Bcc: Krzysztof Czarnecki
Date: Thu, 14 Jun 2012 19:31:48 +0200
Subject: 15 June 2012, 13:00, L016, PEM: Re-Envisioning Lightweight Modeling
 (Krzysztof Czarnecki)
From: Vadim Zaytsev 
To: PEM 

Dear environmentalists,

Hopefully you have not only enjoyed today's presentations at the
Understanding Software Day, but also will be able to quickly recover
in order to enjoy tomorrow's PEM Colloquium. One of today's speakers,
Krzysztof Czarnecki (who talked today about variability, see
http://www.cwi.nl/abstract-cwi-lectures-understanding-variability-in-software-ecosystems),
will give another talk on his joint work with Michal Antkiewicz,
Kacper Bak and Andrzej Wasowski. Please note that the time is
different than usual (we start AFTER lunch, not before it, because
otherwise Paul will not be able to attend), and the room is L016 (next
to L017). See you there!

 Date: 15 June 2012
 Time: 13:00
 Room: L016 (CWI)
 Speaker: Krzysztof Czarnecki
 Title: Re-Envisioning Lightweight Modeling
 Abstract: see below

----------------------------------------------------------------------
          Re-Envisioning Lightweight Modeling

Model-driven engineering (MDE) has failed to reach the masses. MDE is
the use of models---reduced, purposeful, and comprehensible
representations of a system and its environment---to describe, analyze
and construct a system in a tool-supported way. It is a compelling
idea designed to deal with the ever-growing complexity of today and
tomorrow's software-intensive systems. Yet MDE remains confined to
niches that represent only a small portion of the software industry
today.

In the first part of the talk, I will attempt to uncover the causes
for this failure. Two main forms of MDE exist today: using
domain-specific modeling languages (DSMLs) or the UML, a
general-purpose modeling language. Both forms have failed to stimulate
a wide adoption of MDE. Domain-specific approaches are successful in
specific areas only---thus, being confined to niches by definition.
Another challenge to DSMLs is the high up-front investment necessary
to develop them. General-purpose languages, such as UML, again by
definition, are widely applicable but offer inevitably much less value
in any particular context. To make the situation worse, UML and its
associated tools are heavyweight, rendering UML-based MDE
cost-ineffective in most cases.

In the second part of the talk, I will turn to lightweight
modeling---an emerging form of modeling that may overcome the
limitations of DSMLs and the UML. Lightweight modeling---defined by
Zave as the use of small, abstract models and push-button
verification---focuses on easy-to-use languages, models, and tools and
thus improves the cost-effectiveness of general-purpose modeling.
While lightweight modeling has had some successes in specific areas,
such as network protocol design and verification, I believe that
several improvements are necessary to make the technology attractive
to a much wider audience. I will attempt to re-envision lightweight
modeling from first principles, starting with a lightweight method and
moving to language and tool design to support such a method. The
suggested improvements will include support for transitioning from
informal to formal, from concrete to abstract, and from
general-purpose to domain-specific. They also include support for
modeling and analyzing rich state and behavior and for co-evolution of
abstractions and examples. I will illustrate some of these points
using Clafer (see clafer.org), a lightweight modeling technology under
development.

This talk is based on joint work with Michal Antkiewicz, Kacper Bak,
and Andrzej Wasowski.
----------------------------------------------------------------------
               http://event.cwi.nl/pem
----------------------------------------------------------------------

Yours,
      Vadim.