Advanced Software Engineering - Dipartimento …advingsoft/versioniPDF/Lesson1... · Outline of the course • Models and metamodels • Revisiting UML: UML metamodel • Definition

Advanced Software Engineering

Paolo Bottoni

Lecture 1: Introduction

Outline of the course

• Models and metamodels

• Revisiting UML: UML metamodel

• Definition of Domain Specific Languages

• Model transformations – Horizontal transformations

• Within-language: Refactoring; Between-languages: Model translation

– Vertical transformations • Within-language: Refinement; Between-languages: 2-way generation

• Tools of the trade Formal frameworks: AGG, EMF

Integrated Development Environments: Eclipse, NetBeans

– UML-Based tools

– Web-development: WebRatio

– Generation of Domain Specific Languages: MetaEdit+

• Hands-on: Defining a small application

2 Lesson1Introduction Advanced Software Engineering

Models and metamodels

• System: a set of resources organized to:

– Interact with the environment

• (Software system) Provide services

– Maintain its viability

• Model: a description of the system through different artefacts.

– Allows communication and reasoning about the system

• Metamodel: a model defining a language in which to construct models.

• Layering of models and metamodels (MDA)

– Instances

– Models

– Metamodels

– Metamodeling languages

• Recursive metamodeling

Lesson1Introduction 3 Advanced Software Engineering

A metamodel for system descriptions

(from L. Balmelli, D. Brown, M. Cantor, M. Mott, “Model-driven systems development”,

IBM Systems Journal, 45(3):569-585, 2006)


Metamodels and domain models

• Metamodel as abstract syntax

• Domain model as constraints on possible realisations

• From a metamodel to a model

– Model instance of language defined by metamodel

– Metamodel contains metaclasses and meta-relationships

– Model classes and relationships instances of meta-elements

• From domain model to system model

– System model specialises and extends domain model

– Reuse or specialisation of notions

– Instances define system configuration


From domain models to metamodels

• Transformations of classes into metaclasses

• Transformation of classes into stereotypes

• Definition of mechanisms for class generation

• Constraints on possible realisations


Extent of domains

• General domains

– E.g. banking, business, automotive, physics, media

• Technical spaces

– E.g. software platforms, UI generation, real-time

applications

• Specific domains

– E.g. a bank, a company, EJB, CORBA


An example of Model-based development

From use case diagrams …

… to UML class diagrams …

… to Java

• Abstract representations on metamodel graph

• Graph transformations on metamodel elements

• Three different metamodels involved


CIM

(Use case

language)





Metamodel representation

(UML Class Diagram language)

Transformations

1 : Classifier

name=X

mk: Attribute

: UseCase

name=Y

ownedUseCase

2 : Class

name=X

:=

2 : Class

name=X

1 : Classifier

name=X mk: Attribute

: Operation

name=lowerCase(Y)

ownedOperation

: Classifier

name=X

mk: Attribute

: UseCase

ownedUseCase :=

… to class diagrams

:1: Classifier

name=X :=

: Class

name=X

1 : Classifier

name=X

mk: Attribute mk: Attribute


: Property

ownedProperty

Analogous rule

for properties

Transformations Result

: Class

name="Account"

: Operation

name="deposit"

ownedOperation

: Operation

name="withdraw"

: Operation

name="check"

Account

deposit()

check()

withdraw()


: Operation

name=“transfer"

balance : float = 0.0

: Property

name=“balance"

ownedAttribute

: Class

name=“Transaction" ownedOperation Transaction

transfer()

Maintaining the connection


Metamodel representation

wth: Operation

name=“withdraw”

dep: Operation

name=“deposit”

chk: Operation

name=“check”

ac : Class

name=“Account”

ownedOperation

ownedOperation

ownedOperation

PIM

(Class diagram language) Account

withdraw()

deposit()

check()

From UML class diagrams …


Transformations

: Class

name=X

mk: Attribute

: Operation

name=Y

ownedOperation :=

1 : Class

name=X

mk: Attribute

: Operation

name=Y

ownedOperation

2 : class

name=X

:=

2 : class

name=X

1 : Class

name=X

mk: Attribute

: method

name=Y

: type

name=“boolean”

… to Java

1 : Class

name=X :=

: class


1 : Class



Analogous rule

for attribute

name=“withdraw”

dep: method

name=“deposit”

chk: method

name=“check”

ac : class

name=“Account”

wth: method Metamodel representation : type

name=“boolean”

: type

name=“boolean”

: type

name=“boolean”

PSM

(Java language) class Account {

boolean withdraw();

boolean deposit();

boolean check();

}

Further transformations needed to flesh out methods

Resulting Java code


Maintaining the relation between models


Creating operations


Aggiunta metodi


Specification of behaviours

wth: method

name=“withdraw”

public boolean withdraw(float amount) {

if(amount<0) return false;

else if(amount>=0) {

if(balance<amount) return false;

else if (balance>=amount) {

removeAmountFromBalance(amount);

sendSignal("ServeAmountToCustomer");

}

}

return true

} Lesson1Introduction 20 Advanced Software Engineering

Specification of behaviours: SD

public boolean transfer (float amount, Account otherAccount ) {

boolean check1, check2 = false;

check1 = otherAccount.deposit(amount);

if (check1) check2 = withdraw(amount);

return check2;

}


Specification of behaviours: stereotyped SD public boolean transfer (float amount, Account otherAccount, custAccount ) {

boolean check1, check2 = false;

Transaction trans = new Transaction("transfer", amount, custAccount, otherAccount, this);

check1 = trans.PREPARE("deposit");

if (check1) {

check2 = trans.PREPARE("withdraw");

if (check2) trans.COMMIT();

else trans.ROLLBACK();

}

else trans.ROLLBACK();

return check2;

}


Integrated Development Environments

• Eclipse – IDE for Java

– Plugins provide extensions to other modeling or programming languages

– Set of transformations available: refactoring

– Own meta-modeling facility

• Netbeans – Support to different languages integrated

– Set of transformations available: refactoring, pattern conformance.

– Support to generation towards different platforms

• MetaEdit+ – Definition of metamodels.

– Generation of code to manage models

– Distinction between library and generated code

• WebRatio – (I)DE for Web development and deployment

– Based on WebML, extension of UML + navigation information


Recommended readings

• OMG documentation on UML and MDA

• Stefano Ceri, Piero Fraternali, Aldo Bongio, Marco

Brambilla, Sara Comai, Maristella Matera, Designing Data

Intensive Web Applications, Morgan Kauffman, 2003

• Markus Völter, Thomas Stahl, Jorn Bettin, Arno

Haase, Simon Helsen, Model-Driven Software

Development: Technology, Engineering, Management,

Wiley 2006

• Markus Völter, DSL Engineering - Designing,

Implementing and Using Domain-Specific Languages,

http://voelter.de

• Steven Kelly, Juha-Pekka Tolvanen, Domain-Specific

Modeling: Enabling Full Code Generation, Wiley 2008


http://voelter.de/

Lesson1Introduction 25

http://wwwusers.di.uniroma1.it/~advingsoft

[email protected]

Advanced Software Engineering

http://wwwusers.di.uniroma1.it/~advingsoft

Advanced Software Engineering - Dipartimento …advingsoft/versioniPDF/Lesson1... · Outline of the course • Models and metamodels • Revisiting UML: UML metamodel • Definition

Documents