CompilationAndDeployment

Installing from Source Code

The HyperGraphDB source code resides in the Subversion repository and releases are structured following standard Subversion conventions:

• The main branch containing latest development code can be found under trunk.
• Each release can be found under tags.

HyperGraphDB consists of a core component implementing the nuts and bolts of the database itself, together with several example applications that use HyperGraphDB in different domains. Some of the applications have dependencies on other applications, and all have a dependency on the management component.

Source Code Directory Structure

Source code is organized in the following top-level directories:

apps	All HyperGraphDB application modules. There is one directory per application module under apps.
core	The core of HyperGraphDB, including the distributed database functionality.
viewer	A component for visualization of HyperGraphDB graphs.
test	Unit tests, mainly for the core component.

The directory structure under core, viewer, and each application module roughly follows this pattern:

src/	The source code. Sub-divided under java, possibly config and possibly other languages.
jars/	Third-party Java library dependencies for that particular component.
etc/	Optional, self-explanatory directories depending on the component.

Compiling the Source

HyperGraphDB uses the ANT build system. You can compile the entire package, or individual components with the top-level ant script. You can also compile individual components from their own home directories by running their respective ant scripts.

Tip: Use the command ant -p to list all available targets for a project script.

Compiling the HyperGraphDB Core

There are a number of different targets that can be built from core, depending on whether you will use HyperGraphDB as a distributed database, or simply as an embedded database. The following table describes the three targets used to build HyperGraphDB for deployment:

Target	Result	Contents
full-jar	hgdbfull.jar	Contains both the database management and the distributed version code.
core-jar	hypergraphdb.jar	Contains just the embedded database version.
peer-jar	hgpeer.jar	Contains just the peer/distributed version related code.

The full-jar target merges hypergraphdb.jar with hgpeer.jar to give you all the functionality you need to create both local and distributed databases. The code for hgpeer.jar is contained entirely in the org.hypergraphdb.peer package.

Compiling the Application Modules

Each application module is essentially contained in one .jar file. The build.xml file in the apps directory contains the target for each respective application. The next table describes each of these targets:

Target	Description
loadWordNet	Loads WordNet into a HyperGraphDB instance.
management	Builds the HyperGraphDB Management JAR.
prolog	Builds the tuProlog JAR.
sail	Builds the HyperGraphDB Sail JAR, for use with the [http://www.openrdf.org Seseme RDF framework.
tm	Builds the HyperGraphDB TopicMaps JAR
wordnet	Builds the HyperGraphDB WordNet Jar
xsd	Build HyperGraphDB XSD JAR

Compiling with GCJ

The following has only been tested on Ubuntu 8.10, but should work on any system where GCJ is installed.

First, compile the Berkeley DB Java interface with GCJ as follows. In the Berkeley DB distribution top folder run:

gcj -fjni -o dbjava.o -c ``find java/src -name '*.java' | grep -v debug``

This command produces an object file, dbjava.o that needs to be linked to HyperGraphDB. If you're compiling on a 64-bit system, add the option -fPIC to the above command line. Then in 'hypergraphdb/core' run:

gcj -shared --classpath=jars/db.jar dbjava.o hypergraphdb.jar -o hgdb.so

Again, add -fPIC if you are building on a 64-bit system. This will produce a shared library called hgdb.so that can be loaded from C++ applications.

Deployment

Deployment of HyperGraphDB and/or its application modules is described in Deploying Applications.