Unlike a wrapper, which adds another layer of execution, a call spec simply publishes the existence of a Java method. So, when you call the method (through its call spec), the run-time system dispatches the call with minimal overhead.

The run-time contexts for Java Stored Procedures are:

• Functions and Procedures
• Database Triggers
• Object-Relational Methods

Method main() Client-based Java applications declare a single, top-level method (main()) that defines the profile of an application. As with applets, server-based applications have no such "inner loop". Instead, they are driven by logically independent clients. Each client begins a session, calls its server-side logic modules through top-level entry points, and eventually ends the session. The server environment hides the managing of sessions, networks, and other shared resources from hosted Java programs.

The GUI A server cannot provide GUIs, but it can supply the logic that drives them. For example, the JServer JVM does not supply the basic GUI components found in the JDK's Abstract Windowing Toolkit (AWT). However, all AWT Java classes are available within the server environment. So, your programs can use AWT functionality, as long as they do not attempt to materialize a GUI on the server.

The IDE The JServer JVM is oriented to Java application deployment, not development. You can write and unit-test applications in your favorite IDE, then deploy them for execution within the RDBMS. Java's binary compatibility allows you to work in any IDE, then upload Java class files to the server. You need not move your Java source files to the database. Instead, you can use powerful client-side IDEs to maintain Java applications that are deployed on the server.

Multi-threading Multi-threaded Java programs execute on the Oracle8i database server without modification. However, in a server environment, Java threads do not increase concurrency (and therefore throughput). Throughput is affected only by MTS mode, the number of OS processes used by the RDBMS, and various tuning methods.

Before porting a multi-threaded application to the server, make sure you understand how threads work with the JServer JVM. The important differences are that on the server:

• Threads run sequentially, not concurrently.
• Threads within a call die when the call ends.
• Threads are cooperative, not preemptive, so if one thread enters an infinite loop, no other threads can run.

The JServer JVM embeds the standard Java namespace in RDBMS schemas. This feature lets Java programs access Java objects stored in Oracle databases and application servers across the enterprise. In addition, the JVM is tightly integrated with the scalable, shared memory architecture of the RDBMS. Java programs use call, session, and object lifetimes efficiently without your intervention. So, you can scale JServer and middle-tier Java business objects, even when they have session-long state.

This section briefly describes the main components of the JServer JVM and some of the facilities they provide.

Library Manager To store Java classes in an Oracle database, you use the command-line utility loadjava, which employs SQL CREATE JAVA statements to do its work. When invoked by the CREATE JAVA {SOURCE | CLASS | RESOURCE} statement, the library manager loads Java source, class, or resource files into the database. You never access these Java schema objects directly; only the JServer JVM uses them.

Memory Manager Automated storage management is one of Java's key features. In particular, the Java run-time system requires automatic garbage collection (deallocation of memory held by unused objects). The memory manager uses memory allocation techniques tuned to object lifetimes. Objects that survive beyond call boundaries are migrated to appropriate memory areas. Also, the memory manager minimizes the footprint per session by sharing immutable object state such as class definitions and final static variables.

Compiler The JServer JVM includes a standard Java 2 (also known as JDK 1.2) Java compiler. When invoked by the CREATE JAVA SOURCE statement, it translates Java source files into architecture-neutral, one-byte instructions known as bytecodes. Each bytecode consists of an opcode followed by its operands. The resulting Java class files, which conform fully to the Java standard, are submitted to the interpreter at run time.

Interpreter To execute Java programs, the JServer JVM includes a standard Java 2 bytecode interpreter. The interpreter and associated Java run-time system execute standard Java class files. For high throughput, the interpreter runs on the Multi-Threaded Server, which manages sessions and schedules the execution of Java programs. The run-time system supports native methods and call-in/call-out from the host environment.

Class Loader In response to requests from the run-time system, the Java class loader locates, loads, and initializes Java classes stored in the database. The class loader reads the class, then generates the data structures needed to execute it. Immutable data and metadata are loaded into initialize-once shared memory. As a result, less memory is required per session. The class loader attempts to resolve external references when necessary. Also, it invokes the Java compiler automatically when Java class files must be recompiled (and the source files are available).

Verifier Java class files are fully portable and conform to a well-defined format. The verifier prevents the inadvertent use of "spoofed" Java class files, which might alter program flow or violate access restrictions. Oracle security and Java security work with the verifier to protect your applications and data.

Note: Although your own code is interpreted, the JServer JVM uses natively compiled versions of the core Java class libraries, object request broker (ORB), SQLJ translator, and JDBC drivers.

Server-Side JDBC Internal Driver JDBC is a standard set of Java classes providing vendor-independent access to relational data. Specified by Sun Microsystems and modeled after ODBC (Open Database Connectivity) and the X/Open SQL CLI (Call Level Interface), the JDBC classes supply standard features such as simultaneous connections to several databases, transaction management, simple queries, calls to stored procedures, and streaming access to LONG column data.

Using low-level entry points, a specially tuned JDBC driver runs directly inside the RDBMS, thereby providing the fastest access to Oracle data from Java stored procedures. The server-side internal JDBC driver complies fully with the Sun Microsystems JDBC specification. Tightly integrated with the RDBMS, it supports Oracle-specific datatypes, NLS character sets, and stored procedures. Additionally, the client-side and server-side JDBC APIs are the same, which makes it easy to partition applications.

Server-Side SQLJ Translator SQLJ enables you to embed SQL statements in Java programs. It is more concise than JDBC and more amenable to static analysis and type checking. The SQLJ preprocessor, itself a Java program, takes as input a Java source file in which SQLJ clauses are embedded. Then, it translates the SQLJ clauses into Java class definitions that implement the specified SQL statements. The Java type system ensures that objects of those classes are called with the correct arguments.

A highly optimized SQLJ translator runs directly inside the RDBMS, where it provides run-time access to Oracle data using the server-side internal JDBC driver. SQLJ forms can include queries, DML, DDL, transaction control statements, and calls to stored procedures. The client-side and server-side SQLJ APIs are identical, which makes it easy to partition applications.

JServer Accelerator The JServer Accelerator is a native-code compiler that speeds up the execution of Java programs by eliminating interpreter overhead. It translates standard Java class files into specialized C source files that are processed by a platform-dependent C compiler into shared libraries, which the JServer JVM can load dynamically. Unlike just-in-time (JIT) compilers, which rely on processor-specific code, the JServer Accelerator is portable to all OS and hardware platforms. To speed up your applications, the JServer JVM has natively compiled versions of the core Java class libraries, ORB, SQLJ translator, and JDBC drivers.

Although the Java programs you load into the database are interpreted, they use natively compiled facilities. In addition, the core JDK classes and supplied Oracle classes that the programs use are natively compiled. As Figure the figure below shows, natively compiled code executes up to ten times faster than interpreted code. So, the more native code your program uses, the faster it executes.

Use your favorite Java IDE to create classes, or simply reuse existing classes that meet your needs. Oracle's Java facilities support many Java development tools and client-side programmatic interfaces. For example, the JServer JVM accepts programs developed in popular Java IDEs such as Symantec's Visual Cafe, Oracle's JDeveloper, and Borland's JBuilder. In the example below, you create the public class Oscar. It has a single method named quote(), which returns a quotation from Oscar Wilde.

    public class Oscar {
      // return a quotation from Oscar Wilde
      public static String quote() {
        return "I can resist everything except temptation.";
      }
    }

    javac Oscar.java

Step 2: Load and Resolve the Java Classes

Using the utility loadjava, you can upload Java source, class, and resource files into an Oracle database, where they are stored as Java schema objects. You can run loadjava from the command line or from an application, and you can specify several options including a resolver. In the example below, loadjava connects to the database using the default JDBC OCI driver. You must specify the username and password. By default, class Oscar is loaded into the logon schema (in this case, scott).

    % loadjava -user scott/tiger Oscar.class

Step 3: Publish the Java Classes

For each Java method callable from SQL, you must write a call spec, which exposes the method's top-level entry point to Oracle. Typically, only a few call specs are needed, but if you like, Oracle's JDeveloper can generate them for you. In the following example, from SQL*Plus, you connect to the database, then define a top-level call spec for method quote():

    SQL> connect scott/tiger
    SQL> CREATE FUNCTION oscar_quote RETURN VARCHAR2
    2 AS LANGUAGE JAVA
    3 NAME 'Oscar.quote() return java.lang.String';

Step 4: Call the Stored Procedures

You can call Java stored procedures from SQL DML statements, PL/SQL blocks, and PL/SQL subprograms. Using the SQL CALL statement, you can also call them from the top level (from SQL*Plus, for example) and from database triggers. In the following example, you declare a SQL*Plus host variable:

    SQL> VARIABLE theQuote VARCHAR2(50);

    SQL> CALL oscar_quote() INTO :theQuote;
    SQL> PRINT theQuote;

    THEQUOTE
    --------------------------------------------------
    I can resist everything except temptation.

Step 5: If Necessary, Debug the Stored Procedures

Your Java stored procedures execute remotely on a server, which typically resides on a separate machine. However, the JDK debugger (jdb) cannot debug remote Java programs, so JServer provides a way to debug them. The class DebugProxy makes remote Java programs appear to be local. It lets any debugger that supports the sun.tools.debug.Agent protocol connect to a program as if the program were local. The proxy forwards requests to the server and returns results to the debugger.