoked substatements are,
the restrictions applied by log-bin-trust-routine-creators=0 are superfluous for procedures.
They still need to apply to functions where function calls are written to the binlog (for example as "DO myfunc(3)").
We rename the variable to log-bin-trust-function-creators but allow the old name until some future version (and issue a warning if old name is used).


mysql-test/mysql-test-run.pl:
  update to new option name
mysql-test/mysql-test-run.sh:
  update to new option name
mysql-test/mysql_test_run_new.c:
  update to new option name
mysql-test/r/rpl_sp.result:
  result update
mysql-test/t/rpl_sp-slave.opt:
  we need to skip this error to not hit BUG#14769
mysql-test/t/rpl_sp.test:
  Test update:
  1) as log-bin-trust-routine-creators now affects only functions, the testing of this option, which was
  mainly done on procedures, is moved to functions
  2) cleanup is simplified; and instead of many SHOW BINLOG EVENTS we do a big one in the end, which is more
  maintainable.
  3) we test a few more function and procedures cases to see how they replicate.
  4) removing out-of-date comments
sql/item_func.cc:
  This warning is wrong since binlogging of functions was changed in August. If a function fails
  in the middle, it will be binlogged with its error code (i.e. properly).
sql/mysql_priv.h:
  variable name changed
sql/mysqld.cc:
  option name changes. A precision about --read-only.
sql/set_var.cc:
  a new class sys_var_trust_routine_creators to be able to issue a "this is a deprecated variable" warning if used.
sql/set_var.h:
  new class to be able to issue a "this is a deprecated variable" warning if used.
sql/share/errmsg.txt:
  routine -> function
sql/sp.cc:
  log-bin-trust-routine-creators now applies only to functions.
sql/sql_parse.cc:
  1) sending ER_FAILED_ROUTINE_BREAK_BINLOG is wrong since August as we don't binlog CALL anymore but instead binlog the substatements;
  the clear_error() goes away too as it was necessary only when we created a binlog event from the "CALL" statement.
  2) log-bin-trust-routine-creators now applies only to functions.
sql/sql_trigger.cc:
  comments.

1 files changed, 3 insertions, 2 deletions

diff --git a/sql/sp.cc b/sql/sp.cc
index 8386c5d58a2..e01d05cf405 100644
--- a/sql/sp.cc
+++ b/sql/sp.cc
@@ -551,12 +551,13 @@ db_create_routine(THD *thd, int type, sp_head *sp)
 	store(sp->m_chistics->comment.str, sp->m_chistics->comment.length,
 	      system_charset_info);
 
-    if (!trust_routine_creators && mysql_bin_log.is_open())
+    if ((sp->m_type == TYPE_ENUM_FUNCTION) &&
+        !trust_function_creators && mysql_bin_log.is_open())
     {
       if (!sp->m_chistics->detistic)
       {
 	/*
-	  Note that for a _function_ this test is not enough; one could use
+	  Note that this test is not perfect; one could use
 	  a non-deterministic read-only function in an update statement.
 	*/
 	enum enum_sp_data_access access=