Monday, November 09, 2009

j2moose - how to reverse engineer j2ee

Saw this interesting research project:
Enabling the evolution of J2EE applications through reverse engineering and quality assurance


To quote:
To address the complexity of enterprise applications, J2EE offers a conglomerate of several technologies, (e.g. Enterprise Java Beans - EJB or Java Server Pages - JSP) using several languages, (e.g. Java, XML or SQL). In this context, simply applying existing reverse engineering and quality assurance techniques developed for object-oriented systems fails due to two major reasons:

analyzing only the Java source code overlooks the information written in other languages such as the XML configurations, the JSP files, the database structure or the SQL statements, and
even when analyzing the Java source code we need to consider the technology specific patterns (e.g. implementing specific interfaces).
This project aims to conduct a systematic study in reverse engineering and quality assurance of J2EE applications. In particular, we target the following questions each of them being addressed in a separate track:

How do we model J2EE to support analysis of the different languages?
What defines internal quality in J2EE applications and how do we measure it?
How do we visualize the diversity of languages to support understanding of J2EE applications?


Doc: http://scg.unibe.ch/archive/projects/Gurt06aJSP.pdf
Master: http://moose.unibe.ch/about
Description: http://scg.unibe.ch/research/hasler07

The implications for me: I want something to do automated marking for J2EE assignments. Basically a first pass would reverse engineer the submitted code, decide/visualise the model with certain matrices and then award a technical mark.
Humans (tutors) would then mark the assignment for completeness, correctness and qualitative measures such as "ease of use", "look and feel", "niceness" :-)

No comments: