[Top] [Contents] [Index] [ ? ]

GNU Classpath Hacker's Guide

This file contains important information you will need to know if you are going to hack on the GNU Classpath project code.

Copyright (C) 1998,1999,2000,2001,2002,2003,2004,2005,2007 Free Software Foundation, Inc.

This document contains important information you'll want to know if you want to hack on GNU Classpath, Essential Libraries for Java, to help create free core class libraries for use with virtual machines and compilers for the java programming language.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

1. Introduction

The GNU Classpath Project is dedicated to providing a 100% free, clean room implementation of the standard core class libraries for compilers and runtime environments for the java programming language. It offers free software developers an alternative core library implementation upon which larger java-like programming environments can be built. The GNU Classpath Project was started in the Spring of 1998 as an official Free Software Foundation project. Most of the volunteers working on GNU Classpath do so in their spare time, but a couple of projects based on GNU Classpath have paid programmers to improve the core libraries. We appreciate everyone's efforts in the past to improve and help the project and look forward to future contributions by old and new members alike.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

2. Requirements

Although GNU Classpath is following an open development model where input from developers is welcome, there are certain base requirements that need to be met by anyone who wants to contribute code to this project. They are mostly dictated by legal requirements and are not arbitrary restrictions chosen by the GNU Classpath team.

You will need to adhere to the following things if you want to donate code to the GNU Classpath project:

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

3. Volunteering to Help

The GNU Classpath project needs volunteers to help us out. People are needed to write unimplemented core packages, to test GNU Classpath on free software programs written in the java programming language, to test it on various platforms, and to port it to platforms that are currently unsupported.

While pretty much all contributions are welcome (but see see section Requirements) it is always preferable that volunteers do the whole job when volunteering for a task. So when you volunteer to write a Java package, please be willing to do the following:

Writing good documentation, tests and fixing bugs should be every developer's top priority in order to reach the elusive release of version 1.0.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

4. Project Goals

The goal of the Classpath project is to produce a free implementation of the standard class library for Java. However, there are other more specific goals as to which platforms should be supported.

Classpath is targeted to support the following operating systems:

  1. Free operating systems. This includes GNU/Linux, GNU/Hurd, and the free BSDs.
  2. Other Unix-like operating systems.
  3. Platforms which currently have no Java support at all.
  4. Other platforms such as MS-Windows.

While free operating systems are the top priority, the other priorities can shift depending on whether or not there is a volunteer to port Classpath to those platforms and to test releases.

Eventually we hope the Classpath will support all JVMs that provide JNI or CNI support. However, the top priority is free JVMs. A small list of Compiler/VM environments that are currently actively incorporating GNU Classpath is below. A more complete overview of projects based on GNU classpath can be found online at the GNU Classpath stories page.

  1. GCJ
  2. jamvm
  3. cacao
  4. Jikes RVM
  5. Kaffe
  6. IKVM

As with OS platform support, this priority list could change if a volunteer comes forward to port, maintain, and test releases for a particular JVM. Since gcj is part of the GNU Compiler Collective it is one of the most important targets. But since it doesn't currently work out of the box with GNU Classpath it is not the easiest target. When hacking on GNU Classpath the easiest solution is to use compilers and runtime environments that work out of the box with it, such as the Eclipse compiler, ecj, and the runtime environments jamvm and cacao. Both Jikes RVM and Kaffe use an included version of GNU Classpath by default, but Kaffe can now use a pre-installed version and Jikes RVM supports using a CVS snapshot as well as the latest release. Working directly with targets such as Jikes RVM, gcj and IKVM is possible but can be a little more difficult as changes have to be merged back into GNU Classpath proper, which requires additional work. Due to a recent switch to the use of 1.5 language features within GNU Classpath, a compiler compatible with these features is required. At present, this includes the Eclipse compiler, ecj, and the OpenJDK compiler.

GNU Classpath currently implements the majority of the 1.4 and 1.5 APIs (binary compatibility is above 95% for both, but does not take into account internal implementations of features such as graphic and sound support). There is support for some 1.6 APIs but this is still nascent. Please do not create classes that depend on features in other packages unless GNU Classpath already contains those features. GNU Classpath has been free of any proprietary dependencies for a long time now and we like to keep it that way. Finishing, polishing up, documenting, testing and debugging current functionality is of higher priority then adding new functionality.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

5. Needed Tools and Libraries

If you want to hack on Classpath, you should at least download and install the following tools and try to familiarize yourself with them. In most cases having these tools installed will be all you really need to know about them. Also note that when working on (snapshot) releases only a 1.5 compiler (plus a free VM from the list above and the libraries listed below) is required. The other tools are only needed when working directly on the CVS version.

All of these tools are available from gnudist.gnu.org via anonymous ftp, except CVS which is available from www.cvshome.org and the Eclipse Compiler for Java, which is available from www.eclipse.org/jdt/core.

Except for the Eclipse Compiler for Java, they are fully documented with texinfo manuals. Texinfo can be browsed with the Emacs editor, or with the text editor of your choice, or transformed into nicely printable Postscript.

Here is a brief description of the purpose of those tools.


GNU make ("gmake") is required for building Classpath.


The GNU Compiler Collection. This contains a C compiler (gcc) for compiling the native C code and a compiler for the java programming language (gcj). You will need at least gcc version 2.95 or higher in order to compile the native code. There is currently no released version of gcj that can compile the Java 1.5 programming language used by GNU Classpath.


The Eclipse Compiler for Java. This is a compiler for the Java 1.5 programming language. It translates source code to bytecode. The Eclipse Foundation makes "ecj.jar" available as the JDT Core Batch Compiler download.


A version control system that maintains a centralized Internet repository of all code in the Classpath system.


This tool automatically creates `Makefile.in' files from `Makefile.am' files. The `Makefile.in' is turned into a `Makefile' by autoconf.

Why use this? Because it automatically generates every makefile target you would ever want (`clean', `install', `dist', etc) in full compliance with the GNU coding standards. It also simplifies Makefile creation in a number of ways that cannot be described here. Read the docs for more info.


Automatically configures a package for the platform on which it is being built and generates the Makefile for that platform.


Handles all of the zillions of hairy platform specific options needed to build shared libraries.


The free GNU replacement for the standard Unix macro processor. Proprietary m4 programs are broken and so GNU m4 is required for autoconf to work though knowing a lot about GNU m4 is not required to work with autoconf.


Larry Wall's scripting language. It is used internally by automake.


Manuals and documentation (like this guide) are written in texinfo. Texinfo is the official documentation format of the GNU project. Texinfo uses a single source file to produce output in a number of formats, both online and printed (dvi, info, html, xml, etc.). This means that instead of writing different documents for online information and another for a printed manual, you need write only one document. And when the work is revised, you need revise only that one document.

For any build environment involving native libraries, recent versions of autoconf, automake, and libtool are required if changes are made that require rebuilding `configure', `Makefile.in', `aclocal.m4', or `config.h.in'.

When working from CVS you can run those tools by executing autogen.sh in the source directory.

For building the Java bytecode (.class files), you can select which compiler should be employed using `--with-javac' or `--with-ecj' as an argument to configure; the present default is ecj if found.

Instead of ecj, you can also use javac, which is available at openjdk.dev.java.net/compiler.

For compiling the native AWT libraries you need to have the following libraries installed (unless `--disable-gtk-peer' is used as an argument to configure):

GTK+ 2.8.x

GTK+ is a multi-platform toolkit for creating graphical user interfaces. It is used as the basis of the GNU desktop project GNOME.


gdk-pixbuf is a GNOME library for representing images.


www.x.org hosts the XTest Extension (libXtst). It is necessary for GdkRobot support in java.awt.

There is a bug in earlier versions of at-spi, atk, and gail, which are used for GNOME accessibility. Prior to version 1.18.0 of these packages, gtk graphical applications should be run without accessibility (clear the GTK_MODULES environment variable).

For building the Qt AWT peer JNI native libraries you have to specify `--enable-qt-peer' and need the following library:


Qt version 4.0.1 or higher. The Qt library is a cross-platform graphics toolkit.

Please note that at the moment most operating systems do not ship Qt version 4.0.1 by default. We recommend using GNU Classpath' Qt support only for its developers and bug reporters. See the wiki for details on how to get it to work.

For building the X AWT peers you have to specify where to find the Escher library on your system using the `--with-escher=ABS.PATH' option. You will need the following library:


Escher version 0.2.3 or higher. The Escher library is an implementation of X protocol and associated libraries written in the Java programming language.

For building the ALSA midi provider code you will need the following library:


ALSA libraries.

The ALSA project provides sound device drivers and associated libraries for the Linux kernel.

Building the ALSA midi provider code can be disabled by passing `--disable-alsa' to configure.

For building the DSSI midi synthesizer provider code you will need the following libraries:


DSSI library for audio processing plugins.


liblo, the Lightweight OSC implementation.


LADSPA, the Linux Audio Developer's Simple Plugin API.


JACK, a low latency audio server.


libsndfile, an audio file I/O library.


fluidsynth, a real-time SoundFont 2 based soft-synth.

The GConf-based backend for java.util.prefs needs the following library headers:


GConf version 2.6.0 (or higher). GConf is used for storing desktop and application configuration settings in GNOME.

The GStreamer backend for javax.sound.sampled (The Java Sound API, not including the MIDI portion) needs the following library headers:


GStreamer version 0.10.10 (or higher). You will also need at least gstreamer-base and gstreamer-plugins-base. More plugins can be used to allow streaming of different sound types but are not a compile time requirement. See README.gstreamer in the source distribution for more informations.

For building gcjwebplugin you'll need the Mozilla plugin support headers and libraries, which are available at www.mozilla.org.

For enabling the com.sun.tools.javac support in tools.zip you will need a jar file containing the Eclipse Java Compiler. Otherwise com.sun.tools.javac will not be included in `tools.zip'.

For building the xmlj JAXP implementation (disabled by default, use configure --enable-xmlj) you need the following libraries:


libxml2 version 2.6.8 or higher.

The libxml2 library is the XML C library for the Gnome desktop.


libxslt version 1.1.11 or higher.

The libxslt library if the XSLT C library for the Gnome desktop.

GNU Classpath comes with a couple of libraries included in the source that are not part of GNU Classpath proper, but that have been included to provide certain needed functionality. All these external libraries should be clearly marked as such. In general we try to use as much as possible the clean upstream versions of these sources. That way merging in new versions will be easier. You should always try to get bug fixes to these files accepted upstream first. Currently we include the following 'external' libraries. Most of these sources are included in the `external' directory. That directory also contains a `README' file explaining how to import newer versions.

JSR166 concurrency support

Can be found in `external/jsr166'. Provides java.util.concurrent and its subpackages. Upstream is Doug Lea's Concurrency Interest Site.

RelaxNG Datatype Interfaces

Can be found in `external/relaxngDatatype'. Provides org.relaxng.datatype and its subpackages. Upstream is http://www.oasis-open.org/committees/relax-ng/.

Simple API for XML (SAX)

Can be found in `external/sax'. Provides org.xml.sax and its subpackages. Upstream is http://www.saxproject.org.

Document Object Model (DOM) bindings

Can be found in `external/w3c_dom'. Provides org.w3c.dom and its subpackages. Upstream locations are listed in `external/w3c_dom/README'.


Can be found in `native/fdlibm'. Provides native implementations of some of the Float and Double operations. Upstream is libgcj, they sync again with the 'real' upstream http://www.netlib.org/fdlibm/readme. See also java.lang.StrictMath.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

6. Installation instructions

This package was designed to use the GNU standard for configuration and makefiles. To build and install do the following:

  1. Configuration

    Run the configure script to configure the package. There are various options you might want to pass to configure to control how the package is built. Consider the following options, configure --help gives a complete list.


    compile Java source (default=`yes').


    compile JNI source (default=`yes').


    compile GTK native peers (default=`yes').


    compile Qt4 native peers (default=`no').


    fully qualified class name of default AWT toolkit (default=`no').


    compile native libxml/xslt library (default=`no').


    enable to use JNI native methods (default=`yes').


    enable build of local Unix sockets.


    define what to install `(zip|flat|both|none)' (default=`zip').


    enable build of the X/Escher peers, with the escher library at `/path/to/escher', either in the form of a JAR file, or a directory containing the .class files of Escher.


    whether to compile C code with `-Werror' which turns any compiler warning into a compilation failure (default=`no').


    generate documentation using gjdoc (default=`no').


    Regenerate the parsers with jay, must be given the path to the jay executable


    use prebuilt glibj.zip class library


    specify jar file containing the Eclipse Java Compiler


    build the experimental GStreamer peer (see `README.gstreamer')

    For more flags run configure --help.

  2. Building

    Type gmake to build the package. There is no longer a dependency problem and we aim to keep it that way.

  3. Installation

    Type gmake install to install everything. This may require being the superuser. The default install path is /usr/local/classpath you may change it by giving configure the `--prefix=<path>' option.

Report bugs to classpath@gnu.org or much better to the GNU Classpath bug tracker at Savannah.

Happy Hacking!

Once installed, GNU Classpath is ready to be used by any VM that supports using the official version of GNU Classpath. Simply ensure that `/usr/local/classpath/share/classpath' is in your CLASSPATH environment variable. You'll also have to set your LD_LIBRARY_PATH variable (or similar system configuration) to include the Classpath native libraries in `/usr/local/classpath/lib/classpath'.

*NOTE* All example paths assume the default prefix is used with configure. If you don't know what this means then the examples are correct.


More information about the VMs that use GNU Classpath can be found in the `README' file.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

7. Building and running with the X AWT peers

In order build the X peers you need the Escher library version 0.2.3 from escher.sourceforge.net. Unpack (and optionally build) the Escher library following the instructions in the downloaded package. Enable the build of the X peers by passing `--with-escher=/path/to/escher' to configure where `/path/to/escher' either points to a directory structure or JAR file containing the Escher classes. For Unix systems it is preferable to also build local socket support by passing `--enable-local-sockets', which accelerates the network communication to the X server significantly.

In this release you have to enable the X peers at runtime by setting the system property awt.toolkit=gnu.java.awt.peer.x.XToolkit by passing `-Dawt.toolkit=gnu.java.awt.peer.x.XToolkit' to the java command when running an application.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

8. Misc. Notes

Compilation is accomplished using a compiler's @file syntax. For our part, we avoid placing make style dependencies as rules upon the compilation of a particular class file and leave this up to the Java compiler instead.

The `--enable-maintainer-mode' option to configure currently does very little and shouldn't be used by ordinary developers or users anyway.

On Windows machines, the native libraries do not currently build, but the Java bytecode library will. GCJ trunk is beginning to work under Cygwin.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

9. Programming Standards

For C source code, follow the GNU Coding Standards. The standards also specify various things like the install directory structure. These should be followed if possible.

For Java source code, please follow the GNU Coding Standards, as much as possible. There are a number of exceptions to the GNU Coding Standards that we make for GNU Classpath as documented in this guide. We will hopefully be providing developers with a code formatting tool that closely matches those rules soon.

For API documentation comments, please follow How to Write Doc Comments for Javadoc. We would like to have a set of guidelines more tailored to GNU Classpath as part of this document.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

9.1 Java source coding style

Here is a list of some specific rules used when hacking on GNU Classpath java source code. We try to follow the standard GNU Coding Standards for that. There are lots of tools that can automatically generate it (although most tools assume C source, not java source code) and it seems as good a standard as any. There are a couple of exceptions and specific rules when hacking on GNU Classpath java source code however. The following lists how code is formatted (and some other code conventions):

Some things are the same as in the normal GNU Coding Standards:

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

10. Working on the code, Working with others

There are a lot of people helping out with GNU Classpath. Here are a couple of practical guidelines to make working together on the code smoother.

The main thing is to always discuss what you are up to on the mailinglist. Making sure that everybody knows who is working on what is the most important thing to make sure we cooperate most effectively.

We maintain a Task List which contains items that you might want to work on.

Before starting to work on something please make sure you read this complete guide. And discuss it on list to make sure your work does not duplicate or interferes with work someone else is already doing. Always make sure that you submit things that are your own work. And that you have paperwork on file (as stated in the requirements section) with the FSF authorizing the use of your additions.

Technically the GNU Classpath project is hosted on Savannah a central point for development, distribution and maintenance of GNU Software. Here you will find the project page, bug reports, pending patches, links to mailing lists, news items and CVS.

You can find instructions on getting a CVS checkout for classpath at https://savannah.gnu.org/cvs/?group=classpath.

You don't have to get CVS commit write access to contribute, but it is sometimes more convenient to be able to add your changes directly to the project CVS. Please contact the GNU Classpath savannah admins to arrange CVS access if you would like to have it.

Make sure to be subscribed to the commit-classpath mailinglist while you are actively hacking on Classpath. You have to send patches (cvs diff -uN) to this list before committing.

We really want to have a pretty open check-in policy. But this means that you should be extra careful if you check something in. If at all in doubt or if you think that something might need extra explaining since it is not completely obvious please make a little announcement about the change on the mailinglist. And if you do commit something without discussing it first and another GNU Classpath hackers asks for extra explanation or suggests to revert a certain commit then please reply to the request by explaining why something should be so or if you agree to revert it. (Just reverting immediately is OK without discussion, but then please don't mix it with other changes and please say so on list.)

Patches that are already approved for libgcj or also OK for Classpath. (But you still have to send a patch/diff to the list.) All other patches require you to think whether or not they are really OK and non-controversial, or if you would like some feedback first on them before committing. We might get real commit rules in the future, for now use your own judgement, but be a bit conservative.

Always contact the GNU Classpath maintainer before adding anything non-trivial that you didn't write yourself and that does not come from libgcj or from another known GNU Classpath or libgcj hacker. If you have been assigned to commit changes on behalf of another project or a company always make sure they come from people who have signed the papers for the FSF and/or fall under the arrangement your company made with the FSF for contributions. Mention in the ChangeLog who actually wrote the patch.

Commits for completely unrelated changes they should be committed separately (especially when doing a formatting change and a logical change, do them in two separate commits). But do try to do a commit of as much things/files that are done at the same time which can logically be seen as part of the same change/cleanup etc.

When the change fixes an important bug or adds nice new functionality please write a short entry for inclusion in the `NEWS' file. If it changes the VM interface you must mention that in both the `NEWS' file and the VM Integration Guide.

All the "rules" are really meant to make sure that GNU Classpath will be maintainable in the long run and to give all the projects that are now using GNU Classpath an accurate view of the changes we make to the code and to see what changed when. If you think the requirements are "unworkable" please try it first for a couple of weeks. If you still feel the same after having some more experience with the project please feel free to bring up suggestions for improvements on the list. But don't just ignore the rules! Other hackers depend on them being followed to be the most productive they can be (given the above constraints).

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

10.1 Working with branches

Sometimes it is necessary to create branch of the source for doing new work that is disruptive to the other hackers, or that needs new language or libraries not yet (easily) available.

After discussing the need for a branch on the main mailinglist with the other hackers explaining the need of a branch and suggestion of the particular branch rules (what will be done on the branch, who will work on it, will there be different commit guidelines then for the mainline trunk and when is the branch estimated to be finished and merged back into the trunk) every GNU Classpath hacker with commit access should feel free to create a branch. There are however a couple of rules that every branch should follow:

If any of these rules are unclear please discuss on the list first.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

10.2 Documenting what changed when with ChangeLog entries

To keep track of who did what when we keep an explicit ChangeLog entry together with the code. This mirrors the CVS commit messages and in general the ChangeLog entry is the same as the CVS commit message. This provides an easy way for people getting a (snapshot) release or without access to the CVS server to see what happened when. We do not generate the ChangeLog file automatically from the CVS server since that is not reliable.

A good ChangeLog entry guideline can be found in the Guile Manual at http://www.gnu.org/software/guile/changelogs/guile-changelogs_3.html.

Here are some example to explain what should or shouldn't be in a ChangeLog entry (and the corresponding commit message):

The above are all just guidelines. We all appreciate the fact that writing ChangeLog entries, using a coding style that is not "your own" and the CVS, patch and diff tools do take some time to getting used to. So don't feel like you have to do it perfect right away or that contributions aren't welcome if they aren't "perfect". We all learn by doing and interacting with each other.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

11. Programming Goals

When you write code for Classpath, write with three things in mind, and in the following order: portability, robustness, and efficiency.

If efficiency breaks portability or robustness, then don't do it the efficient way. If robustness breaks portability, then bye-bye robust code. Of course, as a programmer you would probably like to find sneaky ways to get around the issue so that your code can be all three ... the following chapters will give some hints on how to do this.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

11.1 Portability

The portability goal for Classpath is the following:

  1. native functions for each platform that work across all VMs on that platform
  2. a single classfile set that work across all VMs on all platforms that support the native functions.

For almost all of Classpath, this is a very feasible goal, using a combination of JNI and native interfaces. This is what you should shoot for. For those few places that require knowledge of the Virtual Machine beyond that provided by the Java standards, the VM Interface was designed. Read the Virtual Machine Integration Guide for more information.

Right now the only supported platform is Linux. This will change as that version stabilizes and we begin the effort to port to many other platforms. Jikes RVM runs Classpath on AIX, and generally the Jikes RVM team fixes Classpath to work on that platform.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

11.2 Utility Classes

At the moment, we are not very good at reuse of the JNI code. There have been some attempts, called libclasspath, to create generally useful utility classes. The utility classes are in the directory `native/jni/classpath' and they are mostly declared in `native/jni/classpath/jcl.h'. These utility classes are currently only discussed in Robustness and in Native Efficiency.

There are more utility classes available that could be factored out if a volunteer wants something nice to hack on. The error reporting and exception throwing functions and macros in `native/jni/gtk-peer/gthread-jni.c' might be good candidates for reuse. There are also some generally useful utility functions in `gnu_java_awt_peer_gtk_GtkMainThread.c' that could be split out and put into libclasspath.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

11.3 Robustness

Native code is very easy to make non-robust. (That's one reason Java is so much better!) Here are a few hints to make your native code more robust.

Always check return values for standard functions. It's sometimes easy to forget to check that malloc() return for an error. Don't make that mistake. (In fact, use JCL_malloc() in the jcl library instead-it will check the return value and throw an exception if necessary.)

Always check the return values of JNI functions, or call ExceptionOccurred to check whether an error occurred. You must do this after every JNI call. JNI does not work well when an exception has been raised, and can have unpredictable behavior.

Throw exceptions using JCL_ThrowException. This guarantees that if something is seriously wrong, the exception text will at least get out somewhere (even if it is stderr).

Check for null values of jclasses before you send them to JNI functions. JNI does not behave nicely when you pass a null class to it: it terminates Java with a "JNI Panic."

In general, try to use functions in `native/jni/classpath/jcl.h'. They check exceptions and return values and throw appropriate exceptions.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

11.4 Java Efficiency

For methods which explicitly throw a NullPointerException when an argument is passed which is null, per a Sun specification, do not write code like:

strlen (String foo) throws NullPointerException
  if (foo == null)
    throw new NullPointerException ("foo is null");
  return foo.length ();

Instead, the code should be written as:

strlen (String foo) throws NullPointerException
  return foo.length ();

Explicitly comparing foo to null is unnecessary, as the virtual machine will throw a NullPointerException when length() is invoked. Classpath is designed to be as fast as possible - every optimization, no matter how small, is important.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

11.5 Native Efficiency

You might think that using native methods all over the place would give our implementation of Java speed, speed, blinding speed. You'd be thinking wrong. Would you believe me if I told you that an empty interpreted Java method is typically about three and a half times faster than the equivalent native method?

Bottom line: JNI is overhead incarnate. In Sun's implementation, even the JNI functions you use once you get into Java are slow.

A final problem is efficiency of native code when it comes to things like method calls, fields, finding classes, etc. Generally you should cache things like that in static C variables if you're going to use them over and over again. GetMethodID(), GetFieldID(), and FindClass() are slow. Classpath provides utility libraries for caching methodIDs and fieldIDs in `native/jni/classpath/jnilink.h'. Other native data can be cached between method calls using functions found in `native/jni/classpath/native_state.h'.

Here are a few tips on writing native code efficiently:

Make as few native method calls as possible. Note that this is not the same thing as doing less in native method calls; it just means that, if given the choice between calling two native methods and writing a single native method that does the job of both, it will usually be better to write the single native method. You can even call the other two native methods directly from your native code and not incur the overhead of a method call from Java to C.

Cache jmethodIDs and jfieldIDs wherever you can. String lookups are expensive. The best way to do this is to use the `native/jni/classpath/jnilink.h' library. It will ensure that jmethodIDs are always valid, even if the class is unloaded at some point. In 1.1, jnilink simply caches a NewGlobalRef() to the method's underlying class; however, when 1.2 comes along, it will use a weak reference to allow the class to be unloaded and then re-resolve the jmethodID the next time it is used.

Cache classes that you need to access often. jnilink will help with this as well. The issue here is the same as the methodID and fieldID issue-how to make certain the class reference remains valid.

If you need to associate native C data with your class, use Paul Fisher's native_state library (NSA). It will allow you to get and set state fairly efficiently. Japhar now supports this library, making native state get and set calls as fast as accessing a C variable directly.

If you are using native libraries defined outside of Classpath, then these should be wrapped by a Classpath function instead and defined within a library of their own. This makes porting Classpath's native libraries to new platforms easier in the long run. It would be nice to be able to use Mozilla's NSPR or Apache's APR, as these libraries are already ported to numerous systems and provide all the necessary system functions as well.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

11.6 Security

Security is such a huge topic it probably deserves its own chapter. Most of the current code needs to be audited for security to ensure all of the proper security checks are in place within the Java platform, but also to verify that native code is reasonably secure and avoids common pitfalls, buffer overflows, etc. A good source for information on secure programming is the excellent HOWTO by David Wheeler, Secure Programming for Linux and Unix HOWTO.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

12. API Compatibility

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

12.1 Serialization

Sun has produced documentation concerning much of the information needed to make Classpath serializable compatible with Sun implementations. Part of doing this is to make sure that every class that is Serializable actually defines a field named serialVersionUID with a value that matches the output of serialver on Sun's implementation. The reason for doing this is below.

If a class has a field (of any accessibility) named serialVersionUID of type long, that is what serialver uses. Otherwise it computes a value using some sort of hash function on the names of all method signatures in the .class file. The fact that different compilers create different synthetic method signatures, such as access$0() if an inner class needs access to a private member of an enclosing class, make it impossible for two distinct compilers to reliably generate the same serial #, because their .class files differ. However, once you have a .class file, its serial # is unique, and the computation will give the same result no matter what platform you execute on.

Serialization compatibility can be tested using tools provided with Japitools. These tools can test binary serialization compatibility and also provide information about unknown serialized formats by writing these in XML instead. Japitools is also the primary means of checking API compatibility for GNU Classpath with Sun's Java Platform.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

12.2 Deprecated Methods

Sun has a practice of creating "alias" methods, where a public or protected method is deprecated in favor of a new one that has the same function but a different name. Sun's reasons for doing this vary; as an example, the original name may contain a spelling error or it may not follow Java naming conventions.

Unfortunately, this practice complicates class library code that calls these aliased methods. Library code must still call the deprecated method so that old client code that overrides it continues to work. But library code must also call the new version, because new code is expected to override the new method.

The correct way to handle this (and the way Sun does it) may seem counterintuitive because it means that new code is less efficient than old code: the new method must call the deprecated method, and throughout the library code calls to the old method must be replaced with calls to the new one.

Take the example of a newly-written container laying out a component and wanting to know its preferred size. The Component class has a deprecated preferredSize method and a new method, getPreferredSize. Assume that the container is laying out an old component that overrides preferredSize and a new component that overrides getPreferredSize. If the container calls getPreferredSize and the default implementation of getPreferredSize calls preferredSize, then the old component will have its preferredSize method called and new code will have its getPreferredSize method called.

Even using this calling scheme, an old component may still be laid out improperly if it implements a method, getPreferredSize, that has the same signature as the new Component.getPreferredSize. But that is a general problem - adding new public or protected methods to a widely-used class that calls those methods internally is risky, because existing client code may have already declared methods with the same signature.

The solution may still seem counterintuitive - why not have the deprecated method call the new method, then have the library always call the old method? One problem with that, using the preferred size example again, is that new containers, which will use the non-deprecated getPreferredSize, will not get the preferred size of old components.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

13. Specification Sources

There are a number of specification sources to use when working on Classpath. In general, the only place you'll find your classes specified is in the JavaDoc documentation or possibly in the corresponding white paper. In the case of java.lang, java.io and java.util, you should look at the Java Language Specification.

Here, however, is a list of specs, in order of canonicality:

  1. Clarifications and Amendments to the JLS - 1.1
  2. JLS Updates - 1.1
  3. The 1.0 JLS
  4. JVM spec - 1.1
  5. JNI spec - 1.1
  6. Sun's javadoc - 1.1 (since Sun's is the reference implementation, the javadoc is documentation for the Java platform itself.)
  7. JVMDI spec - 1.2, JNI spec - 1.2 (sometimes gives clues about unspecified things in 1.1; if it was not specified accurately in 1.1, then use the spec for 1.2; also, we are using JVMDI in this project.)
  8. Sun's javadoc - 1.2 (sometimes gives clues about unspecified things in 1.1; if it was not specified accurately in 1.1, then use the spec for 1.2)
  9. The Bug Parade: I have obtained a ton of useful information about how things do work and how they *should* work from the Bug Parade just by searching for related bugs. The submitters are very careful about their use of the spec. And if something is unspecified, usually you can find a request for specification or a response indicating how Sun thinks it should be specified here.

You'll notice that in this document, white papers and specification papers are more canonical than the JavaDoc documentation. This is true in general.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

14. Directory and File Naming Conventions

The Classpath directory structure is laid out in the following manner:

 |       |
 |       |-->awt
 |       |-->io
 |       |-->lang
 |       |-->util
 |       |     |
 |       |     |--->zip
 |       |     |--->jar
 |       |-->net
 |       |-->etc
 |       |
 |       |-->java
 |             |
 |             |-->awt
 |             |-->lang
 |             |-->util
 |             |     |
 |             |     |-->zip
 |             |-->etc
         |    |-->classpath
         |    |-->gtk-peer
         |    |-->java-io
         |    |-->java-lang
         |    |-->java-net
         |    |-->java-util
         |    |-->etc

Here is a brief description of the toplevel directories and their contents.


Contains the source code to the Java packages that make up the core class library. Because this is the public interface to Java, it is important that the public classes, interfaces, methods, and variables are exactly the same as specified in Sun's documentation. The directory structure is laid out just like the java package names. For example, the class java.util.zip would be in the directory java-util.


Internal classes (roughly analogous to Sun's sun.* classes) should go under the `gnu/java' directory. Classes related to a particular public Java package should go in a directory named like that package. For example, classes related to java.util.zip should go under a directory `gnu/java/util/zip'. Sub-packages under the main package name are allowed. For classes spanning multiple public Java packages, pick an appropriate name and see what everybody else thinks.


This directory holds native code needed by the public Java packages. Each package has its own subdirectory, which is the "flattened" name of the package. For example, native method implementations for java.util.zip should go in `native/classpath/java-util'. Classpath actually includes an all Java version of the zip classes, so no native code is required.

Each person working on a package get's his or her own "directory space" underneath each of the toplevel directories. In addition to the general guidelines above, the following standards should be followed:

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

15. Character Conversions

Java uses the Unicode character encoding system internally. This is a sixteen bit (two byte) collection of characters encompassing most of the world's written languages. However, Java programs must often deal with outside interfaces that are byte (eight bit) oriented. For example, a Unix file, a stream of data from a network socket, etc. Beginning with Java 1.1, the Reader and Writer classes provide functionality for dealing with character oriented streams. The classes InputStreamReader and OutputStreamWriter bridge the gap between byte streams and character streams by converting bytes to Unicode characters and vice versa.

In Classpath, InputStreamReader and OutputStreamWriter rely on an internal class called gnu.java.io.EncodingManager to load translators that perform the actual conversion. There are two types of converters, encoders and decoders. Encoders are subclasses of gnu.java.io.encoder.Encoder. This type of converter takes a Java (Unicode) character stream or buffer and converts it to bytes using a specified encoding scheme. Decoders are a subclass of gnu.java.io.decoder.Decoder. This type of converter takes a byte stream or buffer and converts it to Unicode characters. The Encoder and Decoder classes are subclasses of Writer and Reader respectively, and so can be used in contexts that require character streams, but the Classpath implementation currently does not make use of them in this fashion.

The EncodingManager class searches for requested encoders and decoders by name. Since encoders and decoders are separate in Classpath, it is possible to have a decoder without an encoder for a particular encoding scheme, or vice versa. EncodingManager searches the package path specified by the file.encoding.pkg property. The name of the encoder or decoder is appended to the search path to produce the required class name. Note that EncodingManager knows about the default system encoding scheme, which it retrieves from the system property file.encoding, and it will return the proper translator for the default encoding if no scheme is specified. Also, the Classpath standard translator library, which is the gnu.java.io package, is automatically appended to the end of the path.

For efficiency, EncodingManager maintains a cache of translators that it has loaded. This eliminates the need to search for a commonly used translator each time it is requested.

Finally, EncodingManager supports aliasing of encoding scheme names. For example, the ISO Latin-1 encoding scheme can be referred to as "8859_1" or "ISO-8859-1". EncodingManager searches for aliases by looking for the existence of a system property called gnu.java.io.encoding_scheme_alias.<encoding name>. If such a property exists. The value of that property is assumed to be the canonical name of the encoding scheme, and a translator with that name is looked up instead of one with the original name.

Here is an example of how EncodingManager works. A class requests a decoder for the "UTF-8" encoding scheme by calling EncodingManager.getDecoder("UTF-8"). First, an alias is searched for by looking for the system property gnu.java.io.encoding_scheme_alias.UTF-8. In our example, this property exists and has the value "UTF8". That is the actual decoder that will be searched for. Next, EncodingManager looks in its cache for this translator. Assuming it does not find it, it searches the translator path, which is this example consists only of the default gnu.java.io. The "decoder" package name is appended since we are looking for a decoder. ("encoder" would be used if we were looking for an encoder). Then name name of the translator is appended. So EncodingManager attempts to load a translator class called gnu.java.io.decoder.UTF8. If that class is found, an instance of it is returned. If it is not found, a UnsupportedEncodingException.

To write a new translator, it is only necessary to subclass Encoder and/or Decoder. Only a handful of abstract methods need to be implemented. In general, no methods need to be overridden. The needed methods calculate the number of bytes/chars that the translation will generate, convert buffers to/from bytes, and read/write a requested number of characters to/from a stream.

Many common encoding schemes use only eight bits to encode characters. Writing a translator for these encodings is very easy. There are abstract translator classes gnu.java.io.decode.DecoderEightBitLookup and gnu.java.io.encode.EncoderEightBitLookup. These classes implement all of the necessary methods. All that is necessary to create a lookup table array that maps bytes to Unicode characters and set the class variable lookup_table equal to it in a static initializer. Also, a single constructor that takes an appropriate stream as an argument must be supplied. These translators are exceptionally easy to create and there are several of them supplied in the Classpath distribution.

Writing multi-byte or variable-byte encodings is more difficult, but often not especially challenging. The Classpath distribution ships with translators for the UTF8 encoding scheme which uses from one to three bytes to encode Unicode characters. This can serve as an example of how to write such a translator.

Many more translators are needed. All major character encodings should eventually be supported.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

16. Localization

There are many parts of the Java standard runtime library that must be customized to the particular locale the program is being run in. These include the parsing and display of dates, times, and numbers; sorting words alphabetically; breaking sentences into words, etc. In general, Classpath uses general classes for performing these tasks, and customizes their behavior with configuration data specific to a given locale.

In Classpath, all locale specific data is stored in a ListResourceBundle class in the package gnu/java/locale. The basename of the bundle is LocaleInformation. See the documentation for the java.util.ResourceBundle class for details on how the specific locale classes should be named.

ListResourceBundle's are used instead of PropertyResourceBundle's because data more complex than simple strings need to be provided to configure certain Classpath components. Because ListResourceBundle allows an arbitrary Java object to be associated with a given configuration option, it provides the needed flexibility to accomodate Classpath's needs.

Each Java library component that can be localized requires that certain configuration options be specified in the resource bundle for it. It is important that each and every option be supplied for a specific component or a critical runtime error will most likely result.

As a standard, each option should be assigned a name that is a string. If the value is stored in a class or instance variable, then the option should name should have the name name as the variable. Also, the value associated with each option should be a Java object with the same name as the option name (unless a simple scalar value is used). Here is an example:

A class loads a value for the format_string variable from the resource bundle in the specified locale. Here is the code in the library class:

  ListResourceBundle lrb = 
    ListResourceBundle.getBundle ("gnu/java/locale/LocaleInformation", locale);
  String format_string = lrb.getString ("format_string");

In the actual resource bundle class, here is how the configuration option gets defined:

  * This is the format string used for displaying values
private static final String format_string = "%s %d %i";

private static final Object[][] contents =
  { "format_string", format_string }

Note that each variable should be private, final, and static. Each variable should also have a description of what it does as a documentation comment. The getContents() method returns the contents array.

There are many functional areas of the standard class library that are configured using this mechanism. A given locale does not need to support each functional area. But if a functional area is supported, then all of the specified entries for that area must be supplied. In order to determine which functional areas are supported, there is a special key that is queried by the affected class or classes. If this key exists, and has a value that is a Boolean object wrappering the true value, then full support is assumed. Otherwise it is assumed that no support exists for this functional area. Every class using resources for configuration must use this scheme and define a special scheme that indicates the functional area is supported. Simply checking for the resource bundle's existence is not sufficient to ensure that a given functional area is supported.

The following sections define the functional areas that use resources for locale specific configuration in GNU Classpath. Please refer to the documentation for the classes mentioned for details on how these values are used. You may also wish to look at the source file for `gnu/java/locale/LocaleInformation_en' as an example.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

16.1 String Collation

Collation involves the sorting of strings. The Java class library provides a public class called java.text.RuleBasedCollator that performs sorting based on a set of sorting rules.

Note that some languages might be too complex for RuleBasedCollator to handle. In this case an entirely new class might need to be written in lieu of defining this rule string.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

16.2 Break Iteration

The class java.text.BreakIterator breaks text into words, sentences, and lines. It is configured with the following resource bundle entries:

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

16.3 Date Formatting and Parsing

Date formatting and parsing is handled by the java.text.SimpleDateFormat class in most locales. This class is configured by attaching an instance of the java.text.DateFormatSymbols class. That class simply reads properties from our locale specific resource bundle. The following items are required (refer to the documentation of the java.text.DateFormatSymbols class for details io what the actual values should be):

Note that it may not be possible to use this mechanism for all locales. In those cases a special purpose class may need to be written to handle date/time processing.

[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

16.4 Decimal/Currency Formatting and Parsing

NumberFormat is an abstract class for formatting and parsing numbers. The class DecimalFormat provides a concrete subclass that handles this is in a locale independent manner. As with SimpleDateFormat, this class gets information on how to format numbers from a class that wrappers a collection of locale specific formatting values. In this case, the class is DecimalFormatSymbols. That class reads its default values for a locale from the resource bundle. The required entries are:

Note that several of these values are an individual character. These should be wrappered in a String at character position 0, not in a Character object.

[Top] [Contents] [Index] [ ? ]

About This Document

This document was generated by Andrew John Hughes on February, 6 2009 using texi2html 1.76.

The buttons in the navigation panels have the following meaning:

Button Name Go to From 1.2.3 go to
[ < ] Back previous section in reading order 1.2.2
[ > ] Forward next section in reading order 1.2.4
[ << ] FastBack beginning of this chapter or previous chapter 1
[ Up ] Up up section 1.2
[ >> ] FastForward next chapter 2
[Top] Top cover (top) of document  
[Contents] Contents table of contents  
[Index] Index index  
[ ? ] About about (help)  

where the Example assumes that the current position is at Subsubsection One-Two-Three of a document of the following structure:

This document was generated by Andrew John Hughes on February, 6 2009 using texi2html 1.76.