As promised we have evaluated whether our case study on the logistics domain is a good fit for the case study in the extended version of our McGPLE paper. As it turns out: There are a lot of variations, and most of them are context-dependent and dynamic. There are complex mappings from requirements to implementation, and some of the adaptations involves complex reconfiguration of software components (so the issue of safe reconfiguration is also present).

Currently we have applied the OVM method by Pohl et al. for documenting external variability and to a lesser extent internal variability, and we are currently in the process of extending the resulting variability models with dynamic variability and contextual information. In attachement you can find two documents as background infomation for preparing the meeting. Note these documents are confidential and should not be distributed further. The first document is a domain analysis of the logistisc domain. The second document presents the variability models that we designed using the OVM method described in the book of Pohl et al. We are currently in the process of extending these variability models with context information. This is not yet included in these documents.

We therefore would like to organize a meeting to present the case study, the context-dependent adaptations, and our key findings with limitiations of using the OVM method. We see it feasible to hold this meeting on 29 or 30 March. Please indicate when it is feasible for you to have meeting in the morning or afternoon during these 2 days


10h30-10h45: Small recap of MCGPLE paper (Eddy) 10h45-12h45: Presentation on case study + Discussion and feedback on case study documents (Javier) 12h45h-13h00 : Concrete discussion on how to go about approaching the more extended version of the article

Main action points:

  • The consensus was that the context modeling in the case study has not progressed far enough in order to derive specific requirements for dynamic software product lines.
  • ⇒ Action point: Leuven will continue working on this and organize a technical workshop as part of the Stadium project to further refine the context modeling
  • ⇒ Action point: Leuven will investigate work on context modeling from the AI world.
  • The OVM method used in the case study is not necessarily the best method if the goal is to automatically reason over feature configuration.
    • ⇒ Action point: Arnaud will send a list of papers on feature modeling
  • The paper itself should be written as an experience report, probably focussing on subset of the research questions
    • ⇒ Action point deferred until more specific requirements are distilled from the case study.
wp3/minutes29march2010.txt · Last modified: 2011/03/11 11:27 by eddy.truyen