ICSE2010 Day 1 some reflections on papers

May 6, 2010

While I listened to some 6 papers I will reflect here only on the two I found most interesting.

Asuncion, Asuncion & Taylor. Software Traceability with Topic Modeling.

This talk presented a prospective traceability approach., it is prospective in the sense that traceability links are actively built while the developer are working on the project.  They have used topic linking, specifically utilizing the Latent Dirichlet Allocation model used in machine learning to automate the process. Some further work is probably necessary here because the topics seemed to be mere word lists, rather than truly semantic-rich topics. Nevertheless, they symbiotic relationship between the prospective traceability techniques and the automated topic modeling is interesting. There approach does not directly link artifacts, but rather does so indirectly through the architectural model.   They also developed a toolset to demonstrate the techniques working together.  Overall I think this is interesting idea to help with traceability.

Kornstaedt & Reiswich. Staying Afloat in an Expanding Sea of Choices: Emerging Best Practices for Eclipse Rich Client Platform Development.

Eclipse Rich Client platform is a platform for developing and deploying rich client applications. It allows for the deployment of native GUI applications to a variety of desktop operating systems.  Actually coming to grips with the platform and choosing between the options it offer presents a challenge.  It is this challenge that this paper tries to address.

They, in essence, propose some guidance as to how to break down the technical model into deployment units. For this they propose groupings of deployment units according to tool components, domain components, service components, service proxy components, service provider components and library components. Their set of best practices also include things such as naming conventions.

While I am not familiar with the Eclipse Rich Client Platform their suggestions certainly appear sound and sensible (logical). I will argue with the therm “best practices” as these seem to have emerged from their own experience rather than from studying a community empirically. They at best have good practices derived through logical argumentation from their experience. No doubt that this can be useful if a newcomer (say me) want to think about going the Eclipse rich Client Platform direction.