WL#1596 "make mysqldump --master-data --single-transaction able to do online dump of InnoDB AND report reliable

binlog coordinates corresponding to the dump".
The good news is that now mysqldump can be used to get an online backup of InnoDB *which works for
point-in-time recovery and replication slave creation*. Formerly, mysqldump --master-data --single-transaction
used to call in fact mysqldump --master-data, so the dump was not an online dump (took big lock all time of dump).
The only lock which is now taken in this patch is at the beginning of the dump: mysqldump does:
FLUSH TABLES WITH READ LOCK; START TRANSACTION WITH CONSISTENT SNAPSHOT; SHOW MASTER STATUS; UNLOCK TABLES;
so the lock time is in fact the time FLUSH TABLES WITH READ LOCK takes to return (can be 0 or very long, if
a table is undergoing a huge update).
I have done some more minor changes listed in the paragraph of mysqldump.c.
WL#2237 "WITH CONSISTENT SNAPSHOT clause for START TRANSACTION":
it's a START TRANSACTION which additionally starts a consistent read on all
capable storage engine (i.e. InnoDB). So, can serve as a replacement for
BEGIN; SELECT * FROM some_innodb_table LIMIT 1; which starts a consistent read too. 


client/mysqldump.c:
  Main change: mysqldump --single-transaction --master-data is now able to, at the same time,
  take an online dump of InnoDB (using consistent read) AND get the binlog position corresponding to this dump
  (before, using the two options used to silently cancel --single-transaction).
  This uses the new START TRANSACTION WITH CONSISTENT SNAPSHOT syntax.
  Additional changes: 
  a) cleanup:
   - DBerror calls exit() so some code was unneeded
   - no need to call COMMIT at end, leave disconnection do the job
   - mysql_query_with_error_report()
  b) requirements I had heard from colleagues:
   - --master-data now requires an argument, to comment out ("--") the CHANGE MASTER or not
     (commenting had been asked for point-in-time recovery when replication is not necessary).
   - --first-slave is renamed to --lock-all-tables
  c) more sensible behaviours (has been discussed internally):
   - if used with --master-data, --flush-logs is probably intended to get a flush synchronous
     with the dump, not one random flush per dumped db.
   - disabled automatic reconnection as, at least, SQL_MODE would be lost (and also, depending
     on options, LOCK TABLES, BEGIN, FLUSH TABLES WITH READ LOCK).
include/mysqld_error.h:
  an error if START TRANSACTION WITH CONSISTENT SNAPSHOT is called and there is no consistent-read capable storage engine
  (idea ((C) PeterG) is that it's a bit like CREATE TABLE ENGINE=InnoDB when there is no support for InnoDB).
sql/handler.cc:
  new ha_start_consistent_snapshot(), which, inside an existing transaction, starts a consistent read
  (offers an alternative to SELECTing any InnoDB table). Does something only for InnoDB.
  Warning if no suitable engine supported.
sql/handler.h:
  declarations
sql/lex.h:
  symbols for lex
sql/share/czech/errmsg.txt:
  new message
sql/share/danish/errmsg.txt:
  new message
sql/share/dutch/errmsg.txt:
  new message
sql/share/english/errmsg.txt:
  new message
sql/share/estonian/errmsg.txt:
  new message
sql/share/french/errmsg.txt:
  new message
sql/share/german/errmsg.txt:
  new message
sql/share/greek/errmsg.txt:
  new message
sql/share/hungarian/errmsg.txt:
  new message
sql/share/italian/errmsg.txt:
  new message
sql/share/japanese/errmsg.txt:
  new message
sql/share/korean/errmsg.txt:
  new message
sql/share/norwegian-ny/errmsg.txt:
  new message
sql/share/norwegian/errmsg.txt:
  new message
sql/share/polish/errmsg.txt:
  new message
sql/share/portuguese/errmsg.txt:
  new message
sql/share/romanian/errmsg.txt:
  new message
sql/share/russian/errmsg.txt:
  new message
sql/share/serbian/errmsg.txt:
  new message
sql/share/slovak/errmsg.txt:
  new message
sql/share/spanish/errmsg.txt:
  new message
sql/share/swedish/errmsg.txt:
  new message
sql/share/ukrainian/errmsg.txt:
  new message
sql/sql_lex.h:
  new option in lex (transaction options)
sql/sql_parse.cc:
  warning comment (never make UNLOCK TABLES commit a transaction, please);
  support for starting consistent snapshot.
sql/sql_yacc.yy:
  new clause WITH CONSISTENT SNAPSHOT (syntax ok'd by PeterG) for START TRANSACTION.

1 files changed, 18 insertions, 3 deletions

diff --git a/sql/sql_yacc.yy b/sql/sql_yacc.yy
index 4538cb6e6ac..1ef91bcb257 100644
--- a/sql/sql_yacc.yy
+++ b/sql/sql_yacc.yy
@@ -131,6 +131,7 @@ bool my_yyoverflow(short **a, YYSTYPE **b, ulong *yystacksize);
 %token	CLIENT_SYM
 %token	COMMENT_SYM
 %token	COMMIT_SYM
+%token  CONSISTENT_SYM
 %token	COUNT_SYM
 %token	CREATE
 %token	CROSS
@@ -165,6 +166,7 @@ bool my_yyoverflow(short **a, YYSTYPE **b, ulong *yystacksize);
 %token	SELECT_SYM
 %token	SHOW
 %token	SLAVE
+%token  SNAPSHOT_SYM
 %token	SQL_THREAD
 %token	START_SYM
 %token	STD_SYM
@@ -618,6 +620,7 @@ bool my_yyoverflow(short **a, YYSTYPE **b, ulong *yystacksize);
         table_option opt_if_not_exists opt_no_write_to_binlog opt_var_type
         opt_var_ident_type delete_option opt_temporary all_or_any opt_distinct
         opt_ignore_leaves fulltext_options spatial_type union_option
+        start_transaction_opts
 
 %type <ulong_num>
 	ULONG_NUM raid_types merge_insert_types
@@ -2095,10 +2098,20 @@ slave:
 
 
 start:
-	START_SYM TRANSACTION_SYM { Lex->sql_command = SQLCOM_BEGIN;}
-	{}
+	START_SYM TRANSACTION_SYM start_transaction_opts
+        {
+           Lex->sql_command = SQLCOM_BEGIN;
+           Lex->start_transaction_opt= $3;
+        }
 	;
 
+start_transaction_opts:
+        /*empty*/ { $$ = 0; }
+        | WITH CONSISTENT_SYM SNAPSHOT_SYM
+        {
+           $$= MYSQL_START_TRANS_OPT_WITH_CONS_SNAPSHOT;
+        }
+
 slave_thread_opts:
 	{ Lex->slave_thd_opt= 0; }
 	slave_thread_opt_list
@@ -5122,6 +5135,7 @@ keyword:
 	| COMMIT_SYM		{}
 	| COMPRESSED_SYM	{}
 	| CONCURRENT		{}
+	| CONSISTENT_SYM	{}
 	| CUBE_SYM		{}
 	| DATA_SYM		{}
 	| DATETIME		{}
@@ -5262,6 +5276,7 @@ keyword:
 	| SHARE_SYM		{}
 	| SHUTDOWN		{}
 	| SLAVE			{}
+	| SNAPSHOT_SYM		{}
 	| SOUNDS_SYM		{}
 	| SQL_CACHE_SYM		{}
 	| SQL_BUFFER_RESULT	{}
@@ -5888,7 +5903,7 @@ grant_option:
         ;
 
 begin:
-	BEGIN_SYM   { Lex->sql_command = SQLCOM_BEGIN;} opt_work {}
+	BEGIN_SYM   { Lex->sql_command = SQLCOM_BEGIN; Lex->start_transaction_opt= 0;} opt_work {}
 	;
 
 opt_work: