MD2P2 is a workshop about the interrelation of modeldriven development (MDD) and development processes. The workshop provides a forum for researchers and practitioners to exchange experiences on the questions how processes need to adapt or can be adapted when model-driven techniques are applied. We argue that the interrelation between MDD and development processes can be crucial for the success of MDD. For example, the need to adapt a process when introducing MDD can be a reason to decide against an MDD adoption. Further, we aim to give an introduction to foundations and research perspectives. MD2P2 is co-located with ACM/IEEE 17th International Conference on Model Driven Engineering Languages & Systems.
Model-driven development, which includes the synthesis of executable systems from models or the use of abstract languages, e.g. UML, Simulink, or DSLs, and software development processes are used to reach similar targets, such as increasing the quality of software or the e ciency of software development. Although, these goals are approached di erently, MDD and software development processes are not totally independent.
In fact, the literature has a multitude of proposals for MDD speci c development processes. Proposals for the use of MDD in context of established development processes, such as the V-Modell XT or SCRUM, however, sometimes include signi cant adaptations of the development process. Stakeholders' responsibilities might be as a ected as quality assurance activities, which need to respect the structure of the artifacts used. Partly, even process phases and sprints are adapted to enable a combination with MDD. It seems that most adaptations have pragmatic reasons, and aim at supporting a fruitful combination.
However, it is also possible that MDD and a particular development process do not t together well. For example, the bene t of an agile approach might be reduced, if long running transformation chains enforce long running sprints. Further, an MDD approach that requires to build languages and transformations rst before "business value" can be delivered to the customer, con icts super cially with some agile processes.
Thus, companies that aim at adopting MDD also have to face the questions, how appropriate development processes have to look like as well as whether and how existing development processes can be adapted. It is important to consider that development process adaptation can be cost intensive (e.g. due to required training of developers or changes in the involvement of stakeholders).
In summary, the questions whether and how an established development process must be adapted when MDD is introduced are crucial, since they can impact the e ciency of the development processes as well as the costs of the MDD introduction. In face of the growing number of MDD techniques and the variety of MDD approaches applied in practice, there is a need for systematic guidelines or best practices that help with adjusting or tailoring of development processes, when MDD is introduced.
In the following we give an introduction into foundations of the workshop's topic and present various research perspectives. We then provide an overview on this rst edition of the MD2P2 workshop.
Software processes de ne various aspects of development, such as phases or tasks, but also documents that need to be created at di erent points in time. Furthermore, software processes are concerned with \soft" aspects, e.g. teams, skills, communication and roles. Similarly in MDD there are technical aspects, such as automation, through e.g. generators or model transformations, that can predene ne-grained sequences of activities between manual and automated tasks. Further, just as processes, also MDD can, due to concepts such as abstraction, a ect soft aspects as the skill set of developers. Thus, both MDD and software development process can a ect similar and related aspects of development. However, the question how these interrelations can lead to mutual constraints or to synergy e ects is only rarely studied.
Research on the keyword software process tailoring (i.e. on the question how
processes need to be customized due to environmental factors) is rarely
concerned with MDD directly. However, as summarized in the survey of Kalus et
al., programming languages and tool infrastructure are known to be criteria for
process tailoring [
A bit more attention was drawn to this topic in the modeling community [
As our survey on processes that had been adapted for MDD has shown,
in uences between both MDD and software process can be most diverse [
Summing up, there is an awareness that software processes might need or have the potential to adapt, when MDD is applied. However, there is still much research to be done before we fully understand this interrelation. 3
This rst workshop on model-driven development processes and practices aims at calling attention to the question how development processes can be integrated with an MDD approach.
First of all, this is an empirical research question. Therefore, the MD2P2 workshop provides a forum for researchers and practitioners to exchange and discuss experiences on how the use of MDD a ects the development process in practice. For example, there is so far not much knowledge about how MDD is a ected by a maturing development processes. Further, the workshop tries to uncover what happens to the development processes in practice, when MDD is introduced. Empirical data or case studies from practice can help to approach diverse questions: { Which stakeholders are involved in modeling tasks & which stakeholders are not a ected by the integration of MDD? { These questions are interesting for two reasons. On the short term, it might be less expensive to train a smaller set of developers and stakeholders to the new technologies. On the long term, however, only stakeholders who are involved in modeling also have potential to bene t from the higher level of abstraction and the improved automation. { Which (modeling) artifacts are subject to quality assurance activities, e.g. reviews or testing? { MDD de nes the set of artifacts that represent the system under development. This includes the potential for quality assurance, when certain checks can be performed earlier in development. However, it can also change the skills that are required for e.g. reviews and with it the roles in the process. { Are development process phases adapted? Does the number or frequency of iterations change? { The structure of process phases or the iterations are essential for many processes. Changing them might have a major impact on the characteristics of the process. Since there are hints that such changes sometimes occur due to MDD, the question becomes pressing, when and why these adaptations happen. { Is there empirical evidence that the intended MDD e ects occur, e.g. does front-loading actually reduce the number of errors in later phases? { As indicated above, MDD is associated with certain hopes. However, only empirical evidence can prove whether these goals are reached.
In addition to the empirical perspective, it is also necessary to approach the topic more theoretically. Such research can cover systematic investigations of the mechanisms that drive impacts from MDD approaches on development processes or in turn de ne constraints on MDD approaches that are implicitly de ned by development processes. Based on these investigations, researchers might foster the success of MDD, e.g. by providing guidelines, methods, or tools that support practitioners in reusing or adapting development processes when MDD is introduced are required.
Investigations on what aspects of a process are a ected by MDD can concern diverse aspects, e.g.: How are di erent stakeholders integrated in the modeling activities? Can modeling tasks be split over multiple roles and phases? What is the e ect of automated veri cation methods on testing methodologies and philosophies de ned in development processes (e.g. in test-driven development processes)? Is there a need to adapt test and quality assurance activities in development processes, such that the various modeling artifacts are covered appropriately? When is it necessary or bene cial to adapt the number of development process phases or to change the frequency of iterations?
One motivation to answer these questions, is to identify combinations of MDD and processes that do not t together, i.e. where the bene ts of the process or of the MDD approach cannot fully be used. This is a rst step towards tailoring MDD or processes, such that a better t can be reached. A second motivation is that it might be possible to achieve synergies between MDD and development processes. Associated questions are: How can the combination with an MDD approach increase (or decrease) the bene ts of a process? How can the choice or adaptation of a process increase (or decrease) the bene ts of an MDD approach?
Ideally this research leads to guidelines and methods that can support practitioners in reusing or adapting development processes when MDD is used. Furthermore, researchers might come up with tool support for the integration of MDD into a given process. Finally, as both the development processes and MDD evolve and mature, there is the question how synergistic e ects can be maintained over time, i.e. how co-evolution or co-maturation of MDD and development processes can be supported.
To summarize, there are many open questions related to the combination of MDD and software processes. With this workshop we want to strongly encourage more researchers to contribute to the investigation of these questions. 4 First Edition of MD2P2 For this rst edition of the workshop, we received eight papers, of which four have been accepted for inclusion in the proceedings. The accepted papers can be split in two groups:
The rst group of accepted papers deals with the question how processes look
like that are used for the development of MDD technologies, such as
transformations or DSLs. Barisic et al. present of the case study of FlowSL, how a usability
concerns can be considered throughout during DSL development throughout an
agile development approach where MDD tools are used [
The second group of accepted papers provide an empirical view on processes
that are used in combination with MDD. Hovsepyan et al. present an experience
report on the use of MDD in development of safety-critical avionic systems [
We hope that the workshop will help researchers and practitioners to build up a community that shares data and experience. In closing, we would like to thank all authors papers and reviewers, for their contributions, the e ort they invested, and for making this workshop possible.