Research topics
Our research interest is at the intersection of process modelling and software component modelling in order to offer a better way of controlling adaptation and reconfiguration of these software entities. Non-functional properties are (or should be) the guide for selecting the variants during reconfiguration.
Process modelling
Participants : A. Beugnard, F. Dagnat
Current process models are not flexible enough to describe the development processes we use and to allow easy reuse of processes. We are therefore proposing a new process model with a software approach: processes have a specification, possibly many design variants with several implementations (enactments).
Component model
Participants: A. Beugnard, F. Dagnat, MT. Segarra
We propose a new component model. A component is a set of distributed acces points. Component are assembled with local acces point binding. The specification of a component is based on contracts.
We are interested in the process development of this "software entity". We propose a refinement process where a sequence of issues have to be solved. For each issue, solutions are injected into the component model with model weaving techniques.
Reconfiguration / Adaptation
Participants : A. Beugnard, F. Dagnat, MT. Segarra
Our research is two fold. First we derive information from the sequence of solutions we inject in a component variant. This information is used to ease the dynamic reconfiguration of a component with the construction of a reconfiguration plan that describes the actions needed to go from an implementation variant to another one.
The second reconfiguration approach is based on the idea that a reconfiguration is a sequence of actions that go from an implementation to another one. We propose a language to plan this
reconfiguration.
Beyond this work on planning reconfiguration, we also intend to provide a middleware dedicated to reconfiguration. We propose basic mechanisms required for reconfiguration plan execution.
Non-functional properties
Participants : A. Beugnard, Julien Mallet, Siegfried Rouvrais
Non-functional properties are parts of the issues that have to be tackled during the design process. We study how we could use the knowledge of the process to have a better control on the emerging features of the implementation variants.
Modelling
Participants: A. Beugnard, F. Dagnat
We study how we can escape from the traditional UML/EMF strict modelling and how models could be federated with different levels of abstractions or different points of view.
This work is done with the "Société Coopérative d'Intérêt Collectif" (SCIC) OpenFlexo.
- Site hébergé par IMT Atlantique