KABA - Download

Universität Passau Lehrstuhl für Softwaresysteme Innstraße 33 94032 Passau, Germany Phone: +49 851 509-30 91 Fax: +49 851 509-30 92 e-mail: snelting@fmi.uni-passau.de WWW: http://www.infosun.fmi.uni-passau.de/st/

The distribution of KABA is currently split into two parts:

• A front-end which extract member access from Java class files and calculates points-to information
• A graphical user interface which calculates the concepts, visualizes them and provides some (limited) support for interactive reengineering of the new class hierarchy

Important!

• THIS IS A RESEARCH PROTOTYPE. Its quality and stabilty does not meet common standards.
• This work may not be used elsewhere without explicit permission
• This is a work in progress with constant development. There are no guarantees, not even that any particular version will compile.
• No part of the source code contained in this work was written to be useful to or readable by anyone else than it's original author.
• This work will not scale beyond sample applications. It is not a full strength reengineering tool
• The build instructions on this page do not necessarily refer to the latest available version

Front-end

• JDK 1.4.2 for running the front-end (other versions may work)
• a C compiler that can compile native code for this JVM
• JDK 1.3.1 as run-time system for analysed programs (other versions propably will not work)
• GNU Trove (currently used version: 1.0.2, other versions may be usable)
• ANT with support for compiling C programs (versions 1.5.4 and 1.0b1 will work, other versions will propably work)

Graphical User Interface

• GNOME 2.4 Developer Platform
• gcc (currently used version: 3.2, older versions will probably not work, other c++ compilers may work)
• Graphviz (currently used version: 1.10 + additional patches, other versions may work)

Installation & usage instructions

Building the front-end

• unpack the front-end sources and change into their directory
• create a subdirectory requirements/trove-1.0.2/ and copy trove.jar into the created directory.
  If you use a different version of trove adjust the path in build.xml
• use ant to compile the sources.
• create a directory classes/stubclasses. Copy rt.jar from JDK 1.3.1 into this directory.
  If you want to load rt.jar from a different path, adjust build.xml
• use ant stubs to build the stubs for some classes and methods required by the analysis
• test it by doing something like

  mkdir test
  cd test
  javac -d . ../samples/ABC.java
  ../kaba --classpath=. Test --kaba-dump=abc.kaba
  

  It should finish with a very crypting data file abc.kaba containing the analysis results

Building graphviz

• unpack the graphviz sources and apply the patches:

  gzip -dc graphviz-1.10.tar.gz | tar xvf -
  cd graphviz-1.10
  patch -p1 < graphviz-1.10-install.patch
  patch -p1 < graphviz-20030516-assert.patch 
  patch -p1 < graphviz-20030520-debug.patch
  ./autogen.sh
  

• configure, build and install graphviz

  ./configure --prefix=PREFIX --with-dynagraph ---without-tclsh && make && make install
  

  Replace PREFIX with something suitable for your system.

Building the graphical user interface

• unpack the gui sources and change into their directory
• use

  ./configure --with-graphviz-prefix=PREFIX && make
  

  to compile the sources. Use the PREFIX where you installed graphviz. Please pay attention to the error messages generated by configure.
• Use

  ./gui abc.kaba
  

  with the abc.kaba generated above.