3 Key Programming Considerations

This chapter discusses key issues to consider before developing and running your SQLJ application, and also provides a summary and sample applications. The following topics are discussed:

Selection of the JDBC Driver

You must consider which Java Database Connectivity (JDBC) driver will be appropriate for your situation and whether it may be advantageous to use different drivers for translation and run time. You must choose or register the appropriate driver class for each and then specify the driver in your connection URL.

Note:

Your application will require an Oracle JDBC driver if you use Oracle-specific code generation or if you use ISO code generation with Oracle customizer, even if your code does not actually use Oracle-specific features.

This section covers the following topics:

Overview of Oracle JDBC Drivers

Oracle provides the following JDBC drivers:

  • Oracle Call Interface (OCI) driver: For client-side use with an Oracle client installation.

  • Thin driver: A pure Java driver for client-side use, particularly with applets. It does not require an Oracle client installation.

  • Server-side Thin driver: Is functionally the same as the client-side Thin driver, but is for code that runs inside Oracle Database instance and needs to access a remote server.

  • Server-side internal driver: For code that runs inside the target server, that is, inside Oracle Database instance that it must access.

Oracle Database 11g provides client-side drivers compatible with Java Development Kit (JDK) 1.5.

Note:

Remember that your choices may differ between translation time and run time. For example, you may want to use Oracle JDBC OCI driver at translation time for semantics-checking, but Oracle JDBC Thin driver at run time.

Core JDBC Functionality

The core functionality of all Oracle JDBC drivers is the same. They support the same feature set, syntax, programming interfaces, and Oracle extensions.

All Oracle JDBC drivers are supported by the oracle.jdbc.OracleDriver class.

JDBC OCI Driver

Oracle JDBC OCI driver accesses the database by calling the OCI directly from Java, providing the highest compatibility with the different Oracle Database versions. These drivers support installed Oracle Net adapters, including interprocess communication (IPC), named pipes, TCP/IP, and IPX/SPX.

The use of native methods to call C entry points makes the OCI driver dependent on the Oracle platform, requiring an Oracle client installation that includes Oracle Net. Therefore it is not suitable for applets.

Connection strings for the OCI driver are of the following form, where tns is an optional TNS alias or full TNS specification:

jdbc:oracle:oci:@<tns>

Note:

For backward compatibility, oci8 is still acceptable instead of oci.

JDBC Thin Driver

Oracle JDBC Thin driver is a platform-independent, pure Java implementation that uses Java sockets to connect directly to Oracle Database from any Oracle or non-Oracle client. It can be downloaded into a browser simultaneously with the Java applet being run.

The JDBC Thin driver supports only TCP/IP protocol and requires a TNS listener to be listening on TCP/IP sockets from the database server. When the JDBC Thin driver is used with an applet, the client browser must have the capability to support Java sockets.

Connection strings for the JDBC Thin driver are typically of the following form:

jdbc:oracle:thin:@host:port/servicename

See Also:

Oracle Database JDBC Developer's Guide and Reference for information about database service names

In Oracle Database 11g, connection strings using SIDs are deprecated, but are still supported for backward compatibility:

jdbc:oracle:thin:@host:port:sid

JDBC Server-Side Thin Driver

Oracle JDBC server-side Thin driver offers the same functionality as the client-side JDBC Thin driver, but runs inside the database and accesses a remote server. This is useful in accessing one Oracle Database instance from inside another, such as from a Java stored procedure.

Connection strings for the server-side Thin driver are the same as for the client-side Thin driver.

Note:

In order to leave the originating database when using the server-side Thin driver, the user account must have SocketPermission assigned. Refer to the Oracle Database JDBC Developer's Guide and Reference for more information. Also, refer to the Oracle Database Java Developer's Guide for general information about SocketPermission and other permissions.

JDBC Server-Side Internal Driver

Oracle JDBC server-side internal driver provides support for any Java code that runs inside the target Oracle Database instance where the SQL operations are to be performed. The server-side internal driver enables Oracle Java virtual machine (JVM) to communicate directly with the SQL engine. This driver is the default JDBC driver for SQLJ code running as a stored procedure, stored function, or trigger in Oracle Database 11g.

Connection strings for the server-side internal driver are of the following form:

jdbc:oracle:kprb:

If your SQLJ code uses the default connection context, then SQLJ automatically uses this driver for code running in Oracle JVM.

Driver Selection for Translation

Use SQLJ option settings, either on the command line or in a properties file, to choose the driver manager class and specify a driver for translation.

Use the SQLJ -driver option to choose any driver manager class other than OracleDriver, which is the default.

Specify the particular JDBC driver to choose, such as JDBC Thin or JDBC OCI for Oracle Database, as part of the connection URL you specify in the SQLJ -url option.

You will typically, but not necessarily, use the same driver that you use in your source code for the run time connection.

Note:

Remember that the -driver option does not choose a particular driver. It registers a driver class with the driver manager. One driver class might be used for multiple driver protocols, such as OracleDriver, which is used for all of Oracle JDBC protocols.

Driver Selection and Registration for Run Time

To connect to the database at run time, you must register one or more drivers that will understand the URLs you specify for any of your connection instances, whether they are instances of the sqlj.runtime.ref.DefaultContext class or of any connection context classes that you declare.

If you are using an Oracle JDBC driver and create a default connection using the Oracle.connect() method, then SQLJ handles this automatically. The Oracle.connect() method registers the oracle.jdbc.OracleDriver class.

If you are using an Oracle JDBC driver, but do not use Oracle.connect(), then you must manually register the OracleDriver class, as follows:

DriverManager.registerDriver(new oracle.jdbc.OracleDriver());

If you are not using an Oracle JDBC driver, then you must register some appropriate driver class, as follows:

DriverManager.registerDriver(new mydriver.jdbc.driver.MyDriver());

In any case, you must also set your connection URL, user name, and password.

Note:

As an alternative to using the JDBC driver manager in establishing JDBC connections, you can use data sources. You can specify a data source in a with clause, as described in "Declaration WITH Clause". For general information about data sources, refer to the Oracle Database JDBC Developer's Guide and Reference.

Connection Considerations

When deciding what database connection or connections you will need for your SQLJ application, consider the following:

  • Will you need just one database connection or multiple connections?

  • If using multiple connections (possibly to multiple schemas), then will each connection use SQL entities of the same name: tables of the same name, columns of the same name and data types, stored procedures of the same name and signature, and so on?

  • Will you need different connections for translation and run time or will the same suffice for both?

A SQLJ executable statement can specify a particular connection context instance, either of DefaultContext or of a declared connection context class, for its database connection. Alternatively, it can omit the connection context specification and use the default connection, which is an instance of DefaultContext that was previously set as the default.

Note:

If your operations will use different sets of SQL entities, then you will typically want to declare and use additional connection context classes.

This section covers the following topics:

Single Connection or Multiple Connections Using DefaultContext

This section discusses scenarios where you will use connection instances of only the DefaultContext class.

This is typical if you are using a single connection, or multiple connections that use SQL entities with the same names and data types.

Single Connection

For a single connection, use one instance of the DefaultContext class specifying the database URL, user name, and password, when you construct your DefaultContext object.

You can use the connect() method of the oracle.sqlj.runtime.Oracle class to accomplish this. Calling this method automatically initializes the default connection context instance. This method has several signatures, including ones that allow you to specify user name, password, and URL, either directly or using a properties file. In the following example, the properties file connect.properties is used:

Oracle.connect(MyClass.class, "connect.properties");

Note:

The connect.properties file is searched for relative to the specified class. In the example, if MyClass is located in my-package, then connect.properties must be found in the same package location, my-package.

If you use connect.properties, then you must edit it appropriately and package it with your application. In this example, you must also import the oracle.sqlj.runtime.Oracle class.

Alternatively, you can specify user name, password, and URL directly:

Oracle.connect("jdbc:oracle:thin:@localhost:1521/myservice", "scott", "tiger");

In this example, the connection will use the JDBC Thin driver to connect the scott user with the password, tiger, to a database on the computer, localhost, through port 1521, where myservice is the name of the database service for the connection.

Either of these examples creates a special static instance of the DefaultContext class and installs it as your default connection. It is not necessary to do anything with this DefaultContext instance directly.

Once you have completed these steps, you do not need to specify the connection for any of the SQLJ executable statements in your application, if you want them all to use the default connection.

Note that in using a JDBC Thin driver, the URL must include the host name, port number, and service name (or SID, which is deprecated in Oracle Database 11g), as in the preceding example. Also, the database must have a listener running at the specified port. In using the JDBC OCI driver, no service name (or SID) is required if you intend to use the default account of the client, as will be the case in examples in this document. Alternatively, you can use name-value pairs.

The following URL will connect to the default account of the client:

jdbc:oracle:oci:@

Note:

  • Oracle.connect() will not set your default connection if one had already been set. In that case, it returns null. This enables you to use the same code on a client or in the server. If you do want to override your default connection, then use the static setDefaultContext() method of DefaultContext.

  • The Oracle.connect() method defaults to a false setting of the auto-commit flag. However, it also has signatures to set it explicitly. In the Oracle JDBC implementation, the auto-commit flag defaults to true.

  • You can optionally specify getClass() instead of MyClass.class in the Oracle.connect() call, as long as you are not calling getClass() from a static method. The getClass() method is used in some of the SQLJ demo applications.

  • You can access the static DefaultContext instance, which corresponds to your default connection, as follows:

    DefaultContext.getDefaultContext();
    
    

Multiple Connections

For multiple connections, you can create and use additional instances of the DefaultContext class, while optionally still using the default connection.

You can use the Oracle.getConnection() method to instantiate DefaultContext, as in the following examples.

First, consider a case where you want most statements to use the default connection, but other statements to use a different connection. You must create one additional instance of DefaultContext:

DefaultContext ctx = Oracle.getConnection (
   "jdbc:oracle:thin:@localhost2:1521/myservice2", "bill", "lion");

Note:

ctx could also use the scott/tiger schema, if you want to perform multiple sets of operations on the same schema.

When you want to use the default connection, it is not necessary to specify a connection context:

#sql { SQL operation };

This is actually a shortcut for the following:

#sql [DefaultContext.getDefaultContext()] { SQL operation };

When you want to use the additional connection, specify ctx as the connection:

#sql [ctx] { SQL operation };

Next, consider situations where you want to use multiple connections, where each of them is a named DefaultContext instance. This enables you to switch your connection back and forth.

The following statements establish multiple connections to the same schema (in case you want to use multiple Oracle Database sessions or transactions, for example). Instantiate the DefaultContext class for each connection you will need:

DefaultContext ctx1 = Oracle.getConnection
   ("jdbc:oracle:thin:@localhost1:1521/myservice1", "scott", "tiger");
DefaultContext ctx2 = Oracle.getConnection
   ("jdbc:oracle:thin:@localhost1:1521/myservice1", "scott", "tiger");

This creates two connection context instances that would use the same schema, connecting to scott/tiger using service myservice1 on the computer localhost1, using Oracle JDBC Thin driver.

Now, consider a case where you would want multiple connections to different schemas. Again, instantiate the DefaultContext class for each connection you will need:

DefaultContext ctx1 = Oracle.getConnection
   ("jdbc:oracle:thin:@localhost1:1521/myservice1", "scott", "tiger");
DefaultContext ctx2 = Oracle.getConnection
   ("jdbc:oracle:thin:@localhost2:1521/myservice2", "bill", "lion");

This creates two connection context instances that use Oracle JDBC Thin driver but use different schemas. The ctx1 object connects to scott/tiger using service myservice1 on the computer localhost1, while the ctx2 object connects to bill/lion using service myservice2 on the computer localhost2.

There are two ways to switch back and forth between these connections for the SQLJ executable statements in your application:

  • If you switch back and forth frequently, then you can specify the connection for each statement in your application:

    #sql [ctx1] { SQL operation };
    ...
    #sql [ctx2] { SQL operation };
    
    

    Note:

    Include the square brackets around the connection context instance name; they are part of the syntax.
  • If you use either of the connections several times in a row within your code flow, then you can periodically use the static setDefaultContext() method of the DefaultContext class to reset the default connection. This method initializes the default connection context instance. This way, you can avoid specifying connections in your SQLJ statements.

    DefaultContext.setDefaultContext(ctx1);
    #sql { SQL operation };   // These three statements all use ctx1
    #sql { SQL operation };
    #sql { SQL operation };
    ...
    DefaultContext.setDefaultContext(ctx2);
    #sql { SQL operation };   // These three statements all use ctx2
    #sql { SQL operation };
    #sql { SQL operation };
    
    

    Note:

    Because the preceding statements do not specify connection contexts, at translation time they will all be checked against the default connection context.

Closing Connections

It is advisable to close your connection context instances when you are done, preferably in a finally clause of a try block (in case your application terminates with an exception).

The DefaultContext class, as well as any connection context classes that you declare, includes a close() method. Calling this method closes the SQLJ connection context instance and, by default, also closes the underlying JDBC connection instance and the physical connection.

In addition, the oracle.sqlj.runtime.Oracle class has a static close() method to close the default connection only. In the following example, presume ctx is an instance of any connection context class:

...
finally
{
   ctx.close();
}
...

Alternatively, if the finally clause is not within a try block in case a SQL exception is encountered:

...
finally
{
   try { ctx.close(); } catch(SQLException ex) {...}
}
...

Or, to close the default connection, the Oracle class also provides a close() method:

...
finally
{
   Oracle.close();
}
...

Always commit or roll back any pending changes before closing the connection. Whether there would be an implicit COMMIT operation as the connection is closed is not specified in the JDBC standard and may vary from vendor to vendor. For Oracle, there is an implicit COMMIT when a connection is closed, and an implicit ROLLBACK when a connection is garbage-collected without being closed, but it is not advisable to rely on these mechanisms.

Note:

It is also possible to close a connection context instance without closing the underlying connection (in case the underlying connection is shared). Refer to "Closing Shared Connections" for more information.

Multiple Connections Using Declared Connection Context Classes

For multiple connections that use different sets of SQL entities, it is advantageous to use connection context declarations to define additional connection context classes. Having a separate connection context class for each set of SQL entities that you use enables SQLJ to do more rigorous semantics-checking of your code.

More About the Oracle Class

The Oracle SQLJ implementation provides the oracle.sqlj.runtime.Oracle class to simplify the process of creating and using instances of the DefaultContext class.

The static connect() method initializes the default connection context instance, instantiating a DefaultContext object and installing it as your default connection. You do not need to assign or use the DefaultContext instance returned by connect(). If you had already established a default connection, then connect() returns null.

The static getConnection() method simply instantiates a DefaultContext object and returns it. You can use the returned instance as desired.

Both methods register Oracle JDBC driver manager automatically if the oracle.jdbc.OracleDriver class is found in the CLASSPATH. The static close() method closes the default connection.

Signatures of the Oracle.connect() and Oracle.getConnection() Methods

Both the method have signatures that take the following parameter sets as input:

  • URL (String), user name (String), password (String)

  • URL (String), user name (String), password (String), auto-commit flag (boolean)

  • URL (String), java.util.Properties object containing properties for the connection

  • URL (String), java.util.Properties object, auto-commit flag (boolean)

  • URL (String) fully specifying the connection, including user name and password

    The following is an example of the format of a URL string specifying user name (scott) and password (tiger) when using Oracle JDBC drivers, in this case the JDBC Thin driver:

    "jdbc:oracle:thin:scott/tiger@localhost:1521/myservice"
    
    
  • URL (String), auto-commit flag (boolean)

  • A java.lang.Class object for the class relative to which the properties file is loaded, name of properties file (String)

  • A java.lang.Class object, name of properties file (String), auto-commit flag (boolean)

  • A java.lang.Class object, name of properties file (String), user name (String), password (String)

  • A java.lang.Class object, name of properties file (String), user name (String), password (String), auto-commit flag (boolean)

  • JDBC connection object (Connection)

  • SQLJ connection context object

These last two signatures inherit an existing database connection. When you inherit a connection, you will also inherit the auto-commit setting of that connection.

The auto-commit flag specifies whether SQL operations are automatically committed. For the Oracle.connect() and Oracle.getConnection() methods only, the default is false. If that is the setting you want, then you can use one of the signatures that does not take auto-commit as input. However, anytime you use a constructor to create an instance of a connection context class, including DefaultContext, you must specify the auto-commit setting. In the Oracle JDBC implementation, the default for the auto-commit flag is true.

Optional Oracle.close() Method Parameters

In using the Oracle.close() method to close the default connection, you have the option of specifying whether or not to close the underlying physical database connection. By default it is closed. This is relevant if you are sharing this physical connection between multiple connection objects, either SQLJ connection context instances or JDBC connection instances.

You can keep the underlying physical connection open as follows:

Oracle.close(ConnectionContext.KEEP_CONNECTION);

You can close the underlying physical connection (default behavior) as follows:

Oracle.close(ConnectionContext.CLOSE_CONNECTION);

KEEP_CONNECTION and CLOSE_CONNECTION are static constants of the ConnectionContext interface.

More About the DefaultContext Class

The sqlj.runtime.ref.DefaultContext class provides a complete default implementation of a connection context class. As with classes created using a connection context declaration, the DefaultContext class implements the sqlj.runtime.ConnectionContext interface. The DefaultContext class has the same class definition that would have been generated by the SQLJ translator from the declaration:

#sql public context DefaultContext;

DefaultContext Methods

The following are the key methods of the DefaultContext class:

  • getConnection()

    Gets the underlying JDBC connection object. This is useful if you want to have JDBC code in your application, which is one way to use dynamic SQL operations. You can also use the setAutoCommit() method of the underlying JDBC connection object to set the auto-commit flag for the connection.

  • setDefaultContext()

    Sets the default connection your application uses. This is a static method and takes a DefaultContext instance as input. SQLJ executable statements that do not specify a connection context instance will use the default connection that you define using this method or the Oracle.connect() method.

  • getDefaultContext()

    Returns the DefaultContext instance currently defined as the default connection for your application. This is a static method.

  • close()

    Closes the connection context instance.

The getConnection() and close() methods are specified in the sqlj.runtime.ConnectionContext interface.

Note:

On a client, getDefaultContext() returns null if setDefaultContext() was not previously called. However, if a data source object has been bound under "jdbc/defaultDataSource" in JNDI, then the client will use this data source object as its default connection.

In the server, getDefaultContext() returns the default connection, which is the connection to the server itself.

DefaultContext Constructors

It is typical to instantiate DefaultContext using the Oracle.connect() or Oracle.getConnection() method. However, if you want to create an instance directly, then there are five constructors for DefaultContext. The different input parameter sets for these constructors are:

  • URL (String), user name (String), password (String), auto-commit (boolean)

  • URL (String), java.util.Properties object, auto-commit (boolean)

  • URL (String fully specifying connection and including user name and password), auto-commit setting (boolean)

    The following is an example of the format of a URL specifying user name and password when using Oracle JDBC drivers, in this case the JDBC Thin driver:

    "jdbc:oracle:thin:scott/tiger@localhost:1521/myservice"
    
    
  • JDBC connection object (Connection)

  • SQLJ connection context object

The last two signatures inherit an existing database connection. When you inherit a connection, you will also inherit the auto-commit setting of that connection.

Following is an example of constructing a DefaultContext instance:

DefaultContext defctx = new DefaultContext
   ("jdbc:oracle:thin:@localhost:1521/myservice", "scott", "tiger", false);

Notes About Connection Context Constructors:

Note:

You must keep the following in mind when using connection context constructors:
  • It is important to note that connection context class constructors, unlike the Oracle.connect() method, require an auto-commit setting.

  • To use any of the first three constructors listed, you must first register your JDBC driver. This happens automatically if you are using an Oracle JDBC driver and call Oracle.connect(). Refer to "Driver Selection and Registration for Run Time".

  • Connection context classes that you declare generally have the same constructor signatures as the DefaultContext class. However, if you declare a connection context class to be associated with a data source, a different set of constructors is provided. Refer to "Standard Data Source Support" for more information.

  • When using the constructor that takes a JDBC connection object, do not initialize the connection context instance with a null JDBC connection.

  • The auto-commit setting determines whether SQL operations are automatically committed. Refer to "Basic Transaction Control" for more information.

Optional DefaultContext close() Method Parameters

When you close a connection context instance, you have the option of specifying whether or not to close the underlying physical connection. By default it is closed. This is relevant if you are sharing the physical connection between multiple connection objects, either SQLJ connection context instances or JDBC connection instances. The following examples presume a DefaultContext instance defctx.

To keep the underlying physical connection open, use the following:

defctx.close(ConnectionContext.KEEP_CONNECTION);

To close the underlying physical connection, which is the default behavior, use the following:

defctx.close(ConnectionContext.CLOSE_CONNECTION);

KEEP_CONNECTION and CLOSE_CONNECTION are static constants of the ConnectionContext interface.

See Also:

"Closing Shared Connections" for more information about using these parameters and about shared connections

Connection for Translation

If you want to use online semantics-checking during translation, then you must specify a database connection for SQLJ to use. These are referred to as exemplar schemas.

You can use different connections for translation and run time. In fact, it is often necessary or preferable to do so. It might be necessary if you are not developing the application in the same kind of environment that it will run in. But even if the run time connection is available during translation, it might be preferable to create an account with a narrower set of resources so that your online checking will be tighter. This would be true if your application uses only a small subset of the SQL entities available in the run time connection. Your online checking would be tighter and more meaningful if you create an exemplar schema consisting only of SQL entities that your application actually uses.

Use the SQLJ translator connection options, either on the command line or in a properties file, to specify a connection for translation.

Connection for Customization

Generally, Oracle customization does not require a database connection. However, the Oracle SQLJ implementation does support customizer connections. This is useful in two circumstances:

  • If you are using Oracle customizer with the optcols option enabled, then a connection is required. This option allows iterator column type and size definitions for performance optimization.

  • If you are using SQLCheckerCustomizer, a specialized customizer that performs semantics-checking on profiles, then a connection is required if you are using an online checker, which is true by default.

For Oracle-specific code generation, the SQLJ translator has an -optcols option with the same functionality. The SQLCheckerCustomizer is invoked through Oracle customizer harness verify option. Use the customizer harness user, password, url, and driver options to specify connection parameters for whatever customizer you are using, as appropriate.

NULL-Handling

Java primitive types, such as int, double, or float, cannot have null values. You must consider this in choosing your result expression and host expression types.

This section covers the following topics:

Wrapper Classes for NULL-Handling

SQLJ consistently enforces retrieving SQL NULL as Java null, in contrast to JDBC, which retrieves NULL as 0 or false for certain data types. Therefore, do not use Java primitive types in SQLJ for output variables in situations where a SQL NULL may be received, because Java primitive types cannot take null values.

This pertains to result expressions, output or input-output host expressions, and iterator column types. If the receiving Java type is primitive and an attempt is made to retrieve a SQL NULL, then a sqlj.runtime.SQLNullException is thrown and no assignment is made.

To avoid the possibility of NULL being assigned to Java primitives, use the following wrapper classes instead of primitive types:

  • java.lang.Boolean

  • java.lang.Byte

  • java.lang.Short

  • java.lang.Integer

  • java.lang.Long

  • java.lang.Double

  • java.lang.Float

In case you must convert back to a primitive value, each of these wrapper classes has an xxxValue() method. For example, intValue() returns an int value from an Integer object and floatValue() returns a float value from a Float object. For example, presuming intobj is an Integer object:

int j = intobj.intValue();

Note:

  • SQLNullException is a subclass of the standard java.sql.SQLException class.

  • Because Java objects can have null values, there is no need for indicator variables in SQLJ, such as those used in other host languages like C, C++, and COBOL.

Examples of NULL-Handling

The following examples show the use of the java.lang wrapper classes to handle NULL.

Example: Null Input Host Variable

In the following example, a Float object is used to pass a null value to the database:

int empno = 7499;
Float commission = null;

#sql { UPDATE emp SET comm = :commission WHERE empno = :empno };

You cannot use the Java primitive type float to accomplish this.

Example: Null Iterator Rows

In the following example, a Double column type is used in an iterator to allow for the possibility of null data.

For each employee in the emp table whose salary is at least $50,000, the employee name (ENAME) and commission (COMM) are selected into the iterator. Then each row is tested to determine if the COMM field is, in fact, null. If so, then it is processed accordingly.

#sql iterator EmployeeIter (String ename, Double comm);

EmployeeIter ei;
#sql ei = { SELECT ename, comm FROM emp WHERE sal >= 50000 };

while (ei.next())
{
   if (ei.comm() == null) 
      System.out.println(ei.ename() + " is not on commission.");
}
ei.close();
...

Note:

To execute a WHERE clause comparison against NULL, use the following SQL syntax:
...WHERE :x IS NULL

Exception-Handling Basics

This section covers the basics of handling exceptions in SQLJ application, including requirements for error-checking. This section covers the following topics:

SQLJ and JDBC Exception-Handling Requirements

Because SQLJ executable statements result in JDBC calls through sqlj.runtime, and JDBC requires SQL exceptions to be caught or thrown, SQLJ also requires SQL exceptions to be caught or thrown in any block containing SQLJ executable statements. Your source code will generate errors during compilation if you do not include appropriate exception-handling.

Handling SQL exceptions requires the SQLException class, which is included in the standard JDBC java.sql.* package.

Example: Exception Handling

This example demonstrates the basic exception-handling required in SQLJ applications. The code declares a main method with a try/catch block and another method, which throws SQLException when an exception is encountered. The code is as follows:

    /* Import SQLExceptions class.  The SQLException comes from
       JDBC. Executable #sql clauses result in calls to JDBC, so methods
       containing executable #sql clauses must either catch or throw
       SQLException.
    */
import java.sql.* ;
import oracle.sqlj.runtime.Oracle;

    // iterator for the select

#sql iterator MyIter (String ITEM_NAME);

public class TestInstallSQLJ 
{
    //Main method
  public static void main (String args[]) 
  {
    try { 

    // Set the default connection to the URL, user, and password
    // specified in your connect.properties file
      Oracle.connect(TestInstallSQLJ.class, "connect.properties");

      TestInstallSQLJ ti = new TestInstallSQLJ();

    // This method throws SQLException. Therefore, it ic called within a try block
      ti.runExample();

    } catch (SQLException e) { 
      System.err.println("Error running the example: " + e);
    }

  } //End of method main

  //Method that runs the example
  void runExample() throws SQLException
  {
      //Issue SQL command to clear the SALES table
    #sql { DELETE FROM SALES };
    #sql { INSERT INTO SALES(ITEM_NAME) VALUES ('Hello, SQLJ!')};

    MyIter iter;
    #sql iter = { SELECT ITEM_NAME FROM SALES };

    while (iter.next()) {
      System.out.println(iter.ITEM_NAME());
    }
  }
}

Processing Exceptions

This section discusses ways to process and interpret exceptions in your SQLJ application. During run time, exceptions may be raised from any of the following:

  • SQLJ run time

  • JDBC driver

  • RDBMS

Printing Error Text

The example in the previous section showed how to catch SQL exceptions and output the error messages. Part of that code is as follows:

...
try {
...
} catch (SQLException e) { 
      System.err.println("Error running the example: " + e); 
}
...

This will print the error text from the SQLException object.

You can also retrieve error information using the getMessage(), getErrorCode(), and getSQLState() methods the SQLException class.

Printing the error text, as in this example, prints the error message with some additional text, such as SQLException.

Retrieving SQL States and Error Codes

The java.sql.SQLException class and subclasses include the getMessage(), getErrorCode(), and getSQLState() methods. Depending on where the exception or error originated and how they are implemented there, the following methods provide additional information:

  • String getMessage()

    If the error originates in the SQLJ run time or JDBC driver, then this method returns the error message with no prefix. If the error originates in the RDBMS, then it returns the error message prefixed by the ORA number.

  • int getErrorCode()

    If the error originates in the SQLJ run time, then this method returns no meaningful information. If the error originates in the JDBC driver or RDBMS, then it returns the five-digit ORA number as an integer.

  • String getSQLState()

    If the error originates in the SQLJ run time, then this method returns a string with a five-digit code indicating the SQL state. If the error originates in the JDBC driver, then it returns no meaningful information. If the error originates in the RDBMS, then it returns the five-digit SQL state. Your application should have appropriate code to handle null values returned.

The following example prints the error message and also checks the SQL state:

...
try {
...
} catch (SQLException e) { 
      System.err.println("Error running the example: " + e); 
      String sqlState = e.getSQLState();
      System.err.println("SQL state = " + sqlState); 
}
...

Using SQLException Subclasses

For more specific error-checking, use any available and appropriate subclasses of the java.sql.SQLException class.

SQLJ provides the sqlj.runtime.NullException class, which is a subclass of java.sql.SQLException. You can use this exception in situations where a NULL might be returned into a Java primitive variable.

For batch-enabled environments, there is also the standard java.sql.BatchUpdateException subclass. Refer to "Error Conditions During Batch Execution" for further information.

When you use a subclass of SQLException, catch the subclass exception before catching SQLException, as in the following example:

...
try {
...
} catch (SQLNullException ne) {
     System.err.println("Null value encountered: " + ne); }
  catch (SQLException e) { 
     System.err.println("Error running the example: " + e); }
...

This is because a subclass exception can also be caught as a SQLException. If you catch SQLException first, then execution will not proceed to the part where you have coded special processing for the subclass exception.

Basic Transaction Control

This section discusses how to manage data updates. It covers the following topics:

Overview of Transactions

A transaction is a sequence of SQL operations that Oracle treats as a single unit. A transaction begins with the first executable SQL statement after any of the following:

  • Connection to the database

  • COMMIT (committing data updates, either automatically or manually)

  • ROLLBACK (canceling data updates)

A transaction ends with a COMMIT or ROLLBACK operation.

Note:

In Oracle Database 11g, all data definition language (DDL) statements, such as CREATE and ALTER, include an implicit COMMIT. This will commit not only the DDL statement, but all the preceding data manipulation language (DML) statements, such as INSERT, DELETE, and UPDATE, that have not yet been committed or rolled back.

Automatic Commits Versus Manual Commits

In using SQLJ or JDBC, you can either have your data updates automatically committed or commit them manually. In either case, each COMMIT operation starts a new transaction. You can specify that changes be committed automatically by enabling the auto-commit flag. This can be done either when you define a SQLJ connection or by using the setAutoCommit() method of the underlying JDBC connection object of an existing connection. You can use manual control by disabling the auto-commit flag and using SQLJ COMMIT and ROLLBACK statements.

Enabling auto-commit may be more convenient, but gives you less control. For example, you have no option to roll back changes. In addition, some SQLJ or JDBC features are incompatible with auto-commit mode. For example, you must disable the auto-commit flag for update batching or SELECT FOR UPDATE syntax to work properly.

Specifying Auto-Commit as You Define a Connection

When you use the Oracle.connect() or Oracle.getConnection() method to create a DefaultContext instance and define a connection, the auto-commit flag is set to false by default. However, there are signatures of these methods that enable you to set this flag explicitly. The auto-commit flag is always the last parameter.

The following is an example of instantiating DefaultContext and using the default false setting for auto-commit mode:

Oracle.getConnection
   ("jdbc:oracle:thin:@localhost:1521/myservice", "scott", "tiger");

Alternatively, you can specify a true setting as follows:

Oracle.getConnection
   ("jdbc:oracle:thin:@localhost:1521/myservice", "scott", "tiger", true);

If you use a constructor to create a connection context instance, either of DefaultContext or of a declared connection context class, then you must specify the auto-commit setting. Again, it is the last parameter, as in the following example:

DefaultContext ctx = new DefaultContext
   ("jdbc:oracle:thin:@localhost:1521/myservice", "scott", "tiger", false);

If you have reason to create a JDBC Connection instance directly, then the auto-commit flag is set to true by default if your program runs on a client, or false by default if it runs in the server. You cannot specify an auto-commit setting when you create a JDBC Connection instance directly, but you can use the setAutoCommit() method to alter the setting.

Note:

Auto-commit functionality is not supported by the JDBC server-side internal driver.

Modifying Auto-Commit in an Existing Connection

There is typically no reason to change the auto-commit flag setting for an existing connection, but you can if you desire. You can do this by using the setAutoCommit() method of the underlying JDBC connection object.

You can retrieve the underlying JDBC connection object by using the getConnection() method of any SQLJ connection context instance, whether it is an instance of the DefaultContext class or of a connection context class that you declared.

You can accomplish these two steps at once, as follows:

ctx.getConnection().setAutoCommit(false);

or:

ctx.getConnection().setAutoCommit(true);

In these examples, ctx is a SQLJ connection context instance.

Note:

Do not alter the auto-commit setting in the middle of a transaction.

Using Manual COMMIT and ROLLBACK

If you disable the auto-commit flag, then you must manually commit any data updates. To commit any changes that have been executed since the last COMMIT operation, use the SQLJ COMMIT statement, as follows:

#sql { COMMIT };

To roll back any changes that have been executed since the last COMMIT operation, use the SQLJ ROLLBACK statement, as follows:

#sql { ROLLBACK };

Note:

  • Do not use the COMMIT and ROLLBACK commands when auto-commit is enabled. This will result in unspecified behavior, or even SQL exceptions could be raised.

  • You can also roll back to a specified savepoint. Refer to "Using Savepoints".

  • All DDL statements in Oracle SQL syntax include an implicit COMMIT operation. There is no special SQLJ functionality in this regard. Such statements follow standard Oracle SQL rules.

  • If auto-commit mode is off and you close a connection context instance from a client application, then any changes since your last COMMIT will be committed, unless you close the connection context instance with KEEP_CONNECTION. Refer to "Closing Shared Connections" for more information.

Effect of Commits and Rollbacks on Iterators and Result Sets

COMMIT and ROLLBACK operations do not affect open result sets and iterators. The result sets and iterators will still be open. Usually, all that is relevant to their content is the state of the database at the time of execution of the SELECT statements that populated them.

Note:

An exception to this is if you declared an iterator class with sensitivity=SENSITIVE. In this case, changes to the underlying result set may be seen whenever the iterator is scrolled outside of its window size. For more information about scrollable iterators, refer to "Scrollable Iterators". For more information about the underlying scrollable result sets, refer to the Oracle Database JDBC Developer's Guide and Reference.

This also applies to UPDATE, INSERT, and DELETE statements that are executed after the SELECT statements. Execution of these statements does not affect the contents of open result sets and iterators.

Consider a situation where you SELECT, then UPDATE, and then COMMIT. A nonsensitive result set or iterator populated by the SELECT statement will be unaffected by the UPDATE and COMMIT.

As a further example, consider a situation where you UPDATE, then SELECT, and then ROLLBACK. A nonsensitive result set or iterator populated by the SELECT will still contain the updated data, regardless of the subsequent ROLLBACK.

Using Savepoints

The JDBC 3.0 specification added support for savepoints. A savepoint is a defined point in a transaction that you can roll back to, if desired, instead of rolling back the entire transaction. The savepoint is the point in the transaction where the SAVEPOINT statement appears.

In Oracle9i Database Release 2 (9.2), SQLJ first included Oracle-specific syntax to support savepoints. In Oracle Database 11g, SQLJ adds support for ISO savepoint syntax.

Support for ISO SQLJ Savepoint Syntax

In ISO syntax, use a string literal in a SAVEPOINT statement to designate a name for a savepoint. This can be done as follows:

#sql { SAVEPOINT savepoint1 };

If you want to roll back changes to that savepoint, then you can refer to the specified name later in a ROLLBACK TO statement, as follows:

#sql { ROLLBACK TO savepoint1 };

Use a RELEASE SAVEPOINT statement if you no longer need the savepoint:

#sql { RELEASE SAVEPOINT savepoint1 };

Savepoints are saved in the SQLJ execution context, which has methods that parallel the functionality of these three statements.

Because any COMMIT operation ends the transaction, this also releases all savepoints of the transaction.

Oracle SQLJ Savepoint Syntax

In addition to the ISO syntax, the following Oracle-specific syntax for savepoints is supported. Note that the Oracle syntax uses string host expressions, rather than string literals.

You can set a savepoint as follows:

#sql { SET SAVEPOINT :savepoint };

The host expression, savepoint in this example, is a variable that specifies the name of the savepoint as a Java String.

You can roll back to a savepoint as follows:

#sql { ROLLBACK TO :savepoint };

To release a savepoint, use the following SQLJ statement:

#sql { RELEASE :savepoint };

Note:

Oracle-specific syntax will continue to be supported for backward compatibility. Note the following differences between Oracle syntax and ISO syntax:
  • Oracle syntax takes string variables rather than string literals.

  • Oracle syntax uses SET SAVEPOINT instead of SAVEPOINT.

  • Oracle syntax uses RELEASE instead of RELEASE SAVEPOINT.

Summary: First Steps in SQLJ Code

The best way to summarize the SQLJ executable statement features and functionality discussed to this point is by examining short but complete programs. This section presents two such examples.

The first example, presented one step at a time and then again in its entirety, uses a SELECT INTO statement to perform a single-row query of two columns from a table of employees. If you want to run the example, ensure that you change the parameters in the connect.properties file to settings that will let you connect to an appropriate database.

The second example, slightly more complicated, will make use of a SQLJ iterator for a multi-row query.

Import Required Classes

Import any JDBC or SQLJ packages you will need. You will need at least some of the classes in the java.sql package:

import java.sql.*;

You may not need all the java.sql package. Key classes are java.sql.SQLException and any classes that you refer to explicitly. For example, java.sql.Date and java.sql.ResultSet.

You will need the following package for the Oracle class, which you typically use to instantiate DefaultContext objects and establish your default connection:

import oracle.sqlj.runtime.*;

If you will be using any SQLJ run time classes directly in your code, then import the following packages:

import sqlj.runtime.*;
import sqlj.runtime.ref.*;

However, even if your code does not use any SQLJ run time classes directly, it will be sufficient to have them in the CLASSPATH.

Key run time classes include ResultSetIterator and ExecutionContext in the sqlj.runtime package and DefaultContext in the sqlj.runtime.ref package.

Register JDBC Drivers and Set Default Connection

Declare the SimpleExample class with a constructor that uses the static Oracle.connect() method to set the default connection. This also registers Oracle JDBC drivers.

This uses a signature of connect() that takes the URL, user name, and password from the connect.properties file. An example of this file is in the directory ORACLE_HOME/sqlj/demo and also in "Set Up the Run Time Connection".

public class SimpleExample {

  public SimpleExample() throws SQLException {
    // Set default connection (as defined in connect.properties).
    Oracle.connect(getClass(), "connect.properties");
  }

Set Up Exception Handling

Create a main() that calls the SimpleExample constructor and then sets up a try/catch block to handle any SQL exceptions thrown by the runExample() method, which performs the real work of this application:

...
public static void main (String [] args) {
    
   try {
      SimpleExample o1 = new SimpleExample();
      o1.runExample();
   }
   catch (SQLException ex) {
      System.err.println("Error running the example: " + ex);
   }
}
...

You can also use a try/catch block inside a finally clause when you close the connection, presuming the finally clause is not already inside a try/catch block in case of SQL exceptions:

finally
{
   try { Oracle.close(); } catch(SQLException ex) {...}
}

Set Up Host Variables, Execute SQLJ Clause, Process Results

Create a runExample() method that performs the following:

  1. Throws any SQL exceptions to the main() method for processing.

  2. Declares Java host variables.

  3. Executes a SQLJ clause that binds the Java host variables into an embedded SELECT statement and selects the data into the host variables.

  4. Prints the results.

The code for this method is as follows:

void runExample() throws SQLException {
     
     System.out.println( "Running the example--" );
     
     // Declare two Java host variables--
     Float salary;
     String empname;

     // Use SELECT INTO statement to execute query and retrieve values.
      #sql { SELECT ename, sal INTO :empname, :salary FROM emp
             WHERE empno = 7499 };
     
     // Print the results--
     System.out.println("Name is " + empname + ", and Salary is " + salary);
  }
}    // Closing brace of SimpleExample class

This example declares salary and empname as Java host variables. The SQLJ clause then selects data from the ename and sal columns of the emp table and places the data into the host variables. Finally, the values of salary and empname are printed.

Note that this SELECT statement could select only one row of the emp table, because the empno column in the WHERE clause is the primary key of the table.

Example of Single-Row Query using SELECT INTO

This section presents the entire SimpleExample class from the previous step-by-step sections. Because this is a single-row query, no iterator is required.

// Import SQLJ classes:
import sqlj.runtime.*;
import sqlj.runtime.ref.*;
import oracle.sqlj.runtime.*;

// Import standard java.sql package:
import java.sql.*;

public class SimpleExample {

  public SimpleExample() throws SQLException {
    // Set default connection (as defined in connect.properties).
    Oracle.connect(getClass(), "connect.properties");
  }

  public static void main (String [] args) throws SQLException {
    
    try {
      SimpleExample o1 = new SimpleExample();
      o1.runExample();
    }
    catch (SQLException ex) {
      System.err.println("Error running the example: " + ex);
    }
  }

  finally
  {
     try { Oracle.close(); } catch(SQLException ex) {...}
  }

  void runExample() throws SQLException {
     
     System.out.println( "Running the example--" );
     
     // Declare two Java host variables--
     Float salary;
     String empname;

     // Use SELECT INTO statement to execute query and retrieve values.
        #sql { SELECT ename, sal INTO :empname, :salary FROM emp
              WHERE empno = 7499 };
     
     // Print the results--
     System.out.println("Name is " + empname + ", and Salary is " + salary);
  }
}

Set Up a Named Iterator

This example builds on the previous example by adding a named iterator and using it for a multiple-row query.

First, declare the iterator class. Use object types Integer and Float, instead of primitive types int and float, wherever there is the possibility of NULL values.

#sql iterator EmpRecs(
      int empno,       // This column cannot be null, so int is OK.
                       // (If null is possible, use Integer.)
      String ename,
      String job,
      Integer mgr,
      Date hiredate,
      Float sal,
      Float comm,
      int deptno);

Next, instantiate the EmpRecs class and populate it with query results.

EmpRecs employees;

#sql employees = { SELECT empno, ename, job, mgr, hiredate,
                   sal, comm, deptno FROM emp };

Then, use the next() method of the iterator to print the results.

    while (employees.next())  {
      System.out.println( "Name:       " + employees.ename() );
      System.out.println( "EMPNO:      " + employees.empno() );
      System.out.println( "Job:        " + employees.job() );
      System.out.println( "Manager:    " + employees.mgr() );
      System.out.println( "Date hired: " + employees.hiredate() );
      System.out.println( "Salary:     " + employees.sal() );
      System.out.println( "Commission: " + employees.comm() );
      System.out.println( "Department: " + employees.deptno() );
      System.out.println();
    }

Finally, close the iterator.

employees.close();

Example of Multiple-Row Query Using Named Iterator

This example uses a named iterator for a multiple-row query that selects several columns of data from a table of employees.

Apart from use of the named iterator, this example is conceptually similar to the previous single-row query example.

// Import SQLJ classes:
import sqlj.runtime.*;
import sqlj.runtime.ref.*;
import oracle.sqlj.runtime.*;

// Import standard java.sql package:
import java.sql.*;

// Declare a SQLJ iterator.
// Use object types (Integer, Float) for mgr, sal, And comm rather
// than primitive types to allow for possible null selection.

#sql iterator EmpRecs(
      int empno,       // This column cannot be null, so int is OK.
                       // (If null is possible, Integer is required.)
      String ename,
      String job,
      Integer mgr,
      Date hiredate,
      Float sal,
      Float comm,
      int deptno);

// This is the application class.  
public class EmpDemo1App {

   public EmpDemo1App() throws SQLException {
      // Set default connection (as defined in connect.properties).
      Oracle.connect(getClass(), "connect.properties");
   }

  public static void main(String[] args) {

    try {
      EmpDemo1App app = new EmpDemo1App();
      app.runExample();
    }
    catch( SQLException exception ) {
      System.err.println( "Error running the example: " + exception );
    }
  }

  finally
  {
     try { Oracle.close(); } catch(SQLException ex) {...}
  }

  void runExample() throws SQLException  {
    System.out.println("\nRunning the example.\n" );

    // The query creates a new instance of the iterator and stores it in
    // the variable 'employees' of type 'EmpRecs'.  SQLJ translator has
    // automatically declared the iterator so that it has methods for
    // accessing the rows and columns of the result set.

    EmpRecs employees;

    #sql employees = { SELECT empno, ename, job, mgr, hiredate,
                       sal, comm, deptno FROM emp };

    // Print the result using the iterator.

    // Note how the next row is accessed using method 'next()', and how
    // the columns can be accessed with methods that are named after the
    // actual database column names.

    while (employees.next())  {
      System.out.println( "Name:       " + employees.ename() );
      System.out.println( "EMPNO:      " + employees.empno() );
      System.out.println( "Job:        " + employees.job() );
      System.out.println( "Manager:    " + employees.mgr() );
      System.out.println( "Date hired: " + employees.hiredate() );
      System.out.println( "Salary:     " + employees.sal() );
      System.out.println( "Commission: " + employees.comm() );
      System.out.println( "Department: " + employees.deptno() );
      System.out.println();
    }

    // You must close the iterator when it's no longer needed.
    employees.close() ;
  }
}

Oracle-Specific Code Generation (No Profiles)

Throughout this manual there is general and standard discussion of the SQLJ run time layer and SQLJ profiles. However, the Oracle SQLJ implementation, by default, generates Oracle-specific code with direct calls to Oracle JDBC driver instead of generating ISO standard code that calls the SQLJ run time. With Oracle-specific code generation, there are no profile files, and the role of the SQLJ run time layer is greatly reduced during program execution. Oracle-specific code supports all Oracle-specific extended features.

Code generation is determined through the SQLJ translator -codegen option. The default setting for Oracle-specific code generation is -codegen=oracle. Alternatively, you can set -codegen=iso for code generation according to the ISO standard.

This section covers the following topics:

Code Considerations and Limitations with Oracle-Specific Code Generation

When coding a SQLJ application where Oracle-specific code generation will be used, be aware of the following programming considerations and restrictions:

  • To use a nondefault statement cache size, you must include appropriate method calls in your code, because Oracle customizer stmtcache option is unavailable. Refer to "SQLJ Usage Changes with Oracle-Specific Code Generation".

  • Do not mix Oracle-specific generated code with ISO standard generated code in the same application. However, if Oracle-specific code and ISO standard code must share the same connection, do one of the following:

    • Ensure that the Oracle-specific code and ISO standard code use different SQLJ execution context instances. Refer to "Execution Contexts" for information about SQLJ execution contexts.

    • Place a transaction boundary, that is, as a manual COMMIT or ROLLBACK statement, between the two kinds of code.

    This limitation regarding mixing code is especially significant for server-side code, because all Java code running in a given session uses the same JDBC connection and SQLJ connection context.

  • Do not rely on side effects in parameter expressions when values are returned from the database. Oracle-specific code generation does not create temporary variables for evaluation of OUT parameters, IN OUT parameters, SELECT INTO variables, or return arguments on SQL statements.

    For example, avoid statements such as the following:

    #sql { SELECT * FROM EMP INTO :(x[i++]), :(f_with_sideffect()[i++]),
                                  :(a.b[i]) };
    
    

    or:

    #sql x[i++] = { VALUES f(:INOUT (x[i++]), :OUT (f_with_sideffect())) };
    
    

    Evaluation of arguments is performed in place in the generated code. This may result in different behavior than when evaluation is according to ISO SQLJ standards.

  • Type maps for Oracle object functionality assumes that the corresponding Java classes implement the java.sql.SQLData interface, given that JPublisher-generated Java classes do not otherwise require a type map. If you use type maps for Oracle object functionality, then your iterator declarations and connection context declarations must specify the same type maps. Specify this through the with clause.

    For example, if you declare a connection context class as follows:

    #sql context TypeMapContext with (typeMap="MyTypeMap");
    
    

    and you populate an iterator instance from a SQLJ statement that uses an instance of this connection context class, as follows:

    TypeMapContext tmc = new TypeMapContext(...);
    ...
    MyIterator it;
    #sql [tmc] it = ( SELECT pers, addr FROM tab WHERE ...);
    
    

    then the iterator declaration is required to have specified the same type map, as follows:

    #sql iterator MyIterator with (typeMap="MyTypeMap") 
                  (Person pers, Address addr);
    
    

    Note:

    The reason for this restriction is that with Oracle-specific code generation, all iterator getter methods are fully generated as Oracle JDBC calls during translation. To generate the proper calls, the SQLJ translator must know whether an iterator will be used with a particular type map.

SQLJ Usage Changes with Oracle-Specific Code Generation

Some options that were previously available only as Oracle customizer options are useful with Oracle-specific code generation as well. Because profile customization is not applicable with Oracle-specific code generation, these options have been made available through other means.

To alter the statement cache size or disable statement caching when generating Oracle-specific code, use method calls in your code instead of using the customizer stmtcache option. The sqlj.runtime.ref.DefaultContext class, as well as any connection context class you declare, now has the following static methods:

  • setDefaultStmtCacheSize(int)

  • int getDefaultStmtCacheSize()

It also has the following instance methods:

  • setStmtCacheSize(int)

  • int getStmtCacheSize()

By default, statement caching is enabled.

In addition, the following options are available as front-end Oracle SQLJ translator options as well as Oracle customizer options:

  • -optcols: Enable iterator column type and size definitions to optimize performance.

  • -optparams: Enable parameter size definitions to optimize JDBC resource allocation. This option is used in conjunction with optparamdefaults.

  • -optparamdefaults: Set parameter size defaults for particular data types. This option is used in conjunction with optparams.

  • -fixedchar: Enable CHAR comparisons with blank padding for WHERE clauses.

Be aware of the following:

  • Use the -optcols option only if you are using online semantics-checking, where you have used the SQLJ translator -user, -password, and -url options appropriately to request a database connection during translation.

  • The functionality of the -optcols, -optparams, and -optparamdefaults options, including default values, is the same as for the corresponding customizer options.

Server-Side Considerations with Oracle-Specific Code Generation

Consider the following if your SQLJ code will run in the server:

  • The server-side SQLJ translator no longer supports ISO standard generated code. SQLJ source code that is loaded into the server and compiled there will always be translated with the default -codegen=oracle setting.

    Therefore, to use ISO standard generated code in the server, you must translate and compile the SQLJ code on a client and then load the individual components into the server.

  • The caution against mixing Oracle-specific generated code with ISO standard generated code applies to server-side Java code that calls a Java stored procedure or stored function, even if the stored procedure is invoked through a PL/SQL wrapper. This constitutes a recursive call-in. By default, the ExecutionContext object is shared by both the calling module and the called module. Therefore, both modules should be translated with the same -codegen setting.

    If you want to ensure interoperability with code that has been translated with ISO standard code generation, then it is advisable to explicitly instantiate execution context instances, as in the following example:

    public static method() throws SQLException
    {
       Execution Context ec = new ExecutionContext();
       ...
       try {
          ...
          #sql [ec] { SQL operation };
          ...
       } finally { ec.close(); }
    }
    
    

Note:

To avoid resource leakage when using an explicit ExecutionContext instance, ensure that you use the close() method, as shown in this example.

Advantages and Disadvantages of Oracle-Specific Code Generation

Oracle-specific code generation offers following advantages over ISO standard code generation:

  • Applications run more efficiently. The code calls JDBC application programming interfaces (APIs) directly, placing run-time performance directly at the JDBC level. The role of the intermediate SQLJ run time layer is greatly reduced during program execution.

  • Applications are smaller in size.

  • No profile files (.ser) are produced. This is especially convenient if you are loading a translated application into the database or porting it to another system, because there are fewer components.

  • Translation is faster, because there is no profile customization step.

  • During execution, Oracle SQLJ run time and Oracle JDBC driver use the same statement cache resources, so partitioning resources between the two is unnecessary.

  • Having the SQL-specific information appear in the Java class files instead of in separate profile files avoids potential security issues.

  • You need not have to rewrite your code to take advantage of possible future Oracle JDBC performance enhancements, such as enhancements being considered for execution of static SQL code. Future releases of Oracle SQLJ translator will handle this automatically.

  • The use of Java reflection at run time is eliminated, and thus, provides full portability to browser environments.

However. there are a few disadvantages:

ISO Standard Code Generation

This section covers the following topics:

Environment Requirements for ISO Standard Code Generation

The Oracle SQLJ implementation, by default, generates Oracle-specific code with direct calls to Oracle JDBC driver instead of generating ISO standard code that calls the SQLJ run time. The following is a typical environment setup for ISO standard code generation:

  • SQLJ code generation: -codegen=iso

  • SQLJ translation library: translator.jar

  • SQLJ run time library: runtime12.jar with JDK 1.5.x and Oracle JDBC driver 11g release 1 (11.1).

  • JDBC drivers: Oracle 11g release 1 (11.1)

  • JDK version: 1.5.x

SQLJ Translator and SQLJ Run Time

The following section describes the differences in Oracle SQLJ implementation in case of ISO standard code generation:

  • SQLJ translator: Along with the .java file, the translator also produces one or more SQLJ profiles for ISO standard code generation. These profiles contain information about the embedded SQL operations. SQLJ then automatically invokes a Java compiler to produce .class files from the .java file.

  • SQLJ run time: For ISO standard code generation, the SQLJ run time implements the desired actions of the SQL operations by accessing the database using a JDBC driver. The generic ISO SQLJ standard does not require that a SQLJ run time use a JDBC driver to access the database.

    See Also:

    "SQLJ Run Time"

In addition to the translator and run time, there is a component known as the customizer that plays a role. A customizer tailors SQLJ profiles for a particular database implementation and vendor-specific features and data types. By default, for ISO standard code, the SQLJ front end invokes an Oracle customizer to tailor your profiles for Oracle Database instance and Oracle-specific features and data types.

When you use Oracle customizer during translation, your application will require the SQLJ run time and an Oracle JDBC driver when it runs.

Note:

From Oracle Database 10g release 1 (10.1), only Oracle JDBC drivers are supported with SQLJ.

SQLJ Profiles

With ISO standard code generation, SQLJ profiles are serialized Java resources or classes generated by the SQLJ translator, which contain details about the embedded SQL statements. The translator creates these profiles. Then, depending on the translator option settings, it either serializes the profiles and puts them into binary resource files or puts them into .class files.

This section covers the following topics:

Overview of Profiles

SQLJ profiles are used in ISO standard code for implementing the embedded SQL operations in SQLJ executable statements. Profiles contain information about the SQL operations and the types and modes of data being accessed. A profile consists of a collection of entries, where each entry maps to one SQL operation. Each entry fully specifies the corresponding SQL operation, describing each of the parameters used in processing this instruction.

SQLJ generates a profile for each connection context class in your application, where each connection context class corresponds to a particular set of SQL entities you use in your database operations. There is one default connection context class, and you can declare additional classes. The ISO SQLJ standard requires that the profiles be of standard format and content. Therefore, for your application to use vendor-specific extended features, your profiles must be customized. By default, this occurs automatically, with your profiles being customized to use Oracle-specific extended features.

Profile customization enables vendors to add value in the following ways:

  • Vendors can support their own specific data types and SQL syntax. For example, Oracle customizer maps standard JDBC PreparedStatement method calls in translated SQLJ code to OraclePreparedStatement method calls, which provide support for Oracle type extensions.

  • Vendors can improve performance through specific optimizations.

Note:

  • By default, SQLJ profile file names have the .ser extension, but this does not mean that all .ser files are profiles. Other serialized objects can use this extension, and a SQLJ program unit can use serialized objects other than its profiles. Optionally, profiles can be converted to .class files instead of .ser files.

  • A SQLJ profile is not produced if there are no SQLJ executable statements in the source code.

Binary Portability

SQLJ-generated profile files support binary portability. That is, you can port them as is and use them with other kinds of databases or in other environments, if you have not used vendor-specific data types or features. This is true for generated .class files as well.

SQLJ Translation Steps

For ISO standard code generation (-codegen=iso), the translator processes the SQLJ source code, converts SQL operations to SQLJ run-time calls, and generates Java output code and one or more SQLJ profiles. A separate profile is generated for each connection context class in the source code, where a different connection context class is typically used for each interrelated set of SQL entities that is used in the operations.

Generated Java code is put into a .java output file containing the following:

  • Any class definitions and Java code from the .sqlj source file

  • Class definitions created as a result of the SQLJ iterator and connection context declarations

  • A class definition for a specialized class known as the profile-keys class that SQLJ generates and uses in conjunction with the profiles (for ISO standard SQLJ code generation only)

  • Calls to the SQLJ run time to implement the actions of the embedded SQL operations

Generated profiles contain information about all the embedded SQL statements in the SQLJ source code, such as actions to take, data types being manipulated, and tables being accessed. When the application is run, the SQLJ run time accesses the profiles to retrieve the SQL operations and passes them to the JDBC driver.

By default, profiles are put into .ser serialized resource files, but SQLJ can optionally convert the .ser files to .class files as part of the translation.

The compiler compiles the generated Java source file and produces Java .class files as appropriate. This includes a .class file for each class that is defined, each of the SQLJ declarations, and the profile-keys class. The JVM then invokes Oracle customizer or other specified customizer to customize the profiles generated.

General SQLJ Notes

Consider the following when translating and running SQLJ applications for ISO specific code generation:

  • Along with compiling existing .java files on the command line and making them available for type resolution, as for Oracle-specific code generation, you need to:

    • Customize the existing profiles

    • Customize the Java Archive (JAR) files containing profiles

  • SQLJ generates profiles and the profile-keys class only if your source code includes SQLJ executable statements.

  • If you use Oracle customizer during translation, then your application requires Oracle SQLJ run time and an Oracle JDBC driver when it runs, even if your code does not use Oracle-specific features. You can avoid this by specifying -profile=false when you translate, to bypass Oracle-specific customization.

Summary of Translator Input and Output

We have seen what the SQLJ translator takes as input, what it produces as output, and where it places its output in case of Oracle-specific code generation. This section covers the same topics for ISO standard code generation:

Translator Input

Similar to Oracle -specific code generation, the SQLJ translator takes one or more .sqlj source files as input, which can be specified on the command line. The name of the main .sqlj file is based on the public class it defines, if any, else on the first class it defines.

Translator Output

The translation step produces a Java source file for each .sqlj file in the application and at least one application profile for ISO standard code generation, presuming the source code uses SQLJ executable statements.

SQLJ generates Java source files and application profiles as follows:

  • Similar to Oracle-specific code generation, Java source files are .java files with the same base names as the .sqlj files.

  • The application profile files, if applicable, contain information about the SQL operations of the SQLJ application. There is one profile for each connection class that is used in the application. The profiles have names with the same base name as the main .sqlj file and the following extensions:

    _SJProfile0.ser
    _SJProfile1.ser
    _SJProfile2.ser
    ...
    
    

    For example, for MyClass.sqlj the translator produces:

    MyClass_SJProfile0.ser
    
    

    The .ser file extension indicates that the profiles are serialized. The .ser files are binary files.

    Note:

    The -ser2class translator option instructs the translator to generate profiles as .class files instead of .ser files. Other than the file name extension, the naming is the same.

Similar to the compilation step of Oracle-specific code generation, compiling the Java source file into multiple class files generates one .class file for each class defined in the .sqlj source file. But in case of ISO code generation, a .class file is also generated for a class known as the profile-keys class that the translator generates and uses with the profiles to implement the SQL operations. Additional .class files are produced if you declare any SQLJ iterators or connection contexts. Also, like Oracle-specific code generation, separate .class files are produced for any inner classes or anonymous classes in the code.

The .class files are named as follows:

  • Like Oracle-specific code generation, the class file for each class defined consists of the name of the class with the .class extension.

  • The profile-keys class that the translator generates is named according to the base name of the main .sqlj file, plus the following:

    _SJProfileKeys
    
    

    So, the class file has the following extension:

    _SJProfileKeys.class
    
    

    For example, for MyClass.sqlj, the translator together with the compiler produces:

    MyClass_SJProfileKeys.class
    
    
  • Like Oracle-specific code generation, the translator names iterator classes and connection context classes according to how you declare them.

The customization step alters the profiles but produces no additional output.

Note:

It is not necessary to reference SQLJ profiles or the profile-keys class directly. This is all handled automatically.

Output File Locations

The output file locations are the same for both Oracle-specific code generation and ISO standard code generation.

SQLJ Run-Time Processing

This section discusses run-time processing for ISO standard code during program execution.

For ISO standard SQLJ applications, the SQLJ run time reads the profiles and creates connected profiles, which incorporate database connections. Then the following occurs each time the application must access the database:

  1. SQLJ-generated application code uses methods in a SQLJ-generated profile-keys class to access the connected profile and read the relevant SQL operations. There is a mapping between SQLJ executable statements in the application and SQL operations in the profile.

  2. The SQLJ-generated application code calls the SQLJ run time, which reads the SQL operations from the profile.

  3. The SQLJ run time calls the JDBC driver and passes the SQL operations to the driver.

  4. The SQLJ run time passes any input parameters to the JDBC driver.

  5. The JDBC driver executes the SQL operations.

  6. If any data is to be returned, then the database sends it to the JDBC driver, which sends it to the SQLJ run time for use by your application.

Note:

Passing input parameters can also be referred to as binding input parameters or binding host expressions. The terms host variables, host expressions, bind variables, and bind expressions are all used to describe Java variables or expressions that are used as input or output for SQL operations.

Deployment Scenarios

We have discussed how to run Oracle-specific SQLJ code in the following scenarios:

  • From an applet

  • In the server (optionally running the SQLJ translator in the server as well)

There are a few considerations that you need to make while running your ISO standard code from an applet:

  • You must package all the SQLJ run time packages with your applet. The packages are:

    sqlj.runtime
    sqlj.runtime.ref
    sqlj.runtime.profile
    sqlj.runtime.profile.ref
    sqlj.runtime.error
    
    

    Also package the following if you used Oracle customization:

    oracle.sqlj.runtime
    oracle.sqlj.runtime.error
    
    

    These packages are included with your Oracle installation in one of several run time libraries in the ORACLE_HOME/lib directory.

  • Applets using Oracle-specific features require Oracle SQLJ run time to work. Oracle SQLJ run time consists of the classes in the SQLJ run time library file under oracle.sqlj.*. Oracle SQLJ runtime.jar library requires the Java Reflection API, java.lang.reflect.*. Most browsers do not support the Reflection API or impose security restrictions, but the Sun Microsystems Java plug-in provides support for the Reflection API.

    With ISO standard code generation, the following SQLJ language features always require the Java Reflection API, regardless of the version of the SQLJ run time you are using:

    • The CAST statement

    • REF CURSOR parameters or REF CURSOR columns being retrieved from the database as instances of a SQLJ iterator

    • Retrieval of java.sql.Ref, Struct, Array, Blob, or Clob objects

    • Retrieval of SQL objects as instances of Java classes implementing the oracle.sql.ORAData or java.sql.SQLData interfaces

      Note:

      • An exception to the preceding is if you use SQLJ in a mode that is fully compatible with ISO. That is, if you use SQLJ in an environment that complies with J2EE and you translate and run your program with the SQLJ runtime12ee.jar library, and you employ connection context type maps as specified by ISO. In this case, instances of java.sql.Ref, Struct, Array, Blob, Clob, and SQLData are being retrieved without the use of reflection.

      • If you use Oracle-specific code generation, then you will eliminate the use of reflection in all of the instances listed.

Oracle-Specific Code Generation Versus ISO Standard Code Generation

The Oracle SQLJ implementation provides the option of Oracle-specific code generation, where Oracle JDBC calls are generated directly in the code. This is the default behavior. In the case of Oracle-specific code generation, you must be aware of the following:

  • There are no profile files, and therefore, there is no customization step during translation.

  • At run time, SQL operations do not have to go through the SQLJ run time layer, because JDBC calls, instead of the SQLJ run time calls, are directly generated in the translated code.

Requirements and Restrictions for Naming

There are four areas to consider in discussing naming requirements, naming restrictions, and reserved words:

  • The Java namespace, including additional restrictions imposed by SQLJ on the naming of local variables and classes

  • The SQLJ namespace

  • The SQL namespace

  • Source file names

This section covers the following topics:

Java Namespace: Local Variable and Class Naming Restrictions

The Java namespace applies to all standard Java statements and declarations, including the naming of Java classes and local variables. All standard Java naming restrictions apply, and you should avoid the use of Java reserved words.

In addition, SQLJ places minor restrictions on the naming of local variables and classes.

Note:

Naming restrictions particular to host variables are discussed in "Restrictions on Host Expressions".

Local Variable Naming Restrictions

Some of the functionality of the SQLJ translator results in minor restrictions in naming local variables. The SQLJ translator replaces each SQLJ executable statement with a statement block, where the SQLJ executable statement is of the standard syntax:

#sql { SQL operation };  

SQLJ may use temporary variable declarations within a generated statement block. The name of any such temporary variables will include the following prefix:

 __sJT_

Note:

There are two underscores at the beginning and one at the end.

The following declarations are examples of those that might occur in a SQLJ-generated statement block:

int __sJT_index;
Object __sJT_key;
java.sql.PreparedStatement __sJT_stmt;

The string __sJT_ is a reserved prefix for SQLJ-generated variable names. SQLJ programmers must not use this string as a prefix for the following:

  • Names of variables declared in blocks that include executable SQL statements

  • Names of parameters to methods that contain executable SQL statements

  • Names of fields in classes that contain executable SQL statements, or whose subclasses or enclosed classes contain executable SQL statements

Class Naming Restrictions

Be aware of the following minor restrictions in naming classes in SQLJ applications:

  • You must not declare class names that may conflict with SQLJ internal classes. In particular, a top-level class cannot have a name of the following form, where a is the name of an existing class in the SQLJ application:

    a_SJb
    
    

    where, a and b are legal Java identifiers.

    For example, if your application class is Foo in file Foo.sqlj, then SQLJ generates a profile-keys class called Foo_SJProfileKeys. Do not declare a class name that conflicts with this.

  • A class containing SQLJ executable statements must not have a name that is the same as the first component of the name of any package that includes a Java type used in the application. Examples of class names to avoid are java, sqlj, and oracle (case-sensitive). As another example, if your SQLJ statements use host variables whose type is abc.def.MyClass, then you cannot use abc as the name of the class that uses these host variables.

    To avoid this restriction, follow Java naming conventions recommending that package names start in lowercase and class names start in uppercase.

SQLJ Namespace

The SQLJ namespace refers to #sql class declarations and the portion of #sql executable statements outside the curly braces.

Note:

Restrictions particular to the naming of iterator columns are discussed in "Using Named Iterators".

Avoid using the following SQLJ reserved words as class names for declared connection context classes or iterator classes, in with or implements clauses, or in iterator column type declaration lists:

  • iterator

  • context

  • with

For example, do not have an iterator class or instance called iterator or a connection context class or instance called context.

However, note that it is permissible to have a stored function return variable whose name is any of these words.

SQL Namespace

The SQL namespace refers to the portion of a SQLJ executable statement inside the curly braces. Standard SQL naming restrictions apply here.

However, note that host expressions follow rules of the Java namespace, not the SQL namespace. This applies to the name of a host variable and to everything between the outer parentheses of a host expression.

File Name Requirements and Restrictions

SQLJ source files have the .sqlj file name extension. If the source file declares a public class (maximum of one), then the base name of the file must match the name of this class (case-sensitive). If the source file does not declare a public class, then the file name must still be a legal Java identifier, and it is recommended that the file name match the name of the first defined class.

For example, if you define the public class MySource in your source file, then your file name must be:

MySource.sqlj

Note:

These file naming requirements follow the Java Language Specification (JLS) and are not SQLJ-specific. These requirements do not directly apply in Oracle Database 11g, but it is still advisable to adhere to them.

Considerations for SQLJ in the Middle Tier

There are special considerations if you run SQLJ in the middle tier, such as in an Oracle9i Application Server Containers for J2EE (OC4J) environment.

Oracle JDBC drivers provide Oracle-specific interfaces in the oracle.jdbc package. The Oracle SQLJ libraries runtime12.jar and runtime12ee.jar make full use of these interfaces, but these libraries are not compatible with Oracle JDBC implementations prior to Oracle9i Application Server.

In Oracle9i Application Server, connections are established through data sources, which typically return instances of the oracle.jdbc.OracleConnection interface instead of the older oracle.jdbc.driver.OracleConnection class. This is necessary for certain connection functionality, such as distributed transactions (XA). To support such features, connection objects must implement the new interface.

This has the following consequences, relevant in an Oracle9i Application Server middle-tier environment, or any situation where data sources are used:

  • For maximum portability and flexibility of your code, use oracle.jdbc.OracleXXX types instead of oracle.jdbc.driver.OracleXXX types.

  • For custom Java types (typically for SQL objects and collections), implement oracle.sql.ORAData instead of the deprecated oracle.sql.CustomDatum interface.

For general information about SQLJ support for data sources and connection JavaBeans, refer to the following sections: