Antonio Vallecillo

1
Model-Driven Engineering inGISUM (Málaga)

• Antonio Vallecillo
• University of Málaga
• Web Engineering models and notations
• Meeting Munïch, December 2006

2
Agenda

• Málaga node
• Our research topics
• MDWE Proposals Integration
• MDWAF

3
Málaga Node
4
GISUM

• Grupo de Ingeniería del Software de la
  Universidad de Málaga
• Established in 1989 by Prof. José María Troya
• Within the Dept. Lenguajes y Ciencias de la
  Computación (Languages and Computer Science),
  Universidad de Málaga
• Activities (research, teaching)
• centered around distributed software systems
• Composition
• 2 full professors, 18 associate professors, 16
  assistant professors
• 30 PhD students (granted/contracted)
• Present in several organizations related to its
  research activities (OMG, ISO, ITU-T, ...)

5
Main Research lines

• Software components and architecture
• Software Quality
• Model-driven development
• Security
• Embedded systems
• Real time systems
• Collaborative Virtual Environments
• Formal methods
• Genetic Algorithms

6
GISUM Research lines

• Distributed Systems
• Ernesto Pimentel
• Manuel Díaz
• Lidia Fuentes
• Francisco Durán
• Antonio Vallecillo
• Genetic and evolutive algorithms
• Enrique Alba y Carlos Cotta
• Security
• Javier López
• Others (e.g. BD)

7
GISUM Distributed Systems

• Formal methods
• Ernesto Pimentel (leader), Fco. Durán,
• Embedded and P2P systems
• Manuel Díaz (leader),
• Aspects
• Lidia Fuentes (leader),
• ODP, MDD, Quality
• A. Vallecillo (leader)
• Luis Iribarne (UA)
• Manuel F. Bertoa
• José R. Romero (UCO)
• Nathalie Moreno
• José E. Rivera

8
1 Software quality and measurement

• Topics
• A Quality Model for software components
• Usability measures for software componentes
• Software Measurement terminology
• Tools for evaluating commercial software
  components quality
• Evaluation of process quality in industrial
  environments
• People
• A.Vallecillo, Manuel F. Bertoa
• M. Piattini, C.Calero (U. Castilla-La Mancha),
  other research groups
• Funded/Sponsored/Supported by
• MCYT (TIC project), National Network on Quality,
  Proc. Uno, Coritel (Accenture)
• Main results so far
• 2 JCR journals, 1 in progress
• 1 prototype of a tool for evaluating the
  usability of sw components
• 1 PhD (M.F. Bertoa)

9
2 Open Distributed Processing (ODP)

• Topics
• Formalization of ODP viewpoints (using Maude)
• Revision of the RM-ODP
• Edition of ITU-T X.906 ISO/IEC 19793 Use of
  UML for ODP systems specification
• Viewpoint modeling
• Viewpoint consistency
• People
• A. Vallecillo, José R. Romero, Fco. Durán
• ISO/JTC1/SC7 WG19, ITU-T SG17, Bryan Wood,
  Peter Linington, Akira Tanaka, Sandy
  Tyndale-Biscoe, Joaquin Miller
• Funded/Sponsored/Supported by
• MCYT (TIC project), AENOR, NASA/JPL, EDF
• Main results so far
• 3 JCR journals, 1 in progress
• Prototype for executing and analysing ODP
  computational specs in Maude
• 1 PhD soon (José R. Romero)

10
3 Model Driven Development

• Topics
• Model-driven framework for Web Engineering
  (MDWAF)
• UML Profile and Metamodel for WebML
• Model transformations
• Use of UML for ODP system specifications.
  Metamodels and Profiles
• People
• A. Vallecillo, Nathalie Moreno
• Jean Bézivin, Alfonso Pierantonio, Jeff Gray
• Piero Fraternalli
• Funded/Sponsored/Supported by
• MCYT (TIC project), National Network on MDD,
  AENOR
• Main results so far
• 1 JCR journal, 3 journals in progress
• 1 PhD soon (Nathalie)
• MDWAF and WebML metamodels and profiles

11
Agenda

• Málaga Node
• Our research topics
• MDWE Proposals Integration
• MDWAF

12
MDWE proposals integration (1/2)

• Option 1 Individual metamodels and
  transformations between them
• Assumptions
• There is no common metamodel, or
• No agreement is reached w.r.t. a common
  metamodel, or
• The common metamodel is not expressive enough, or
• Transformations are possible between all
  individual metamodels
• Benefits/advantages
• Individuality is respected
• Tools are readily available
• Zoos (model repositories) can be easily built
  and maintained to share models
• Disadvantages
• Integration and interoperation are much more
  difficult
• Sharing tools is complicated
• Too many transformations n(n-1)

13
MDWE proposals integration (2/2)

• Option 2 Common metamodel
• Assumptions
• There is a common metamodel
• An agreement is reached w.r.t. such a common
  metamodel
• The common metamodel is expressive enough
• Transformations are possible to/from all
  individual metamodels
• Benefits/advantages
• Integration and interoperation are easier
• Sharing tools is possible
• Core competencies (Presentation/Information/tools/
  )
• Less transformations between metamodels 2n
• Disadvantages
• Individuality is somehow lost
• Too many assumptions
• Interoperability conflicts between different
  proposals

14
Generic and Core metamodels

• MDWE proposals share a common core of concepts
  and mechanisms
• But they in how they extend it and implement some
  of these mechanisms

15
Generic and Core metamodels
16
A common metamodel?

• Should the common metamodel be
• Just the generic and core metamodel?
• The INTERSECTION of the metamodels?
• The UNION of the metamodels?
• Is such a common metamodel possible?
• How to deal with potential conflicts?
• How to integrate models coming from different
  proposals?
• How to integrate tools from different proposals?
• This is precisely what Nathalie has been
  investigating for the last 2 years

17
Agenda

• Málaga Node
• Our research topics
• MDWE Proposals Integration
• MDWAF

18
Whats MDWAF?

• MDWAF is an architectural framework for
  organizing the models that address the different
  concerns of Web applications
• Far from being yet another Web methodology, the
  aims of MDWAF can be summarized as follows
• to be able to represent, in terms of models and
  relationships between them, the concerns required
  for designing and developing Web applications
  following an architectural separation of
  concerns as prescribed by MDA (PIMs, PSMs,
  transformations, ..)
• to integrate and harmonize the models and
  practices proposed by existing approaches,
  addressing their concerns
• to provide as a common framework (and metamodel)
  in which current proposals could be integrated
  and formulated in terms of the MDA principles,
  hence allowing them to smoothly interoperate and
  complement each other, share tools, etc.

19
MDWAF concerns I
20
MDWAF concerns II

• MDWAF concerns come from
• The concerns covered by MDWE proposals
• New requirements for developing Web applications
  that integrate third party systems
• Models covered by MDWE proposals

21
Why is different the Conceptual Model

• The Conceptual Model can be used
• For specifying the basic structure and contents
  on the Web application (so the rest of the views
  can relate to the elements of that model), and
• To maintain the consistency of the model
  specifications establishing how the different
  viewpoints merge and complement each other
• It plays a relevant role when the Web application
  design integrates third party system models using
  heterogeneous notations, terminology or even
  different platforms on which being executed

22
Modeling MDWAF concerns

• Using MOF metamodels
• Using UML Profiles

23
MDWAF development process

• MDWAF is supported by a methodology to guide the
  development process of a Web application
• From scratch
• Reusing models and tools coming from other Web
  methodologies
• MDWAF does not assume that the design of a Web
  application will always start from scratch
• It is designed to allow the reuse of models and
  tools
• Strategies for addressing integration conflicts
  when integrating models that represent different
  views of the same system
• A set of adapters for solving integration
  mitmaches at model level
• Some issues we have addressed
• When models do not share a common terminology
• When models use services which are provided by
  other models
• When the levels of abstraction do not coincide
• When the model integrates several of our
  framework concerns

24
From scratch
25
Reusing models and Tools

• Proof of concept The Conference Review System
• OO-H/Visual Wade supported the design and PHP
  code generation of the User Interface Level.
• WebML/WebRatio supported the design and
  management of a MySQL database at the Data Level.
  
• The business logic of the application has been
  specified using the Xion action language,
  supported by the the Netsilon tool that generates
  code for PHP and Java.

26
Reusing models and Tools
27
Reusing models and Tools

• The previous combination of tools and
  methodologies will allow us
• to preserve the separation of concerns proposed
  by MDWAF and,
• to generate code re-using different models and
  tools.
• In particular, we have re-used the OO-H and WebML
  models for this example, faithfully taken from
  the Proceedings of IWWOST01.

28
Interoperability

• How MDWAF achieve interoperability
• By defining, e.g., interoperability bridges
  between compatible models coming from different
  proposals, whenever this is possible
• Correspondences between MDWAF and other Web
  Methodologies that allow us to derive the
  counterpart of a Web model

29
Ex Interoperability at the User Interface
Viewpoint
Table 1. Correspondences between MDWAF User
Interface Structure and other methodologies
concepts
Table 2. Correspondences between MDWAF Navigation
model and other methodologies concepts
30
Summary of MDWAF

• It provides a model-based framework for Web
  application design
• It provides a common ground shared by most MDWE
  proposals
• MDWAF can be used both to build Web applications
  from scratch and to design and implement
  applications reusing models and tools coming from
  different proposals
• To be able to achieve this goal, we have
  identified and solved some of the issues that
  need to be addressed

31
Future Research

• Ongoing efforts are focused on tool support
• The majority of MDD CASE tools are not
  Web-specific, and do not allow the smooth
  integration of all OMG tools (OCL, QVT, etc)
• We are working on a tool that integrate all the
  notations and engines that we need, and that can
  automatically generate the OCL constraints and
  the glue (i.e., QVT relations) for the three
  viewpoints based on the Conceptual model, no
  matter whether the models are developed from
  scratch or imported from other methodologies

32
Some issues for discussion

• Which option to go?
• Common metamodel
• Individual zoos automatically maintained
• Which metamodel to use?
• MOF? EMOF? ECORE? KM3?
• Should we be OMG-compatible?
• Using OMG standards for MDA (MOF, UML, QVT, )
• Using other standards and keeping consistency
  (ATLAS way)
• Dont mind.
• Which modeling tool?
• They do not interoperate!
• MagicDraw, EA, Eclipse-based,
• Combinations of those?
• Which model transformation tools?
• Supporting QVT (Together)
• Graph-based (AGG, VIATRA, ATOM3,)
• Others (ATL, AMW, )