|
|
2nd OOPSLA Workshop onGenerative Techniquesin the context of Model Driven Architecture |
Please note that a workshop on Best Practices for Model-Driven Software Development
is scheduled for OOPSLA 2004!
2 Dec 2003: Two additional presentations are available for download.
27 Nov 2003: Workshop results and presentations are now available for download!
Thanks to Peter van Emde Boas we've got an extensive set of photos from the workshop!
Object-oriented technology indisputably provided us
with a better handle on complexity than previous technologies.
Nevertheless, the growing size of applications and the demands for
shorter time-to-market mean that many issues remain, and a combination
of generative and model-driven techniques can help us address them. The
OOPSLA'02 workshop on this topic was highly successful and led to the launch of the Generative Model Transformer
(GMT) open source initiative. A related topic is the possible
definition of a common standard notation for model transformation. The
discussion on this language (MOF 2.0 Query/ Views/ Trans-formations
RFP) is still going on, but several initiatives have started
investigating the feasibility of this approach and we welcome the first
research results.
From http://www.omg.org/mda/executive_overview.htm:
"Model Driven Architecture™ addresses the complete life cycle of designing, deploying, integrating, and managing applications as well as data using open standards. MDA-based standards enable organizations to integrate whatever they already have in place with whatever they build today and whatever they build tomorrow.
Most importantly, MDA enables the creation of standardized Domain Models for specific vertical industries. These standardized models can be realized for multiple platforms now and in the future, easing multiple platform integration issues and protecting IT investments against the uncertainty of changing fashions in platform technology.
The benefits of MDA are significant-to business leaders and developers alike:MDA provides a solid framework that frees system infrastructures to evolve in response to a never-ending parade of platforms, while preserving and leveraging existing technology investments. It enables system integration strategies that are better, faster and cheaper."
- Reduced cost throughout the application life-cycle
- Reduced development time for new applications
- Increased return on technology investments
- Rapid inclusion of emerging technology benefits into their existing systems
Generation languages that don't use models as input
may have practical limits in terms of the complexity of code that can
be generated. This far there is very little in terms of concrete tools
that actually support MDA beyond traditional UML modeling and
skeleton-class generation. Some new tools provide features to define
and instantiate design patterns, but many tools still expose the user
to UML models at the level of abstraction of implementation code and do
not provide the support to implement domain specific languages to raise
the level of abstraction.
The workshop aims to bring together practitioners, researchers,
academics, and students to discuss the state-of-the-art of generative
techniques in the context of model-driven architecture.
Topics of interest include:
The goal is to share experience, assess the
state-of-the-art and the state-of-the-practice, consolidate successful
techniques, and identify the most promising application areas and open
issues for future work.
Potential participants are asked to submit a two-page (or longer)
position paper detailing their experience with model-based generative
techniques, their perspective on one or more of the above topics, and
their planned contribution to the workshop. We seek concrete case
studies and potential topics of discussion in order to ground the
workshop in real-world issues. Please mail your submission (in PDF or
Word) to Jorn Bettin (jorn.bettin@softmetaware.com).
Model-Driven Architecture promises stable models in UML and flexible target platforms. In conjunction with generative techniques MDA will improve the quality of object oriented systems, enable increases in productivity, and will allow us to:
Abstract modeling techniques and visual modeling
notations have been in practical use for decades in various forms.
However their impact this far has been limited by a lack of integration
with sufficiently robust code generation/transformation engines.
Generative techniques have also been in practical use for decades
(e.g., compiler development). However, developing new domain-specific
languages (DSLs), application generators, and component generators has
been extremely hard, requiring being knowledgeable and experienced both
in language design and compiler development.
Template languages in particular seem to be a key element in bridging
the gap from platform independent model to platform specific model, or
from model to implementation, and they warrant further investigation.
Generative techniques are an important element in realizing the vision
of the recent Model-Driven Architecture initiative by the OMG. Although
the first prototypes of UML virtual machines are emerging, model
transformations are still necessary and practical concerns in terms of
system performance lead to generation as a pragmatic, currently viable
alternative for the industrial implementation of MDA.
Jorn Bettin, the founder of SoftMetaWare,
is a software consultant with a special interest in designing
large-scale component systems and techniques to optimize the
productivity of software development teams. SoftMetaWare provides CTO
services and product line engineering expertise to software development
organizations. Over the last eight years Jorn has implemented
automated, model-driven development in several software organizations,
has worked in methodology leadership roles in an IBM product
development lab, and has managed object oriented software development
projects in the domains of Distribution, Manufacturing, Insurance, and
Electricity Trading. Currently, together with Ghica van Emde Boas, he
is leading the Generative Model Transformer
Open Source initiative to develop an MDA tool platform in close
collaboration with several universities and the submitters to the OMG's
MOF 2.0/QVT RFP.
Ghica van Emde Boas is an IT-Architect who has been working with
the IBM SanFrancisco business components framework - probably the
largest Java framework in existence - for the last five years.
Applications developed with SanFrancisco rely heavily on UML
model-based code generation. She developed a model-driven generative
tool, FUUT-je, the principles of which are published in http://www.research.ibm.com/journal/sj/392/vanemdeboas.html.
FUUT-je has been used very successfully in multiple engagements. Since
March 2000 Ghica is working as an independent consultant (URL: www.bronstee.com), after a 30-year career at IBM. Currently, together with Jorn Bettin, she is leading the Generative Model Transformer
Open Source initiative to develop an MDA tool platform in close
collaboration with several universities and the submitters to the OMG's
MOF 2.0/QVT RFP.
Aditya Agrawal is a graduate student at the Institute for
Software Integrated Systems (ISIS) at Vanderbilt University. His area
of specialization is domain specific modeling languages. Currently, he
is working on graph transformations techniques and how they can be used
to specify model to model transformations. In this endeavor he has
developed a graph transformation based language GReAT (Graph Rewriting
and Transformation Language) and has used the language to specify
various transformations.
Ed Willink is a Chief Technical Consultant with Thales Research
and Technology [UK] Ltd. He works primarily in the signal processing
field, designing and analyzing systems and environments. Years of
frustration with the poor level of tool support, led to a Ph.D.
initially titled "Effective Compilation Techniques for Fixed Point
Digital Signal Processors" which morphed into "Meta-Compilation for
C++" in response to difficulties managing a large body of C++ compiler
code. Recent considerations of the specification and portability
problems for Software Radios resulted in the proposal for a Waveform
Description Language, which like so many other problems needs a good
Model Driven Architecture tool to make it practical. Ed is a core
contributor to the Generative Model Transformer
Open Source initiative to develop an MDA tool platform and has been
working on the UMLX concrete graphical transformation syntax.
Jean Bézivin is professor of Computer Science at the
University of Nantes, France and a member of the new ATLAS research
group created in Nantes (INRIA and IRIN). He has been very active in
Europe in the Object-Oriented community, starting the ECOOP series of
conference (with P. Cointe), the TOOLS series of conferences (with B.
Meyer), the OCM meetings (with S. Caussarieu and Y. Gallison) and more
recently the <<UML>> series of conferences (with P.-A.
Muller). He created at the University of Nantes in 1988, one of the
first European Master Programs in Software Engineering completely based
on object-technology. His present research interests include
object-oriented analysis and design, product and process modeling,
transitioning from pure object technology to general model engineering,
and assessing the new emerging paradigm of model transformation.
Expected number of participants: 20-35
Cut-off point: 15 position papers, multiple authors per paper allowed.
If more high quality submissions are received, the organizers may
encourage submitters of papers on closely related topics to collaborate
on a joint position paper.
Potential participants will be asked to submit a
two-page (or longer) position paper detailing their experience with
model-driven and/or generative techniques, their perspective on one or
more of the workshop themes, and their planned contribution to the
workshop. We will encourage submitters to submit concrete programming
problems and/or other challenging issues from the practice. We will
interact with the potential participants via e-mail to select a set of
problems and issues to be discussed in focus groups during workshop.
All accepted position papers will be published on this workshop page
prior to the workshop and the participants are asked to read the papers
prior to the workshop.
All position papers in a zip file (2.6 MB)
The workshop will aim to foster discussion and interaction rather than presentations. Presentations will serve to introduce a case study, provide triggers for discussion in the second part of the workshop by presenting a controversial point of view, or by introducing a new point of view. All participants will be given a chance to make a short presentation (5-10 min). However, due to the large number of participants, discussions will be structured around a set of issues/questions provided by the organizers based on the topics covered by the position papers.