(13)DJ: A Java-based Constraint Language and System

Title of the Research Project, Principal Investigator

(1) Title of the Research Project

(2) Principal Investigator

Contents

(1) Purpose of the report

DJ is a language that amalgamates Java and Constraint Programming. As an extension of Java, DJ significantly simplifies the process of constructing GUIs and Java applets. Developing GUIs including applets in Java is a time-consuming process. The users have to choose appropriate layout managers and sometimes have to determine the sizes and positions of graphical components. In contrast, in DJ, the users only need to specify the components that compose a GUI and the relationships among the components by using constraints. The layout for the components is automatically determined by the system. As a constraint programming language, DJ improves the current constraint languages in that problems and solutions can be described in the same language. And most importantly, because DJ is a compiling language that uses Java as the object language, solutions of Constraint Satisfaction Problems (CSPs) can be distributed on the World Wide Web as Java applets.

During the last fiscal year, we developed a compiler for DJ in B-Prolog (http://www.cad.mse.kyutech.ac.jp/people/zhou/bprolog.html). The objectives for this fiscal year are as follows:

(a) To extend the base classes and constraints to make it easier to use DJ develop various applications.
(b) To extend DJ and enable it handle events.
(c) To write more example applications.
(d) To make both the system and manual more user-friendly.

(2) Current status of the project

We have introduced several new constraints, most of which are layout constraints, into DJ. The constraints include table, tree, notOverlap, line, circle, etc. We are still working on implementing some of the constraints.

We have ported a Java interface (developed by Nobukuri Kino) to B-Prolog. With this interface, it is possible now to display partial solutions without generating Java programs and compiling them. We plan to implement an environment through which users can control the search process and pick up satisfactory solutions.

We are implementing two applications, one for developing time tables for classes at Kyushu Institute of Technology and the other on drawing proof trees.

(3) Major result at this moment

Some results are described in (2). In addition, we have done the following extensions and improvements:

(a) it is possible now to attach actions to components that will be taken when the components are clicked. We developed two example programs that uses this new functionality: One is a working calculator and the other is a WEB page navigator. We plan to write more example programs that use this new feature.

(b) Arrays in the previous version were static. We have made it possible for users to define dynamic arrays, which are very important to write GUIs for which the number of components are not known in advance.

(c) We re-implemented the parser to improve the performance and to enable support inner classes in JDK 1.1.

(4) Image of the resulting software

The software will become more robust. We will check every functionality by using several test programs. The applications will also be helpful to this end.

The software will become much easier to use: The reasons are:

1. There is an environment through which users can control the search process and select solutions.
2. There will be a large number of base classes and constraints that can make it dramatically easy to write many applications.
3. More example programs will be available that illustrates the programming techniques for DJ.