MySQL developing Store Procedure

MySQL Stored Procedure Programming

A practical guide

Marco Tusa

Senior Consultant
marco.tusa@mysql.com

Copyright 2008 Marco Tusa, MySQL AB 1

The World’s Most Popular Open Source
Database
Stored Procedure Programming

3. Stored Programming Fundamentals
4. Stored Program Construction
5. Using MySQL Stored Programs in Applications
6. Optimizing Stored Programs



Stored Programming Fundamentals



• Introduction to MySQL Stored Programs
• MySQL Stored Programming Tutorial
• Language Fundamentals
• Blocks, Conditional Statements, and Iterative Programming
• Using SQL in Stored Programming
• Error Handling


Introduction to MySQL Stored Programs
• What is a Stored Program?
• MySQL SP
– Store Procedure
– Sore Function
– Trigger
• MySQL Implementation
– ANSI SQL:2003 SQL/PSM (Persistent Stored Module) specification the only available open
standard for these types of programs (like IBM's DB2 ).
– The MySQL stored program language is a block-structured language (like Pascal) that
includes familiar commands for manipulating variables, implementing conditional execution,
performing iterative processing, and handling errors.


Introduction to MySQL Stored Programs
• Why Use Stored Programs?
– The use of stored programs can lead to a more secure database.
– Stored programs offer a mechanism to abstract data access routines, which can improve the
maintainability of your code as underlying data structures evolve.
– Stored programs can reduce network traffic, because the program can work on the data
from within the server, rather than having to transfer the data across the network.
– Stored programs can be used to implement common routines accessible from multiple
applications possibly using otherwise incompatible frameworks executed either within or
from outside the database server.
– Database-centric logic can be isolated in stored programs and implemented by
programmers with more specialized, database experience.
– The use of stored programs can, under some circumstances, improve the portability of your
application.


MySQL Stored Programming Tutorial
What we will do

• How to create a stored program
• How to pass information in and out of the stored program
• How to interact with the database
• How to create procedures, functions, and triggers in the MySQL
stored program language



What do I need?
• A MySQL server

• A text editor such as vi, emacs, or Notepad

• The MySQL Query Browser


My First Store Procedure
DELIMITER$$
DROP PROCEDURE IF EXIST HelloWorld()$$
CREATE PROCEDURE HelloWorld()
BEGIN
SELECT “Hello World”;
END$$
DELIMITER ;
mysql> CALL HelloWorld();
+-------------+
| Hello World |
+-------------+
| Hello World |
+-------------+
1 row in set (0.01 sec)
Query OK, 0 rows affected (0.01 sec)


• Variable
– Local variables can be declared within stored procedures using the DECLARE statement. Variable names
follow the same naming rules as MySQL table column names and can be of any MySQL data type. You can
give variables an initial value with the DEFAULT clause and assign them new values using the SET
command.
• Parameters
– Place parameters within parentheses that are located immediately after the name of the stored procedure.
Each parameter has a name, a data type, and, optionally, a mode. Valid modes are IN (read-only), INOUT
(read-write), and OUT (write-only). The IN mode is the default.


• Conditional Execution
– IF THEN (ELSEIF) (ELSE) END IF
– CASE

• Loops
– LOOP
– WHILE
– REPEAT … UNTILL


Interacting with the Database

• SELECT INTO is fine for single-row queries
• CURSORS
• Unbound result set (return result set to calling application)
• Non SELECT actions (update, insert, delete)
• Internal CALL to other Store Procedure


Stored Functions

Stored functions are similar to stored procedures: they are named program units that
contain one or more MySQL statements. They differ from procedures in the following
ways:
• The parameter list of a function may contain only IN parameters. OUT and INOUT
parameters are not allowed. Specifying the IN keyword is neither required nor
allowed.

• The function itself must return a single value, whose type is defined in the header of
the function.

• Functions can be called from within SQL statements.

• A function may not return a result set.


CREATE FUNCTION `FirstToUppercase`(stringIn VARCHAR(255) ) RETURNS varchar(255) CHARSET latin1
BEGIN
DECLARE toConvert VARCHAR(255);
DECLARE firstLetter CHAR(1);
DECLARE allTerm VARCHAR(255);
SET toConvert = stringIn;
IF ISNULL(toConvert) = 1 THEN
RETURN(NULL);
END IF;
SET firstLetter = UCASE(SUBSTRING(toConvert,1,1));
SET allTerm = SUBSTRING(toConvert,2);
RETURN (CONCAT(firstLetter,allTerm));
END $$
DELIMITER ;

tusa@localhost:3306 [world]> SELECT FirstToUppercase('adamo');
+---------------------------+
| FirstToUppercase('adamo') |
+---------------------------+
| Adamo |
+---------------------------+


Triggers
• A trigger is a special type of stored program that fires when a table is modified by an
INSERT, UPDATE, or DELETE (DML) statement. Triggers implement functionality
that must take place whenever a certain change occurs to the table. Because triggers
are attached directly to the table, application code cannot bypass database triggers.

• Typical uses of triggers include the implementation of critical business logic, the
denormalization of data for performance reasons, and the auditing of changes made
to a table. Triggers can be defined to fire before or after a specific DML statement
executes.


DELIMITER $$
CREATE TRIGGER nameInUppercase
BEFORE INSERT ON city FOR EACH ROW
BEGIN
DECLARE dummy INT;
IF ISNULL(NEW.Name)<1 THEN
SET NEW.Name=FirstToUppercase(NEW.Name);
END IF;
END$$
DELIMITER ;

insert into city (Name) values ('adamo');
tusa@localhost:3306 [world]> Select Name from city where Name like "Ad%" Order by name limit 1;
+-------+
| Name |
+-------+
| Adamo |
+-------+



Language Fundamentals

TO DO


Blocks, Conditional Statements, and Iterative Programming

• Simple loop
– Continues until you issue a LEAVE statement to terminate the loop

• REPEAT UNTIL loop
– Continues until an expression evaluates as true

• WHILE loop
– Continues as long as an expression evaluates as true



• MySQL stored programs consist of one or more blocks
• Each block commences with a BEGIN statement and is terminated by an END
statement.
CREATE {PROCEDURE|FUNCTION|TRIGGER} program_name
BEGIN
program_statements
END;
• The purpose of a block is twofold:
– To logically group related code segments
– To control the scope of variables and other objects

• You can define a variable within a block that is not visible outside the block.
Furthermore, you can declare a variable within a block that overrides the definition of
a variable with the same name declared outside of the block.


• You can also name a block with a label. Labeling a block has the following
advantages:
– It improves code readabilityfor instance, by allowing you to quickly match the BEGIN statement with its
associated END statement.
– It allows you to terminate block execution with the LEAVE statement (see the section describing this
statement later in this chapter).

A simplified representation of the structure of a block is:
[label:] BEGIN
variable and condition declarations
cursor declarations
handler declarations
program code
END[label];

• Nested Blocks



Conditional Control
Conditional control or "flow of control"statements allow you to execute code based on the value of
some expression. Conditional control statements allow you to take different actions depending
on the value of such an expression, which could refer to parameters to the stored program, to
data in the database, or to other variable data (such as the day of the week or the time of the
day).
The MySQL stored program language supports two conditional control statements :
– IF
– CASE.
Both IF and CASE perform very similar functions, and there is always a way to rewrite an IF
statement as a CASE statement or vice versa. Usually, choosing between IF and CASE is a
matter of personal preference or programming standards. However, there are circumstances in
which one type of statement is more readable or efficient than the other.


IF expression THEN
statements that execute if the expression is TRUE
ELSEIF expression THEN
statements that execute if expression1 is TRUE
ELSE
statements that execute if all the preceding expressions are FALSE or NULL
END IF;

From our previous example (function FirstToUppercase() )

IF ISNULL(toConvert) = 1 THEN
RETURN(NULL);
END IF



The CASE statement is an alternative conditional execution or flow
control statement. Anything that can be done with CASE statements
can be done with IF statements (and vice versa), but CASE
statements are often more readable and efficient when multiple
conditions need to be evaluated, especially when the conditions all

compare the output from a single expression


• Simple CASE statement condition on top
CASE expression
WHEN value THEN
statements
[WHEN value THEN
statements ...]
[ELSE
statements]
END CASE;
• Searched CASE statement condition case by case (IF)
CASE
WHEN condition/expression THEN
statements
[WHEN condition/expression THEN
statements...]
[ELSE
statements]
END CASE;


Why / when to use a loop?

• A program that supports a user interface may run a main loop that waits for,
and then processes, user keystrokes (this doesn't apply to stored programs,
however).
• Many mathematical algorithms can be implemented only by loops in
computer programs.
• When processing a file, a program may loop through each record in the file
and perform computations.
• A database program may loop through the rows returned by a SELECT
statement



• In MySQL we have 3 type of loops:
– Simple
– Repeat … Until
– While



Simple loop
SET i=0;
loop1: LOOP
SET i=i+1;
IF i>=10 THEN /*Last number - exit loop*/
LEAVE loop1;
ELSEIF MOD(i,2)=0 THEN /*Even number - try again*/
ITERATE loop1;
END IF;
SELECT CONCAT(i," is an odd number");
END LOOP loop1;


REPEAT … UNTIL

The REPEAT and UNTIL statements can be used to create a loop that
continues until some logical condition is met.
SET i=0;
loop1: REPEAT
SET i=i+1;
IF MOD(i,2)<>0 THEN /*Even number - try again*/
Select concat(i," is an odd number");
END IF;
UNTIL i >= 10
END REPEAT;


While loop
• A WHILE loop executes as long as a condition is true. If the condition is not
true to begin with, then the loop will never executeunlike the REPEAT loop,
which is guaranteed to execute at least once.

SET i=1;
loop1: WHILE i<=10 DO
IF MOD(i,2)<>0 THEN /*Even number - try again*/
SELECT CONCAT(i," is an odd number");
END IF;
SET i=i+1;
END WHILE loop1;


Using SQL in Stored Programming
Several type of SQL
• Simple (non-SELECT) SQL statements that do not return a result set can be freely embedded
within stored procedures and functions.

• A SELECT statement that returns only a single row can pass its result INTO local variables.

• A SELECT statement that returns multiple rows can form the basis for a cursor that allows you to
loop through each row, taking whatever action you deem appropriate for that row.

• Any SELECT statement can be included in a stored procedure (but not in a stored function)
"unbound" by an INTO clause or a CURSOR statement. The result set from such a SQL
statement will be returned to the calling program (but not, alas, to a calling stored procedure).

• SQL statements can be prepared dynamically using MySQL server-side prepared statements (in
stored procedures only).



• When we include a SQL statement that does not return a result set
such as an UPDATE, INSERT, or SET statement within a stored
program, it will execute exactly as it would if it were executed in
some other context (such as if it were called from PHP or issued
from the MySQL command line).


CREATE PROCEDURE simple_sqls( )
BEGIN
DECLARE i INT DEFAULT 1;
/* Example of a utility statement */
SET autocommit=0;
/* Example of DDL statements */
DROP TABLE IF EXISTS test_table ;
CREATE TABLE test_table
(id INT PRIMARY KEY,
some_data VARCHAR(30))
ENGINE=innodb;
/* Example of an INSERT using a procedure variable */
WHILE (i<=10) DO
INSERT INTO TEST_TABLE VALUES(i,CONCAT("record ",i));
SET i=i+1;
END WHILE;
/* Example of an UPDATE using procedure variables*/
SET i=5;
UPDATE test_table
SET some_data=CONCAT("I updated row ",i)
WHERE id=i;
/* DELETE with a procedure variable */
DELETE FROM test_table
WHERE id>i;
END;


If you have a SELECT statement that returns only a single row, you can return
that row into stored program variables by using the INTO statement within
the SELECT statement.

SELECT expression1 [, expression2 ....]
INTO variable1 [, variable2 ...]
other SELECT statement clauses



DELIMITER $$
DROP PROCEDURE IF EXISTS `world`.` GetCountryNameContinentByCode `$$
CREATE PROCEDURE `world`.`GetCountryNameContinentByCode`(codeIn CHAR(3))
Begin
DECLARE Cname VARCHAR(52);
DECLARE Ccontinent VARCHAR(50);

SELECT Name, Continent INTO Cname, Ccontinent FROM country WHERE code = TRIM(codeIn);
SELECT Cname, Ccontinent;

END $$
DELIMITER ;

tusa@localhost:3306 [world]> CALL GetCountryNameContinentByCode('AFG');
+-------------+------------+
| Cname | Ccontinent |
+-------------+------------+
| Afghanistan | Asia |
+-------------+------------+
Query OK, 0 rows affected (0.00 sec)



CURSOR

To handle a SELECT statement that returns more than one row, we must create and then
manipulate a cursor. A cursor is an object that provides programmatic access to the
result set returned by your SELECT statement. Use a cursor to iterate through the
rows in the result set and take action for each row individually.

In MySQL for now the CURSORS are forward ONLY. This means you can scroll the
cursor from the first to the last not vice versa.


DELIMITER $$
OPEN c1;
DROP PROCEDURE IF EXISTS `world`.`CURSOR_DEMO`$$ city_loop: LOOP
CREATE DEFINER=`tusa`@`%` PROCEDURE
`world`.`CURSOR_DEMO`(countryName VARCHAR(53)) SET s_code = '';
BEGIN FETCH c1 into i_ID,s_name, s_district, s_code;

IF s_code LIKE countryName THEN
DECLARE i_ID INT; INSERT INTO city_table (id, name,
DECLARE s_name VARCHAR(255); district,CountryName) VALUES ( i_ID,s_name,
DECLARE s_district VARCHAR(255); s_district,s_code);
DECLARE s_code VARCHAR(53); END IF;

DECLARE l_done INT DEFAULT 0; IF l_done=1 THEN
LEAVE city_loop;
DECLARE c1 CURSOR FOR END IF;
SELECT ID, city.Name, District,country.name FROM
city,country END LOOP city_loop;
where city.CountryCode = country.code and
country.name=countryName order by CountryCode,District; END$$

DECLARE CONTINUE HANDLER FOR NOT FOUND SET l_done=1; DELIMITER ;

DROP TABLE IF EXISTS city_table;
CREATE TABLE city_table (id INT, name VARCHAR(255),
district varchar(255), CountryName char(53))
ENGINE=memory;



tusa@localhost:3306 [world]> CALL CURSOR_DEMO('Italy');
+------+-----------------------+----------------------+------+
| id | name | district | code |
+------+-----------------------+----------------------+------+
| 1498 | Pescara | Abruzzit | ITA |
| 1479 | Taranto | Apulia | ITA |
| 1488 | Foggia | Apulia | ITA |
..
| 1503 | Trento | Trentino-Alto Adige | ITA |
| 1501 | Terni | Umbria | ITA |
| 1487 | Perugia | Umbria | ITA |
| 1474 | Venezia | Veneto | ITA |
| 1478 | Padova | Veneto | ITA |
| 1500 | Vicenza | Veneto | ITA |
| 1476 | Verona | Veneto | ITA |
+------+-----------------------+----------------------+------+
58 rows in set (0.05 sec)

Query OK, 0 rows affected, 1 warning (0.11 sec)



From previous example
• Cursor declaration
• Exception handling
• Non SELECT statement
• Block definition city_loop
• Query filtering by s_code
• Loop exit on condition (LEAVE)

Exercise
– Use all possible combination of LOOPS and Conditional control



Unbounded Result Sets

MySQL stored procedures (but not functions) can return result sets to the calling
program (though not, unfortunately, directly to another stored procedure).

A result set is returned from a stored procedure whenever a SQL statement that
returns a result set is not associated with either an INTO clause or a cursor.

We call these SQL statements unbounded.

Such SQL statements will usually be SELECT statements, although other statements
that return result sets SHOW, EXPLAIN, DESC, and so on can also be included
within the stored procedure.


DELIMITER $$
DROP PROCEDURE IF EXISTS `world`.`UNBOUND_DEMO` $$
CREATE DEFINER=`tusa`@`%` PROCEDURE `UNBOUND_DEMO`(districtName VARCHAR(35))
BEGIN
SELECT ID, Name, CountryCode,District,Population FROM City where District=districtName order by
Name;
Select ID, Name, CountryCode,District,Population FROM City
WHERE District=districtName AND Population = (Select MAX(Population) from city where
District=districtName );
END $$
DELIMITER ;
tusa@localhost:3306 [world]> CALL UNBOUND_DEMO('Lombardia');
+------+---------+-------------+-----------+------------+
| ID | Name | CountryCode | District | Population |
+------+---------+-------------+-----------+------------+
| 1497 | Bergamo | ITA | Lombardia | 117837 |
| 1480 | Brescia | ITA | Lombardia | 191317 |
| 1465 | Milano | ITA | Lombardia | 1300977 |
| 1496 | Monza | ITA | Lombardia | 119516 |
+------+---------+-------------+-----------+------------+

+------+--------+-------------+-----------+------------+
| ID | Name | CountryCode | District | Population |
+------+--------+-------------+-----------+------------+
| 1465 | Milano | ITA | Lombardia | 1300977 |
+------+--------+-------------+-----------+------------+

How to Handle an UNBOUNDED Result Set in a Java Application
public static void main(String[] args) {
try {
Class.forName("com.mysql.jdbc.Driver").newInstance();
Connection myConnection = DriverManager.getConnection(
"jdbc:mysql://138.132.167.150:3306/world1?user=grp1&password=grp1");
CallableStatement cs = myConnection.prepareCall(
"{CALL UNBOUND_DEMO(?)}");
cs.setString(1, "Lombardia");
boolean moreResultSets = cs.execute();
while (moreResultSets) {
ResultSet rs = cs.getResultSet();
ResultSetMetaData rsmd = rs.getMetaData();

StringBuffer buffer = new StringBuffer();
for (int i = 1; i <= rsmd.getColumnCount(); i++)
buffer.append(rsmd.getColumnName(i)).append("t");
System.out.println(buffer.toString());

while (rs.next()) {
buffer.setLength(0);
for (int i = 1; i <= rsmd.getColumnCount(); i++)
buffer.append(rs.getString(i)).append("t");
System.out.println(buffer.toString());
}
moreResultSets = cs.getMoreResults();
}
} catch (Exception ex) {
ex.printStackTrace();
Copyright 2008 Marco Tusa, MySQL
}
AB 41
}

Performing Dynamic SQL with Prepared Statements
MySQL supports a facility known as server-side prepared statements , which provides an API-
independent way of preparing a SQL statement for repeated execution efficiently and securely.

Prepared statements are interesting from a stored programming perspective because they allow
us to create dynamic SQL calls.

Now, the idea of prepared statements is to reduce the overhead of re-parsing (preparing) a SQL
statement for execution if all that has changed is a few data values, and to enhance security by
allowing SQL statement parameters to be supplied in a way that prevents SQL injection.

Stored procedures don't need prepared statements for these reasons, since the SQL
statements in stored procedures are already "prepared" for execution.

However, prepared statements come in handy in stored programs, because they allow you to
execute dynamic SQL from within a procedure (but not from within a trigger or function). A
SQL statement is dynamic if it is constructed at runtime (whereas a static SQL statement is one
that is constructed at the time of compilation of the program unit).

DELIMITER $$

DROP PROCEDURE IF EXISTS `world`.`DYNAMIC2_DEMO` $$
CREATE DEFINER=`tusa`@`%` PROCEDURE `DYNAMIC2_DEMO`(districtName VARCHAR(35))
BEGIN
DECLARE sql_string VARCHAR(400);
loop_city: BEGIN
CREATE TABLE city_table (id INT, name VARCHAR(255), district varchar(255), CountryName char(53))
ENGINE=memory;

SET sql_string = 'Insert INTO city_table (ID,Name, District, CountryName ) SELECT ID,Name,
District, CountryCode FROM City where ID= ?' ;
SET @SQL = sql_string;
PREPARE s1 FROM @SQL;
SET @x = 100;
REPEAT SET @x = @x + 1;
EXECUTE s1 USING @x;
UNTIL @x > 110
END REPEAT;
DEALLOCATE PREPARE s1;
SELECT * FROM city_table;
END loop_city;
END $$
DELIMITER ;



tusa@localhost:3306 [world]> call DYNAMIC2_DEMO('');
+------+-----------------------+---------------------+-------------+
| id | name | district | CountryName |
+------+-----------------------+---------------------+-------------+
| 101 | Godoy Cruz | Mendoza | ARG |
| 102 | Posadas | Misiones | ARG |
| 103 | GuaymallΘn | Mendoza | ARG |
| 104 | Santiago del Estero | Santiago del Estero | ARG |
| 105 | San Salvador de Jujuy | Jujuy | ARG |
| 106 | Hurlingham | Buenos Aires | ARG |
| 107 | NeuquΘn | NeuquΘn | ARG |
| 108 | Ituzaing≤ | Buenos Aires | ARG |
| 109 | San Fernando | Buenos Aires | ARG |
| 110 | Formosa | Formosa | ARG |
| 111 | Las Heras | Mendoza | ARG |
+------+-----------------------+---------------------+-------------+



When a stored program encounters an error condition, execution ceases
and an error is returned to the calling application.

That's the default behavior.

What if we need a different kind of behavior? What if, for example, we want
to trap that error, log it, or report on it, and then continue execution of our
application? For that kind of control, we need to define exception handlers in
our programs.

When developing MySQL stored programs, a very common scenario
fetching to the end of a result set also requires that we define an exception
handler.



Missing SQL:2003 Feature

There is no way to examine the current MySQL error code or SQLSTATE code. This means that
in an exception handler based on a generic condition such as SQLEXCEPTION, you have no
way of knowing what error just occurred.

You cannot raise an exception of your own to indicate an application-specific error or to re-signal
an exception after first catching the exception and examining its context.

MySQL use two set of ERROR code:
– SQLSTATE
– MySQL Error Code


Which Error code set should we use?
In theory, using the SQLSTATE codes will make your code more portable to other database
platforms and might therefore seem to be the best choice. However, there are a number of
reasons to use MySQL error codes rather than SQLSTATE codes when writing MySQL stored
programs:

In reality, it is unlikely that you will move your stored programs to another RDBMS. The Oracle
and SQL Server stored program languages are totally incompatible with MySQL. The DB2 stored
program language is somewhat compatible (both are based on the SQL:2003 standard). It is very
likely, however, that you will use MySQL-specific syntax as you write your application, which will
prevent your stored code from being portable.

Check http://dev.mysql.com/doc/refman/5.1/en/error-messages-server.html for full reference.


Errors Handlers
• Condition Handlers
– Exit
– Continue
DECLARE {CONTINUE | EXIT} HANDLER FOR {SQLSTATE sqlstate_code| MySQL error
code| condition_name} handler_actions

DECLARE EXIT HANDLER FOR 1062 /* MySQL ERROR code Duplicate key*/ SET duplicate_key=1;
INSERT INTO city (ID,name,CountryCode,Population)
VALUES(in_ID,in_name,in_CountryCode,in_Pop);
END;

IF duplicate_key=1 THEN
SELECT CONCAT('Failed to insert ',in_dept_name,
': duplicate key') as "Result";
END IF;



To improve the readability of the handlers is possible to define a condition declaration, which
associates a MySQL error code or SQLSTATE code with a meaningful name that can be used in
the handler declarations. The syntax for a condition declaration is:

DECLARE condition_name CONDITION FOR {SQLSTATE sqlstate_code | MySQL_error_code};

Once declared the condition name, it is possible to use it in the code instead of a MySQL error
code or SQLSTATE code. So instead of the following declaration:

DECLARE CONTINUE HANDLER FOR 1216 MySQL_statements;

Use the following more readable declaration:

DECLARE foreign_key_error CONDITION FOR 1216;

DECLARE CONTINUE HANDLER FOR foreign_key_error MySQL_statements;


DECLARE sqlcode INT DEFAULT 0;
DECLARE status_message VARCHAR(50);

DECLARE CONTINUE HANDLER FOR duplicate_key
BEGIN
SET sqlcode=1052;
SET status_message='Duplicate key error';
END;

DECLARE CONTINUE HANDLER FOR foreign_key_violated
BEGIN
SET sqlcode=1216;
SET status_message='Foreign key violated';
END;

DECLARE CONTINUE HANDLER FOR NOT FOUND
BEGIN
SET sqlcode=1329;
SET status_message='No record found';
END;


JAVA/JDBC Methods

• getErrorCode( ) Returns the MySQL-specific error code
• getSQLState( ) Returns the ANSI-standard SQLSTATE code
• getMessage( ) Returns the full text of the error message

Stored procedure error handling in Java/JDBC
try {
Class.forName("com.mysql.jdbc.Driver").newInstance( );

String ConnectionString="jdbc:mysql://" + hostname + "/" + database + "?user=" +
username + "&password=" + password;
System.out.println(ConnectionString);
Connection conn = DriverManager.getConnection(ConnectionString);
Statement stmt=conn.createStatement( );
stmt.execute("call error_test_proc(1)");
}
catch(SQLException SQLEx) {
System.out.println("MySQL error: "+SQLEx.getErrorCode( )+
" SQLSTATE:" +SQLEx.getSQLState( ));
System.out.println(SQLEx.getMessage( ));
}

Stored Program Construction
PART 2

• Creating and Maintaining Stored Programs
• Transaction Management
• MySQL Built-in Functions
• Stored Functions
• Triggers


Creating and Maintaining Stored Programs

There are three main ways you can edit your stored program code and submit it
to MySQL:

• Edit the stored program using a standard editor such as vi, Emacs, or
Notepad, and then use the MySQL command-line console to submit the
statements.

• Edit and create the stored program inside the MySQL Query Browser.

• Use a third-party graphical tool such as Quest Software's Toad for MySQL
to create the stored program



The CREATE PROCEDURE statement creates a stored procedure. The syntax for the statement is:
CREATE PROCEDURE procedure_name ([parameter[,...])
[LANGUAGE SQL]
[ [NOT] DETERMINISTIC
]
[ {CONTAINS SQL|MODIFIES SQL DATA|READS SQL DATA|NO SQL} ]
[SQL SECURITY {DEFINER|INVOKER} ]
[COMMENT comment_string]
procedure_statements

The parameter list consists of a comma-separated list of arguments that can be provided to the
stored procedure, each parameter is of the form:
[{IN|OUT|INOUT} ] parameter_name datatype

By default, parameters are of the IN type: this means that their values must be specified by the
calling program and that any modifications made to the parameter in the stored program cannot
be accessed from the calling program. OUT parameters, on the other hand, can be modified by
the stored program, and the modified values can be retrieved from the calling program.



The ALTER statement change the SQL SECURITY characteristic of a
stored procedure or stored function, or change the comment associated with
the procedure or function. This statement cannot currently be issued for
triggers. The syntax of this statement is shown below:

ALTER {PROCEDURE|FUNCTION} procedure_or_function_name
[SQL SECURITY {DEFINER|INVOKER}]
[COMMENT comment_string ]



The CREATE FUNCTION statement creates a stored function. This statement has a very
similar syntax to CREATE PROCEDURE:
CREATE FUNCTION function_name ([parameter[,...])
RETURNS datatype
[LANGUAGE SQL]
[ [NOT] DETERMINISTIC ]
[ { CONTAINS SQL|NO SQL|MODIFIES SQL DATA|READS SQL DATA} ]
[SQL SECURITY {DEFINER|INVOKER} ]
[COMMENT comment_string]
function_statements

• CREATE FUNCTION includes a mandatory RETURNS statement that specifies the
data type that will be returned from the function call.
• With CREATE FUNCTION, you cannot specify the IN, OUT, or INOUT modifiers to
parameters. All parameters are implicitly IN parameters.
• The function body must contain one or more RETURN statements, which terminate
function execution and return the specified result to the calling program.



The CREATE TRIGGER statement creates a trigger. Its syntax follows:
CREATE [DEFINER = { user|CURRENT_USER }] TRIGGER trigger_name
{BEFORE|AFTER}
{UPDATE|INSERT|DELETE}
ON table_name
FOR EACH ROW
trigger_statements
DEFINER
This optional clause specifies the security privileges that the trigger code will assume when it is
invoked. The default CURRENT_USER setting results in the trigger executing with the privileges
of the account that executes the CREATE TRIGGER statement. Specifying a user allows the
trigger to execute with the privileges of another account.
BEFORE or AFTER
These clauses control the sequence in which the trigger will fire either before or after the
triggering statement is executed.

UPDATE, INSERT, or DELETE
Action bounded to the trigger


Getting Information About Stored Programs
tusa@localhost:3306 [world]> SHOW Procedure status;
+-------+-------------------------------+-----------+---------++--------------------+
| Db | Name | Type | Definer || Database Collation |
+-------+-------------------------------+-----------+---------++--------------------+
| world | CURSOR_DEMO | PROCEDURE | tusa@% |dish_ci |
| world | DYNAMIC2_DEMO | PROCEDURE | tusa@% |dish_ci |
| world | DYNAMIC_DEMO | PROCEDURE | tusa@% |dish_ci |
| world | GetCountryNameContinentByCode | PROCEDURE | tusa@% |dish_ci |
| world | UNBOUND_DEMO | PROCEDURE | tusa@% |dish_ci |
+-------+-------------------------------+-----------+---------++--------------------+

tusa@localhost:3306 [world]> SHOW FUNCTION status;
+-------+------------------+----------+---------+----------------------+--------------------+
| Db | Name | Type | Definer | collation_connection | Database Collation |
+-------+------------------+----------+---------+----------------------+--------------------+
| world | FirstToUppercase | FUNCTION | tusa@% | utf8_general_ci | latin1_swedish_ci |
+-------+------------------+----------+---------+----------------------+--------------------+



Getting Information about Triggers
tusa@localhost:3306 [information_schema]> select * from triggersG
*************************** 1. row ***************************
TRIGGER_CATALOG: NULL
TRIGGER_SCHEMA: world
TRIGGER_NAME: nameInUppercase
EVENT_MANIPULATION: INSERT
EVENT_OBJECT_CATALOG: NULL
….
CREATED: NULL
SQL_MODE: STRICT_TRANS_TABLES,NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION
DEFINER: tusa@%
CHARACTER_SET_CLIENT: utf8
COLLATION_CONNECTION: utf8_general_ci
DATABASE_COLLATION: latin1_swedish_ci



Getting Procedure Information recap

• USE SHOW FOR:
– Procedure
– Function

• USE INFORMATION_SCHEMA FOR:
– ALL Including Body
• USE ROUTINES Table for the Procedures and Functions
• USE TRIGGERS Table for Triggers


Transaction Management

A transaction is a set of one or more SQL statements that are logically grouped
together and that must be either applied to the database in their entirety or not
applied at all.
Stored programs provide an excellent mechanism for defining, encapsulating, and
managing transactions.

Without the features available in stored programs, the calling program would need to
issue the relevant SQL statements for the transaction and provide the logic to control
locking and handle transaction failure. With MySQL stored program support , it is
possible to encapsulate the multiple, interdependent SQL statements of the
transaction into a single stored program. The application code, such as a PHP
program, calls the stored program and transfers the responsibility for transaction
management to the program executing in the database server.


Are you ACID?
Atomic
The transaction is indivisible either all the statements in the transaction are applied to the
database, or none are.

Consistent
The database remains in a consistent state before and after transaction execution.

Isolated
While multiple transactions can be executed by one or more users simultaneously, one
transaction should not see the effects of other concurrent transactions.

Durable
Once a transaction is saved to the database (an action referred to in database programming
circles as a commit), its changes are expected to persist. Even if the user turns off her computer
or the database server goes down, the changes will be saved. This usually means that the result
of the transaction must be written to a nonvolatile form of storage, such as a hard disk
(alternatively, it could be redundantly stored in multiple memory stores, written to battery-backed
memory, or written to solid state disk).


Transaction support in MySQL ENGINES

• MyISAM does not support transactions. Using a non transactional storage engine is
fine for certain applications in particular those that are overwhelmingly read-only.
InnoDB

• InnoDB is the most popular transaction-safe MySQL storage engine. It supports ACID
transactions as well as row-level locking and multi-version concurrency.

• Falcon in MySQL 6.X

• Any other ENGINE linked to MySQL core, that support transaction (ACID).


Isolation Level
The ANSI standard defines four isolation levels, all of which are supported by MySQL when using the InnoDB engine:

READ UNCOMMITTED
This is the lowest possible isolation level. Sometimes called dirty read, this level permits a transaction to read
rows that have not yet been committed.
READ COMMITTED
At this isolation level, only committed rows can be seen by a transaction. Furthermore, any changes committed
after a statement commences execution cannot be seen. For example, if you have a long-running SELECT
statement in session A that queries from the BOOKS table, and session B inserts a row into BOOKS while A's
query is still running, that new row will not be visible to the SELECT.
REPEATABLE READ
At this isolation level, no changes to the database that are made by other sessions since the transaction
commenced can be seen within the transaction, until the transaction is committed or rolled back (cancelled). This
means that if you re-execute a SELECT within your transaction, it will always show the same results (other than
any updates that occurred in the same transaction).
SERIALIZABLE
At this isolation level, every transaction is completely isolated so that transactions behave as if they had executed
serially, one after the other. In order to achieve this, the RDBMS will typically lock every row that is read, so other
sessions may not modify that data until the transaction is done with it. The locks are released when you commit or
cancel the transaction.


Transaction management statements
START TRANSACTION
Signifies the commencement of a new transaction. If an existing transaction is already in progress, then START
TRANSACTION will issue an implicit COMMIT . When you issue START TRANSACTION, the autocommit
property is effectively and implicitly set to 0 until the transaction ends.
COMMIT
Saves all changes made in the transaction to the database and then terminates a transaction. COMMIT also
releases any locks that might be in effect, whether they are explicit locks from FOR UPDATE or LOCK TABLES
or implicit locks acquired as a result of executing DML statements.
ROLLBACK
Undoes any changes to the database made by the transaction and then terminates that transaction. Like
COMMIT, ROLLBACK releases any locks held by the transaction.
SAVEPOINT savepoint_name
Creates a named savepoint identifier that can be the target of a ROLLBACK TO SAVEPOINT statement.
ROLLBACK TO SAVEPOINT savepoint_name
Performs a rollback on all statements that have been executed since the specified savepoint was created.
SET TRANSACTION
Allows you to change the isolation level of your transaction.
LOCK TABLES
Allows you to explicitly lock one or more tables. Note that LOCK TABLES implicitly closes any currently open
transactions . Explicitly commit or roll back your transaction before any LOCK TABLES statements.

First Transaction

CREATE PROCEDURE <Proc_NAME> (par1, par2,par3)
BEGIN
SET AUTOCOMMIT = 0;
<... Block ...>
COMMIT;
END;

BEGIN
START TRANSACTION;
<... Block ...>
COMMIT;
END;



Some statements will perform IMPLICIT COMMIT:
ALTER FUNCTION; DROP PROCEDURE;
ALTER PROCEDURE; DROP TABLE;
ALTER TABLE; BEGIN; UNLOCK TABLES;
CREATE DATABASE; LOAD MASTER DATA;
CREATE FUNCTION; LOCK TABLES;
CREATE INDEX; RENAME TABLE;
CREATE PROCEDURE; TRUNCATE TABLE;
CREATE TABLE; SET AUTOCOMMIT=1;
DROP DATABASE; START TRANSACTION.
DROP FUNCTION;
DROP INDEX;


Using SAVEPOINTS

Save points allow you to perform a partial rollback of the changes in your transaction.
If you issue an unqualified ROLLBACK, any and all changes in your current session
are erased. If, however, you place a SAVEPOINT statement in your program, then
you can roll back to that point in your program (and your transaction). In other words,
any changes made before that statement can still be saved to the database with a
COMMIT.

Generally, save points are intended to allow you to recover from a statement-level
error without having to abort and restart your transaction. In these circumstances, the
transaction includes one or more statements that might fail, yet should not force the
invalidation of the entire transaction. Usually you will want to roll back to a savepoint,
as part of handling the error, and then take the appropriate action, as indicated by the
particular error that was raised.



BEGIN
DECLARE txn_error INTEGER DEFAULT 0 ;

DECLARE CONTINUE HANDLER FOR SQLEXCEPTION BEGIN
SET txn_error=1;
END;

START TRANSACTION;

SAVEPOINT savepoint_xxx;
<... Block ...>
IF txn_error=1 THEN
ROLLBACK TO savepoint_xxx;
ELSE
<... Block ...>
END IF;

COMMIT;
END;


CREATE DEFINER=`tusa`@`%` PROCEDURE
Transaction Management IF s_code LIKE countryName THEN
`CURSOR_DEMO_TRANSACTION`(countryName VARCHAR(53)) INSERT INTO city_table (id, name,
BEGIN district,CountryName, transactionId) VALUES ( i_ID,s_name,
DECLARE i_ID INT; s_district,s_code,@transactionId);
DECLARE s_name VARCHAR(255); SET s_DELETE_ID = CONCAT_WS(',', s_DELETE_ID, i_ID);
DECLARE s_district VARCHAR(255); END IF;
DECLARE s_code VARCHAR(53);
DECLARE s_DELETE_ID VARCHAR(2000); IF l_done=1 THEN
DECLARE l_done INT DEFAULT 0; LEAVE city_loop;
DECLARE txn_error INT DEFAULT 0; END IF;
DECLARE c1 CURSOR FOR
SELECT ID, city.Name, District,country.name FROM END LOOP city_loop;
city,country CLOSE c1;
where city.CountryCode = country.code and Select * from city_table;
country.name=countryName order by CountryCode,District;

DECLARE CONTINUE HANDLER FOR NOT FOUND SET l_done=1;
DECLARE CONTINUE HANDLER FOR SQLEXCEPTION SET txn_error=1; SAVEPOINT delete_tmp_recset;
DELETE from city_table where transactionId =
DROP TABLE IF EXISTS city_table; @transactionId;
CREATE TABLE city_table (id INT, name VARCHAR(255),
district varchar(255), CountryName char(53), IF txn_error=1 THEN
transactionId INT) ROLLBACK TO delete_tmp_recset;
ENGINE=InnoDB; SELECT "Table not purge";
END IF;
START TRANSACTION;
SET @transactionId = FLOOR(1+ (RAND() * 50000000)); COMMIT;
OPEN c1;
SET s_DELETE_ID=''; END $$
city_loop: LOOP
SET s_code = '';
FETCH c1 into i_ID,s_name, s_district, s_code;


Transactions and Locks
The ACID properties of a transaction can only be implemented by restricting simultaneous
changes to the database. This is achieved by placing locks on modified data. These locks persist
until the transaction issues a COMMIT or ROLLBACK statement.

Situations in Which Locks Arise
When an UPDATE statement is executed, all rows modified will be locked.
An INSERT statement will cause any primary or unique key records to be locked. This will
prevent a concurrent insert of a statement with an identical primary key.

You can lock entire tables with the LOCK TABLES statement. This is not generally
recommended, because it not only reduces concurrency, it operates above the storage engine
layer, which might mean that any storage engine deadlock resolution mechanisms may be
ineffectual.

If you use the FOR UPDATE or LOCK IN SHARE MODE clauses in a SELECT statement, all of
the rows returned by that SELECT statement will be locked.


Dead Lock
Deadlocks can occur in any database system, but in row-level locking databases like
MySQL/InnoDB, the possibility of a deadlock is usually low. You can further reduce the frequency
of deadlocks by locking rows or tables in a consistent order, and by keeping your transactions as
short as possible.
CREATE PROCEDURE deadlocktrap(par,par2,par3)
IF deadlock=0 THEN
BEGIN
LEAVE tfer_loop;
DECLARE deadlock INT DEFAULT 0;
ELSE
DECLARE attempts INT DEFAULT 0;
SET attempts=attempts+1;
END IF;
tfer_loop:WHILE (attempts<3) DO
END WHILE tfer_loop;
BEGIN
IF deadlock=1 THEN
DECLARE deadlock_detected
CONDITION FOR 1213; SET out_status=-1;
DECLARE EXIT HANDLER FOR SET out_message="Failed with deadlock
deadlock_detected for 3 attempts";
BEGIN ELSE
ROLLBACK; SET out_status=0;
SET deadlock=1; SET out_message=CONCAT("OK
(",attempts," deadlocks)");
END;
END IF;
SET deadlock=0;
END;
START TRANSACTION;
< ... BLOCK ...>
COMMIT;
END;


Lock Timeouts
A deadlock is the most severe result of locking. Yet, in many other situations, a
program in one session may be unable to read or write a particular row, because it is
locked by another session. In this case, the program can and by default will wait for a
certain period of time for the lock to be released. It will then either acquire the lock or
time out. You can set the length of time a session will wait for an InnoDB lock to be
released by setting the value of the innodb_lock_wait_timeout configuration value,
which has a default of 50 seconds.

When a timeout occurs, MySQL/InnoDB will roll back the transaction and issue an
error code 1205, as shown in :

mysql> SELECT * FROM account_balance FOR UPDATE;
ERROR 1205 (HY000): Lock wait timeout exceeded; try restarting transaction


Optimistic and Pessimistic Locking Strategies
“If your transaction reads data that subsequently participates in an UPDATE, INSERT, or
DELETE, you need to take steps to ensure that the integrity of your transaction is not jeopardized
by the possibility of another transaction changing the relevant data between the time you read it
and the time you update it.”

The pessimistic locking strategy
Assume that concurrent updates are quite likely to occur, and write programs to prevent them from happening.
Generally, this means you will need to lock rows as they are read. Other transactions that want to update the
row must wait until the "pessimistic transaction" ends.

The optimistic locking strategy
Assume that it is unlikely that anyone will update a row between the time we view it and the time we update it.
Since we cannot be sure that this assumption is true, we must then, at the last possible moment, make sure
that the row has not been updated. If the row has been updated, the transaction cannot be trusted and will
have to be aborted.

Which is the good one?
STAY ON PESSIMISTIC!!!!



Some Advices on Transaction Design
A well-designed transaction should have the following properties:

• The integrity of the database will be maintained at all times.

• The duration and coverage of locks will be minimized. Locks should be applied to as
few rows as possible and maintained for the shortest possible duration.

• Rollbacks will be minimal transactions that eventually issue a rollback have
needlessly consumed resources.

• User expectations about the persistence of data will be met. For instance, a user who
clicks a Save or Apply button has a reasonable expectation that the data will not
disappear if he subsequently clicks Cancel on another page.
/.


Some Advices on Transaction Design (continue)
Keep transactions small
A transaction should generally include as small a logical unit of work as possible to reduce the duration of locks.
Avoid a transaction design that encourages rollbacks
For instance, rather than trying an insert and rolling back if there is a "duplicate key" error, check for the existence
of the key value before issuing the DML.
Avoid savepoints whenever possible
The existence of a savepoint may indicate that you have failed to check for success criteria before issuing a DML
statement and may indicate a transaction design that encourages rollbacks.
Lock rows that you SELECT if the results of the SELECT statement affect DML executed later in the
transaction. Pessimistic locking is easy to implement and is a robust solution. However, issue SELECTs with
FOR UPDATE as late in the transaction as possible to minimize duration of locks.
Consider optimistic locking for throughput-critical transactions
Optimistic locking requires more coding (to handle failed transactions) and may lead to user frustration if the
optimism is misplaced. However, optimistic locking can reduce lock duration and thereby increase throughput for
high-volume transactions.
Explicitly commence transactions and avoid leaving transactions "dangling"
Stored programs that issue transactional statements should generally take responsibility for commencing and
terminating the transaction, rather than assuming that some external program is going to handle a COMMIT or
ROLLBACK.


While these are reasonable guidelines, there are sometimes trade-offs that you will need to consider:

Unlike any other MySQL statement, the COMMIT statement always requires a physical write to
disk to complete. Therefore, plan it carfully.

Checking all possible success criteria before issuing a DML statement might be overly expensive
in some cases. It might be preferable to let a DML statement fail and then roll back to a savepoint
under certain circumstances.

The trade-offs for the optimistic and pessimistic locking strategies are heavily dependent on the
characteristics of your application.

Modular design considerations may sometimes lead you to write a stored program in such a way
that the control of the overall transaction is delegated to a higher-level program.


Using MySQL Stored Programs with Java
How to connect using JDBC
Class.forName("com.mysql.jdbc.Driver").newInstance( );
Connection myConnection = DriverManager.getConnection( "jdbc:mysql://localhost:3306/test?
user=root&password=secret");

Issuing a non SELECT statement
Statement stmt1 = myConnection.createStatement( );
stmt1.executeUpdate("set autocommit=0");

Issuing a SELECT statement
Statement stmt2 = myConnection.createStatement( );
ResultSet results = stmt2.executeQuery("SELECT id, name " +
" FROM city");
while(results.next( ))
{
int ID = results.getInt(“id"); // Get column by name
String Name = results.getString(2); // Got column by number
System.out.println(ID + ":" + Name);
}
results.close( );


Catching a SQLException
static public void catchingException(Connection connection)
{
try
{
Statement s1 = connection.createStatement( );
s1.executeUpdate("CREATE TABLE DEMO" +
" (MyInt INT," +
" MyString VARCHAR(30))");
}
catch(SQLException exception)
{
System.out.println("Error while creating demo tables: " +
exception.getErrorCode( ) +
" SQLSTATE:" + exception.getSQLState( ));
exception.printStackTrace( );
}
}


Using Stored Programs in JDBC

Stored program calls are very similar to standard JDBC calls. A stored program strongly
resembles a prepared statement that executes a query, with the following exceptions:

• A stored program can return more than one result set.

• A stored procedure can be associated with output as well as input parameters. This means that
we need a way to retrieve the altered values from any stored procedure parameters that are
defined as OUT or INOUT.


A quick example
DROP PROCEDURE IF EXISTS `world`.`DYNAMIC_DEMO_RETURN` $$
CREATE DEFINER=`tusa`@`%` PROCEDURE `DYNAMIC_DEMO_RETURN`(districtName VARCHAR(35), OUT maxPop INT )
BEGIN
SELECT CountryCode, District,SUM(Population) AS POP FROM City
GROUP BY District, CountryCode ORDER BY POP DESC LIMIT 1;

SELECT * FROM City WHERE District = districtName ORDER BY POPULATION DESC;

SELECT MAX(POPULATION) INTO maxPop FROM City WHERE District = districtName ORDER BY POPULATION DESC;

END $$

/ JAVA /
CallableStatement callableStmt =myConnection.prepareCall("{CALL sp_test_inout_rs2(?,?)}");
callableStmt.registerOutParameter(2,Types.INTEGER);
callableStmt.setString(1, ‘Liguria’);
callableStmt.execute( );

/.

A quick example (continue)
private void executeProcedure(Connection connection, String sqlText)
throws SQLException {
CallableStatement cs = connection.prepareCall("{CALL " + sqlText + "}");
boolean moreResultSets = cs.execute( );
while (moreResultSets) {
ResultSet rs = cs.getResultSet( );
ResultSetMetaData rsmd = rs.getMetaData( );

StringBuffer buffer = new StringBuffer( );
for (int i = 1; i <= rsmd.getColumnCount( ); i++)
buffer.append(rsmd.getColumnName(i)).append("t");
System.out.println(buffer.toString( ));
while (rs.next( )) {
buffer.setLength(0);
for (int i = 1; i <= rsmd.getColumnCount( ); i++)
buffer.append(rs.getString(i)).append("t");
System.out.println(buffer.toString( ));
}
moreResultSets = cs.getMoreResults( );
}
}
To retrieve the OUT PARAMETER System.out.println("Out parameter = " + callableStmt.getInt(2));


Optimizing Stored Programs
Permissions Required for Stored Programs

• CREATE ROUTINE
– Allows a user to create new stored programs.

• ALTER ROUTINE
– Allows a user to alter the security mode, SQL mode, or comment for an existing
stored program.

• EXECUTE
– Allows a user to execute a stored procedure or function.

GRANT CREATE ROUTINE ON world.* TO sp_creator;



Stored program code differs from any other kind of code that might execute
against the database in that it can have database privileges that are
different from those of the account that executes the stored program.

Normally, when we execute some SQL whether it is inside the MySQL
client, a PHP program, or whatever the activities that the SQL will perform
(read table X, update table Y, etc.) will be checked against the privileges
that are associated with the database account to which we are connected.

If our account lacks privilege to perform the activity, the SQL statement will
fail with the appropriate error.


Stored programs can be defined to act in the two ways.

• SQL SECURITY INVOKER clause is included in the CREATE
PROCEDURE or CREATE FUNCTION statement used to create the
program.

• SQL SECURITY DEFINER (the default) is specified instead, then the
stored program executes with the privilege of the account that created the
stored program, rather than the account that is executing the stored
program.

Known as definer rights, this execution mode can be a very powerful way
of restricting ad hoc table modifications and avoiding security breaches.


DEFAULT Create Procedure:
CREATE DEFINER=`tusa`@`%` PROCEDURE `world`.`DYNAMIC_DEMO`(districtName
VARCHAR(35))
BEGIN

DEFINER
VARCHAR(35)) SQL SECURITY DEFINER
BEGIN
...

INVOKER
VARCHAR(35)) SQL SECURITY INVOKER
BEGIN


Tuning Optimization

TO DO


MySQL stored programs can often add to application functionality and
developer efficiency, and there are certainly many cases where the use of a
procedural language such as the MySQL stored program language can do
things that a non procedural language like SQL cannot.
There are also a number of reasons why a MySQL stored program approach
may offer performance improvements over a traditional SQL approach:

• It provides a procedural approach (SQL is a declarative, non procedural
language)
• It reduces client-server traffic
• It allows us to divide and conquer complex statements


How FAST are our Store Procedure?
Ok … We are doing our best but still a long way in front of us:


If you want to do the tests by yourself: CREATE PROCEDURE sp_nprimes(p_num int)
PROCEDURE N_PRIMES BEGIN
( p_num NUMBER) DECLARE i INT;
IS DECLARE j INT;
i INT; DECLARE nprimes INT;
j INT; DECLARE isprime INT;
nprimes INT;
isprime INT; SET i=2;
BEGIN SET nprimes=0;
i:=2;
nprimes:=0; main_loop:
<<main_loop>> WHILE (i<p_num) do
WHILE (i<p_num) LOOP SET isprime=1;
isprime:=1; SET j=2;
j:=2; divisor_loop:
<<divisor_loop>> WHILE (j<i) DO
WHILE (j<i) LOOP IF (MOD(i,j)=0) THEN
IF (MOD(i,j)=0) THEN SET isprime=0;
isprime:=0; LEAVE divisor_loop;
EXIT divisor_loop; END IF;
END IF; SET j=j+1;
j:=j+1; END WHILE ;
END LOOP ; IF (isprime=1) THEN
IF (isprime=1) THEN SET nprimes=nprimes+1;
nprimes:=nprimes+1; END IF;
END IF; SET i=i+1;
i:=i+1; END WHILE;
Copyright 2008 Marco Tusa, MySQL AB
END LOOP; SELECT CONCAT(nprimes,' prime numbers less than 91
dbms_output.put_line(nprimes||' prime numbers less ',p_num);
than '||p_num); END;

Stored Programs as an Alternative to Expensive SQL

Sometimes we can use a stored program to perform query or DML
operations that perform badly in standard SQL.

This usually happens when the "pure" SQL statement becomes overly
complex because of limitations in the SQL syntax or when the MySQL
optimizer isn't able to come up with a sensible plan for your SQL query.


IF and CASE Statements

When constructing IF and CASE statements, try to minimize the number of
comparisons that these statements are likely to make by testing for the most
likely scenarios first.

For instance, in the code in the next slide, the first statement maintains
counts of various percentages.

Assuming that the input data is evenly distributed, the first IF condition
(percentage>95) will match about once in every 20 executions.
On the other hand, the final condition will match in three out of four
executions. So this means that for 75% of the cases, all four comparisons
will need to be evaluated.


Non Optimized
IF (percentage>95) THEN
SET Above95=Above95+1;
ELSEIF (percentage >=90) THEN
SET Range90to95=Range90to95+1;
ELSEIF (percentage >=75) THEN
ELSE
SET LessThan75=LessThan75+1;
END IF;
Optimized
IF (percentage<75) THEN
SET LessThan75=LessThan75+1;
ELSEIF (percentage >=75 AND percentage<90) THEN
ELSEIF (percentage >=90 and percentage <=95) THEN
ELSE
SET Above95=Above95+1;
END IF;

Looks simple and the effect is relevant:


Using Recursion

A recursive routine is one that invokes itself.

Recursive routines often offer elegant solutions to complex programming
problems, but they also tend to consume large amounts of memory.

They are also likely to be less efficient and less scalable than
implementations based on iterative execution.


Recursive Not Recursive
CREATE PROCEDURE nonrec_fib(n INT,OUT
CREATE PROCEDURE rec_fib(n INT,OUT out_fib out_fib INT)
INT) BEGIN
BEGIN DECLARE m INT default 0;
DECLARE n_1 INT; DECLARE k INT DEFAULT 1;
DECLARE n_2 INT; DECLARE i INT;
DECLARE tmp INT;
IF (n=0) THEN
SET out_fib=0; SET m=0;
ELSEIF (n=1) then SET k=1;
SET out_fib=1; SET i=1;
ELSE
CALL rec_fib(n-1,n_1); WHILE (i<=n) DO
CALL rec_fib(n-2,n_2); SET tmp=m+k;
SET out_fib=(n_1 + n_2); SET m=k;
END IF; SET k=tmp;
END SET i=i+1;
END WHILE;
SET out_fib=m;
Copyright 2008 Marco Tusa, MySQL AB END 97

The difference is quite impressive and evident


When you need to retrieve only a single row from a SELECT statement,
using the INTO clause is far easier than declaring, opening, fetching from,
and closing a cursor. But does the INTO clause generate some additional
work for MySQL or could the INTO clause be more efficient than a cursor?


Trigger Overhead
Every database trigger is associated with a specific DML operation
(INSERT, UPDATE, or DELETE) on a specific table the trigger code will
execute whenever that DML operation occurs on that table.

Furthermore, all MySQL 5.x triggers are of the FOR EACH ROW type,
which means that the trigger code will execute once for each row affected by
the DML operation.

Given that a single DML operation might potentially affect thousands of
rows, should we be concerned that our triggers might have a negative effect
on DML performance?

Absolutely!
100

When using Trigger be ALWAYS sure to have the right indexes.

101


Questions
&
Answers
102

Exercises
Warming UP
List all regions using English as non primary language;
From this list enumerate all cities with a population between 10.000 and
500.000.
From the first list identify the city with the largest population and the one with
the smaller.

From the list cities with largest population, identify which one is the one with
bigger Life Expectancy

Don’t use internal functions (like MAX() MIN() AVG()), ONLY one join is allowed
per STORE PROCEDURE.

103

Movie Shop
Implementare:
GRP1 sistema per inserire/modificare/cancellare Film, caricando/gestendo I dati testuali su tabella
FILM_TEXT.
Gestire in connessione Attori, Categoria, Lingua, Paese.
Report per Paese, Lingua, Attori e Categoria

GRP3 sistema per inserire/modificare/cancellare Clienti, Indirizzi.
Report per cliente costo totale, per cateoria, per lingua, per Mese, tempo di noleggio totale,
ammontare per mese.

GRP4 sistema per gestire i noleggi inserire/modificare/cancellare.
Utilizzazione/report per cliente, categoria, paese, costo noleggio per film, identificare il film piu
noleggiato, il film che ha portato maggiore incasso per anno e per mese.

GRP2 sistema per inserire/modificare/cancellare le tabelle di referenza, (citta, paese, categorie,
lingue) report per ogni referenza con valori totali/annuali/mensili.

104

Stored

105

References
http://dev.mysql.com/doc/refman/5.1/en/stored-procedures.html

http://downloads.mysql.com/docs/world.sql.gz
http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference.html

by Guy Harrison (Author), Steven Feuerstein (Author)

106

MySQL developing Store Procedure

More Related Content

What's hot (20)

Similar to MySQL developing Store Procedure (20)

More from Marco Tusa (20)

Recently uploaded (20)

MySQL developing Store Procedure