Introduction to Introduction to Model-Model-Driven EngineeringDriven Engineering

Olivier Barais(Univ. Rennes 1 & INRIA)

Triskell Team @ IRISACampus de Beaulieu

F-35042 Rennes CedexTel : +33 299 842 541 Fax : +33 299 847 171

e-mail : barais@irisa.frD’après le cours de Jean-Marc Jézéquel

(or: Why I'd like write program that write (or: Why I'd like write program that write programs rather than write programs)programs rather than write programs)

Problématique du génie logiciel Problématique du génie logiciel aujourd’huiaujourd’hui

Olivier Barais 3

Modern Software ProblemsModern Software Problems

Importance of non-functional propertiesImportance of non-functional properties– distributed systems, parallel & asynchronousdistributed systems, parallel & asynchronous– quality of service quality of service : reliability, latency, performance...: reliability, latency, performance...

Flexibility of functional aspectsFlexibility of functional aspects: Product Lines: Product Lines– notion of notion of product lines product lines (space, time)(space, time)

Versions(Time)

Variants (Functionalities)

Timeto

Market!

Timeto

Market!1.1

1.2

1.01.0 1.0 1.0 1.0 1.0

1.1 1.1 1.1 1.1 1.1

1.2 1.2 1.2 1.2 1.2

1.3 1.3 1.3 1.3 1.3

1.4 1.4 1.4 1.4 1.4

Olivier Barais 4

Problems addressed in SEProblems addressed in SE 1960’s: Cope with inherent complexity of 1960’s: Cope with inherent complexity of

software (Correctness)software (Correctness)– Milestone: Floyd ‘Milestone: Floyd ‘assigning meaning to programs’assigning meaning to programs’

» More than 10 years to mature. More than 10 years to mature. Mature <> solved!Mature <> solved!

1965 1975 1985 1995 2005 Time

Size of« big »projects(LOC)

108

107

106

105

104 Compiler

Olivier Barais 5

Problems addressed in SEProblems addressed in SE 1970’s: Cope with project size1970’s: Cope with project size

– Milestone: Parnas, Yourdon: Milestone: Parnas, Yourdon: modularity & structuremodularity & structure » More than 10 years to matureMore than 10 years to mature

1965 1975 1985 1995 2005 Time

Size of« big »projects(LOC)

108

107

106

105

104 Compiler

Packet Switching System

Olivier Barais 6

Problems addressed in SEProblems addressed in SE 1980’s: Cope with variability in requirements1980’s: Cope with variability in requirements

– Milestone: Jackson, Meyer: Milestone: Jackson, Meyer: modelingmodeling, , object orientationobject orientation » More than 10 years to matureMore than 10 years to mature

1965 1975 1985 1995 2005 Time

Size of« big »projects(LOC)

108

107

106

105

104 Compiler

Packet Switching System

Nuclear Submarine Control

– Nokia’s GSM infrastructure:Nokia’s GSM infrastructure:»50% of requirements changed 50% of requirements changed

after after they were frozenthey were frozen »60% of these changed at least twice!60% of these changed at least twice!

Olivier Barais 8

Problems addressed in SEProblems addressed in SE 1990’s: Cope with distributed systems and 1990’s: Cope with distributed systems and

mass deployment:mass deployment:– Milestone: MS Milestone: MS (COM),(COM), Szyperski: Szyperski: product-lines &product-lines & componentscomponents

1965 1975 1985 1995 2005 Time

Size of« big »projects(LOC)

108

107

106

105

104 Compiler

Packet Switching System

Nuclear Submarine Control

GSM Network

Component = deployment unitComponent = deployment unit– focus on non-functional propertiesfocus on non-functional properties– installation/execution conceptinstallation/execution concept

»Explicit Explicit (contractual) (contractual) dependenciesdependencies»Configuration and connectionConfiguration and connection

Olivier Barais 9

OO approach:OO approach:Models and Components Models and Components

frameworksframeworks

• Guarantees ?Functional , synchronization, performance, QoS

• Changeable software, from distributed/unconnected sourceseven after delivery, by the end user

Olivier Barais 10

Problems addressed in SEProblems addressed in SE 2000’s: pervasive software integration, 2000’s: pervasive software integration,

accelerating technological changes accelerating technological changes (platforms)(platforms)

– Milestone: ?Milestone: ?

1965 1975 1985 1995 2005 Time

Size of« big »projects(LOC)

108

107

106

105

104 Compiler

Packet Switching System

Nuclear Submarine Control

GSM Network

Windows 2000

Olivier Barais 11

From the object as the From the object as the onlyonly one concept one concept– As e.g. in SmalltalkAs e.g. in Smalltalk

To a multitude of conceptsTo a multitude of concepts

Once upon a time…Once upon a time… software development looked simple software development looked simple

Design patternsDesign patterns

Collaborations

Required port Provided Port

<<Component>>Decoder

DecoderIReceiverI

DataIDataI

ComponentsComponentsCOM+DCOM

CORBAIIOP

MicrosoftC# & .Net

XMLSOAP

Sun’s Java & EJB

HTTPHTML

+ until the next ultimate middleware platform (~2005)

ProprietaryMiddleware(eg. automotive)

It's difficult -- in fact, next to impossible – for a large enterprise It's difficult -- in fact, next to impossible – for a large enterprise to standardize on a single middleware platform.to standardize on a single middleware platform. (R. Soley) (R. Soley)

MiddlewareMiddleware (middle war)

AspectsAspects

Olivier Barais 12

CollaborationsCollaborations

Objects should be as simple as possibleObjects should be as simple as possible– To enable modular understandingTo enable modular understanding

But then where is the complexity?But then where is the complexity?– It is in the way objects interact!It is in the way objects interact!– Cf. Collaborations as a Cf. Collaborations as a standalone diagram in UMLstandalone diagram in UML((T. Reenskaug’s works)

Olivier Barais 13

Design PatternsDesign Patterns Embody Embody architectural know-howarchitectural know-how of experts of experts As much about problems as about solutionsAs much about problems as about solutions

– pairs problem/solution in a contextpairs problem/solution in a context About non-functional forces About non-functional forces

– reusability, portability, and extensibility…reusability, portability, and extensibility… Not about classes & objects but Not about classes & objects but

collaborationscollaborations– Actually, design pattern applications Actually, design pattern applications areare parameterized parameterized

collaborationscollaborations

Olivier Barais 15

Middleware or Middle War?Middleware or Middle War?

COM+DCOM

CORBAIIOP

MicrosoftC# & .Net

XMLSOAP

Sun’s Java & EJB

HTTPHTML

No clear winner until now And probably not in the

near future Migration is expensive

and disruptive

The OMG tried to send in the ‘’blue helmets’’ 

with the MDA initiative

+ until the next ultimate middleware platform (~2005)

ProprietaryMiddleware(eg. automotive)

It's difficult -- in fact, next to impossible – for a large enterprise It's difficult -- in fact, next to impossible – for a large enterprise to standardize on a single middleware platform.to standardize on a single middleware platform. (R. Soley) (R. Soley)

Olivier Barais 16

Aspect Oriented ProgrammingAspect Oriented Programming Kiczales et al., ECOOP’97Kiczales et al., ECOOP’97

– MIT’s one of 10 key technologies for 2010MIT’s one of 10 key technologies for 2010 Encapsulation of cross-cutting concerns in OO Encapsulation of cross-cutting concerns in OO

programsprograms– Persistence, contract checking, etc.Persistence, contract checking, etc.

Weaving at some specific points (join points) in the Weaving at some specific points (join points) in the program executionprogram execution– Hence more than macros on steroidsHence more than macros on steroids

AspectJ for AOP in JavaAspectJ for AOP in Java– Some clumsiness in describing dynamic join pointsSome clumsiness in describing dynamic join points

What about Aspect Oriented Design ?What about Aspect Oriented Design ?

Olivier Barais 17

good modularitygood modularity

XML parsing in org.apache.tomcatXML parsing in org.apache.tomcat– red shows relevant lines of codered shows relevant lines of code– nicely fits in one boxnicely fits in one box

XML parsing

Olivier Barais 18

good modularitygood modularity

URL pattern matching in org.apache.tomcatURL pattern matching in org.apache.tomcat– red shows relevant lines of codered shows relevant lines of code– nicely fits in two boxes (using inheritance)nicely fits in two boxes (using inheritance)

URL pattern matching

Olivier Barais 19

problems like…problems like…

where is logging in org.apache.tomcatwhere is logging in org.apache.tomcat– red shows lines of code that handle logging red shows lines of code that handle logging – not in just one placenot in just one place– not even in a small number of placesnot even in a small number of places

logging is not modularized

Olivier Barais 20

AspectJ™ is…AspectJ™ is… a small and well-integrated extension to Javaa small and well-integrated extension to Java a general-purpose AO languagea general-purpose AO language

– just as Java is a general-purpose OO language just as Java is a general-purpose OO language freely available implementationfreely available implementation

– compiler is Open Sourcecompiler is Open Source includes IDE supportincludes IDE support

– emacs, JBuilder, Forteemacs, JBuilder, Forte user feedback is driving language designuser feedback is driving language design

– users@aspectj.orgusers@aspectj.org– support@aspectj.orgsupport@aspectj.org

currently at 1..5.4 releasecurrently at 1..5.4 release

Olivier Barais 21

EExpected benefits of using AOPxpected benefits of using AOP

good modularity,good modularity, even for crosscutting concerns even for crosscutting concerns– less tangled codeless tangled code– more natural codemore natural code– shorter codeshorter code– easier maintenance and evolutioneasier maintenance and evolution

» easier to reason about, debug, changeeasier to reason about, debug, change

– more reusablemore reusable» library aspectslibrary aspects» plug and play aspects when appropriateplug and play aspects when appropriate

État de la pratiqueÉtat de la pratique

Olivier Barais 23

Des logiciels complexes...Des logiciels complexes... Logiciels de grande tailleLogiciels de grande taille

– des millions de lignes de codedes millions de lignes de code– des équipes nombreusesdes équipes nombreuses– durée de vie importantedurée de vie importante– des lignes de produitsdes lignes de produits– plateformes technologiques complexesplateformes technologiques complexes– évolution continueévolution continue

Logiciels complexesLogiciels complexes Logiciels critiquesLogiciels critiques ......

Olivier Barais 24

Évolution des acteursÉvolution des acteurs

scientifique

programmeur

utilisateur

ingénieurs logiciels

utilisateurs

Time

Olivier Barais 25

Oui, Oui, Oui, mais ...Oui, Oui, Oui, mais ...

La mentalité de "l'informaticien moyen" La mentalité de "l'informaticien moyen" a t elle radicalement évoluée ?a t elle radicalement évoluée ?

Du "Codeur", au "Programmeur", à l' "Ingénieur Logiciel", ...Du "Codeur", au "Programmeur", à l' "Ingénieur Logiciel", ...

L' "Ingénieur logiciel"L' "Ingénieur logiciel"– prouve-t-il ses programmes ?prouve-t-il ses programmes ?– maintient-il à jour la documentation, les spécifications ?maintient-il à jour la documentation, les spécifications ?

Olivier Barais 26

État de la pratiqueÉtat de la pratique

50 ans après les débuts de l'informatique,50 ans après les débuts de l'informatique,

pour "l'informaticien moyen"pour "l'informaticien moyen"

la seule chose importante dans le la seule chose importante dans le logiciel c'est ...logiciel c'est ...

le Code !le Code !

=> Ingénierie Dirigée par le code=> Ingénierie Dirigée par le code

Olivier Barais 27

L'industrie logicielle aujourd'huiL'industrie logicielle aujourd'hui

end users

componentdevelopers

architects

assemblers

standardizationorganization

toolvendorsapplication

testers

siteadministrators

componenttesters

middlewareproviders

Olivier Barais 28

middlewareproviders

end users

componentdevelopers

architects

assemblers

standardizationorganization

toolvendorsapplication

testers

siteadministrators

componenttesters

De nombreux acteurs

• des préoccupations différentes

• des métiers différents

• des compétences variées

• des outils variés

• différents éléments logiciels

Séparation des préoccupations

L'industrie logicielle aujourd'huiL'industrie logicielle aujourd'hui

Olivier Barais 29

Point de vue desPoint de vue des

locataireslocataires

Point de vue desPoint de vue des

plombiersplombiers

Point de vue desPoint de vue des

électriciensélectriciens

Point de vue desPoint de vue des

architectesarchitectes

Point de vue desPoint de vue des

paysagistespaysagistes

Point de vue duPoint de vue du

cadastrecadastre

Point de vue desPoint de vue des

assureursassureurs

Point de vue desPoint de vue des

pompierspompiers

Point de vue desPoint de vue des

notairesnotaires

Point de vue desPoint de vue des

promoteurspromoteurs

Séparations des préoccupationsSéparations des préoccupations

système

Olivier Barais 30

Point de vue duPoint de vue du

propriétairepropriétairePoint de vue duPoint de vue du

plombierplombier

Point de vue de l'Point de vue de l'

électricienélectricien

Point de vue duPoint de vue du

maçonmaçonPoint de vue de l'Point de vue de l'

architectearchitecte

Point de vue duPoint de vue du

cadastrecadastre

système

Utile même pourUtile même pourdes systèmes des systèmes "moins" complexes"moins" complexes

Séparations des préoccupationsSéparations des préoccupations

Olivier Barais 31

ProblématiqueProblématique

Complexité croissante des logicielsComplexité croissante des logiciels Séparations des préoccupationsSéparations des préoccupations Séparations des métiersSéparations des métiers Multiplicité des besoinsMultiplicité des besoins Multiplicité des plateformesMultiplicité des plateformes Évolution permanenteÉvolution permanente

Logiciel =Logiciel = Code ? Code ? Est-ce la solution ?Est-ce la solution ?

Olivier Barais 32

Multiples modèles d'un même Multiples modèles d'un même systèmesystème

modèlesmodèlespour les architectespour les architectes

modèlesmodèlespour les notairespour les notaires

modèlesmodèlespour lespour lespompierspompiers

système

modèlesmodèlespour lepour lecadastrecadastre

modèlesmodèlespour l'assureurpour l'assureurcadastrecadastre

modèlesmodèlespour les paysagistespour les paysagistes

modèlesmodèlespour les électricienspour les électriciens

modèlesmodèlespour les plombierspour les plombiers

modèlesmodèlespour les promoteurspour les promoteurs

Olivier Barais 33

Ingénierie Dirigée par leIngénierie Dirigée par le CodeCode ouou

Ingénierie Dirigée par lesIngénierie Dirigée par les ModèlesModèles ? ?

Telle est la question...

Olivier Barais 34

En attente En attente du code

En vérification

En attente du montant

carte insérée

code frappémauvais code

En distribution

code bon

En attente retrait carte

carte retirée

montant correct

montant incorrect

billets retirés

le compte de P.le distrib. la banquepaul

retirer 500 fr

débiter 500 fr

la carte de P.

lire n° compte

la reserve

retirer 500 fr sur le compte n

sortir 5 billets de 100 frprendre la carte

confirmer

Client1..4 0..*

titulaires

Consortium

Compte

numérosolde...

1..*

0..1

1

Banque

numéronom

Distributeur0..*

1

0..*

1..*

signataire1

0..*CarteBleue

CoderetraitMax

1..*

AcceptéPar>

le compte de paul

le distributeur la banque

la carte de P.

la reserve

paul

4 : débiter(500)1 : retirer 500 fr

2 : lire n° de compte

5 : confirmer

6 : sortir 5 billet de 100 fr

3 : retirer 500 fr sur compte n

le compte de paulle compte de paul

le distributeur la banque

la carte de P.

la reserve

paul

4 : débiter(500)1 : retirer 500 fr

2 : lire n° de compte

5 : confirmer

6 : sortir 5 billet de 100 fr

3 : retirer 500 fr sur compte n

ClientClient

AssurerLaMaintenance

AssurerLaMaintenance

TechnicienTechnicien

AjouterDesBillets

AjouterDesBillets Transporteur

DeBillets

DistributeurDeBilletDistributeurDeBillet

FaireUnVirementFaireUnVirement

RetirerDeLArgentRetirerDeLArgent

ConsulterSonCompteConsulterSonCompte

système

Olivier Barais 35

Oui, Oui, Oui, mais ...Oui, Oui, Oui, mais ...

pour "l'informaticien moyen"pour "l'informaticien moyen"

la seule chose importante dans le la seule chose importante dans le logiciel c'est ...logiciel c'est ...

le Code !le Code !

Olivier Barais 36

En attente En attente du code

En vérification

En attente du montant

carte insérée

code frappémauvais code

En distribution

code bon

En attente retrait carte

carte retirée

montant correct

montant incorrect

billets retirés

le compte de P.le distrib. la banquepaul

retirer 500 fr

débiter 500 fr

la carte de P.

lire n° compte

la reserve

retirer 500 fr sur le compte n

sortir 5 billets de 100 frprendre la carte

confirmer

Client1..4 0..*

titulaires

Consortium

Compte

numérosolde...

1..*

0..1

1

Banque

numéronom

Distributeur0..*

1

0..*

1..*

signataire1

0..*CarteBleue

CoderetraitMax

1..*

AcceptéPar>

le compte de paul

le distributeur la banque

la carte de P.

la reserve

paul

4 : débiter(500)1 : retirer 500 fr

2 : lire n° de compte

5 : confirmer

6 : sortir 5 billet de 100 fr

3 : retirer 500 fr sur compte n

le compte de paulle compte de paul

le distributeur la banque

la carte de P.

la reserve

paul

4 : débiter(500)1 : retirer 500 fr

2 : lire n° de compte

5 : confirmer

6 : sortir 5 billet de 100 fr

3 : retirer 500 fr sur compte n

ClientClient

AssurerLaMaintenance

AssurerLaMaintenance

TechnicienTechnicien

AjouterDesBillets

AjouterDesBillets Transporteur

DeBillets

DistributeurDeBilletDistributeurDeBillet

FaireUnVirementFaireUnVirement

RetirerDeLArgentRetirerDeLArgent

ConsulterSonCompteConsulterSonCompte

class Compte {private Vector opérations ;private int découvertMax ;private Client titulaire ;

public int lireSolde() { int r = 0 ;for (i=0;opérations.size();i++)

r = r = opérations.at(i).montant ;return r ; }

public void débiter( int montant ) { if (montant<=0

|| lireSolde()-montant < lireSolde)throw new Exception() ;

...}

class Compte {private Vector opérations ;private int découvertMax ;private Client titulaire ;

public int lireSolde() { int r = 0 ;for (i=0;opérations.size();i++)

r = r = opérations.at(i).montant ;return r ; }

public void débiter( int montant ) { if (montant<=0

|| lireSolde()-montant < lireSolde)throw new Exception() ;

...}

class Compte {private Vector opérations ;private int découvertMax ;private Client titulaire ;

public int lireSolde() { int r = 0 ;for (i=0;opérations.size();i++)

r = r = opérations.at(i).montant ;return r ; }

public void débiter( int montant ) { if (montant<=0

|| lireSolde()-montant < lireSolde)throw new Exception() ;

...}

class Compte {private Vector opérations ;private int découvertMax ;private Client titulaire ;

public int lireSolde() { int r = 0 ;for (i=0;opérations.size();i++)

r = r = opérations.at(i).montant ;return r ; }

public void débiter( int montant ) { if (montant<=0

|| lireSolde()-montant < lireSolde)throw new Exception() ;

...}

?

??

?

?

Olivier Barais 37

SynthèseSynthèse Des Modèles plutôt que du CodeDes Modèles plutôt que du Code

– Un modèle est la simplification/abstraction de la réalitéUn modèle est la simplification/abstraction de la réalité– Nous construisons donc des modèles afin de mieux comprendre les Nous construisons donc des modèles afin de mieux comprendre les

systèmes que nous développonssystèmes que nous développons– Nous modélisons des systèmes complexes parce que nous somme Nous modélisons des systèmes complexes parce que nous somme

incapables de les comprendre dans leur totalitéincapables de les comprendre dans leur totalité– Le code ne permet pas de simplifier/abstraire la réalitéLe code ne permet pas de simplifier/abstraire la réalité

Pourquoi modéliserPourquoi modéliser

Olivier Barais 39

Why modeling: master complexityWhy modeling: master complexity Modeling, in the broadest sense, is the Modeling, in the broadest sense, is the cost-effective use of cost-effective use of

something in place of something else for some cognitive something in place of something else for some cognitive purposepurpose. It allows us to use something that is . It allows us to use something that is simplersimpler, , safersafer or or cheapercheaper than reality instead of reality for some purpose. than reality instead of reality for some purpose.

A model represents reality for the given purpose; the model A model represents reality for the given purpose; the model is an abstraction of reality in the sense that it cannot is an abstraction of reality in the sense that it cannot represent all aspects of reality. This allows us to deal with represent all aspects of reality. This allows us to deal with the world in a simplified manner, avoiding the complexity, the world in a simplified manner, avoiding the complexity, danger and irreversibility of reality.danger and irreversibility of reality.

Jeff RothenbergJeff Rothenberg..

Olivier Barais 40

Modeling in Science & EngineeringModeling in Science & Engineering A Model is a A Model is a simplifiedsimplified representation of an representation of an aspectaspect of the World for a specific of the World for a specific purposepurpose

M0(the world)

M1(modelingspace) Is represented by

Specificity of Engineering:Model something not yet existing (in order to build it)

Specificity of Engineering:Model something not yet existing (in order to build it)

+Applicant()+ApplicantInfo()+MakeApplication()

-companyName : CString-experience : CString-reference1 : CString-reference2 : CString-reference3 : CString

Applicant

+Person()+PersonInfo()

-personID : unsigned long-surname : CString-givenName : CString-middleInitial : char-streetAddress : CString-postCode : CString-countryname : CString-eMailAddress : CString

Person

-is taught by

1

-teaches

0..*+CourseSession()+CourseSessionInfo()

-courseSessionID : unsigned long-courseDate : unsigned long-courseID : unsigned long-courseLocation : CString

CourseSession

+AppStatus()+AppStatusInfo()

-statusCode : char-statusName : CString

AppStatus

+CourseRegistration()+CourseRegistrationInfo()

-registrationDate : unsigned long-completionFlag : bool-confirmedDate : unsigned long

CourseRegistration

+Test()+TestInfo()

-testScore : unsigned long

Test

+Application()+ApplicationInfo()

-productNr : unsigned long-certificationLevel : unsigned long-applicationDate : unsigned long

Application

+PermittedStatusChange()+StatusChangeInfo()

-fromStatus : char-toStatus : char

PermittedStatusChange

+ExamSession()+ExamSessionInfo()

-examSession : unsigned long-examlocation : CString-examDate : unsigned long

ExamSession

-gives0..*

-is achieved1

-is made by

1

-makes

0..*

-allows change in

0..*

-has a

1..*

-is taken by1

-takes0..*

-is made by a1

-made a1..*

-is in1

-is filled by0..*

-uses

1

-is used in

0..*

-applies to a0..*

-is for a1

+Exam()+ExamInfo()

-examID : unsigned long-certificationLevel : unsigned long

Exam

+Employee()+GetCurrentAge()+EmployeeInfo()

-jobType : CString-roomNr : unsigned long-department : CString-division : CString-jobTitle : CString-manager : unsigned long-headsDept : CString-headsDivision : CString-mobileNr : CString-birthDate : unsigned long

Employee

+registrationform()

RegistrationForm

-uses**

ApplicantApplicantList PersonList

findApplicant()

ApplicationRegForm

Applicant()

findPerson()

addPerson()

addApplication()

Application()

MakeApplication()

ApplicationList

Olivier Barais 41

UML: one model, 4 main UML: one model, 4 main dimensions, multiple viewsdimensions, multiple views

Device

start()stop()

Operator<<Actor>>

0..*1 0..*1

controls

Class diagram

: Operator

: Device

start( )

stop( )

Sequence diagram

UseCase diagram

: Operator

ControllingSite RemoteSite

TCP/IP : Device

Implementation diagram

: Operator

Olivier Barais 51

Model and Reality in SoftwareModel and Reality in Software Sun Tse: Sun Tse: Do not take the map for the realityDo not take the map for the reality MagritteMagritte

Software Software Models: from contemplative to productiveModels: from contemplative to productive

Olivier Barais 52

Modeling Modeling andand WeavingWeaving

DesignModel

Use CaseModel

SecurityModel

QoSModel Business

Model

ObjectModel

TestModel

UIModel

PlatformModel

CodeModel

tester

Challenges:-Product Families-Reuse of Weaving Process-Automatic Weaving

Challenges:-Product Families-Reuse of Weaving Process-Automatic Weaving

Olivier Barais 53

Assigning Meaning to ModelsAssigning Meaning to Models If a UML model If a UML model is no longer is no longer just just

– fancy pictures to decorate your roomfancy pictures to decorate your room– a graphical syntax for C++/Java/C#/Eiffel...a graphical syntax for C++/Java/C#/Eiffel...

Then tools must be able to manipulate Then tools must be able to manipulate modelsmodels– Let’s make a model Let’s make a model of what a model is!of what a model is!– => => meta-modelingmeta-modeling

» & meta-meta-modeling..& meta-meta-modeling..

ConstraintNamespace

Package

GeneralizableElement

0..*

0..*+supertype

{ordered}0..*

+subtype

0..*

Generalizes

Classifier

Class AssociationDataType

Feature

BehavioralFeature StructuralFeature

Operation

AssociationEnd

Reference

0..*

1

+referent0..*

+referencedEnd1

RefersTo

MofAttribute

ModelElement

0..*0..1

+containedElement

{ordered}

0..*

+container

0..1

Contains

0..*

1..*

0..*

+constrainedElement

1..*Constrains

Olivier Barais 54

UML2 UML2 meta-model meta-model

(part.)(part.)

Olivier Barais 55

Zoom: commentsZoom: comments

Olivier Barais 56

GeneralizationsGeneralizations

Olivier Barais 58

The 4 layers in practiceThe 4 layers in practice

Olivier Barais 59

Comparing Abstract Syntax SystemsComparing Abstract Syntax Systems

Pascal LanguageGrammar

A specificPascal Program

A specificexecution

of a Pascal program

EBNF MOF

The UMLmeta-Model

A Specificphenomenon

corresponding toa UML Model

A SpecificUML Model

Technology #2(MOF + OCL)

M3

M2

M1

Technology #1(formal grammars

attribute grammars,etc.)

A XMLdocument

A XML DTDOr Schema

A XMLdocument

A XML DTDor Schema

Technology #3(XML Meta-Language)

KIFTheories

Upper LevelOntologies

Technology #4(Ontology engineering)

[XMI=MOF+XML+OCL]

+DescriptionLogics

+ConceptualGraphs+etc.

+ Xlink, Xpath, XSLT+ RDF, OIL, DAML+ etc.

(From J. Bézivin)

Olivier Barais 60

MDA: the OMG new visionMDA: the OMG new vision

"OMG is in the ideal position to "OMG is in the ideal position to provide the model-provide the model-based standardsbased standards that are necessary to that are necessary to extend extend integration beyond the middleware approachintegration beyond the middleware approach… Now … Now is the time to put this plan into effect. is the time to put this plan into effect. Now is the time Now is the time for the Model Driven Architecturfor the Model Driven Architecturee."."

Richard Soley & OMG staff, Richard Soley & OMG staff,

MDA Whitepaper Draft 3.2MDA Whitepaper Draft 3.2

November 27, 2000November 27, 2000

Olivier Barais 61

Mappings to multiple and Mappings to multiple and evolving platformsevolving platforms

COM+DCOM

CORBA C#.Net

XMLSOAP

JavaEJB HTTP

HTML

MOF & UML as the core

Organization assets expressed as models

Model transformations to map to technology specific platforms

Platform neutral models basedon UML & MOF

Platform neutral models basedon UML & MOF

Olivier Barais 62

The core idea of MDA:The core idea of MDA:PIMs & PSMsPIMs & PSMs

MDA modelsMDA models– PIMPIM: Platform Independent Model: Platform Independent Model

» Business Model of a system abstracting away the Business Model of a system abstracting away the deployementdeployement details of a systemdetails of a system

» Example: the UML model of the GPS systemExample: the UML model of the GPS system– PSMPSM: Platform Specific Model: Platform Specific Model

» Operational model including platform specific aspectsOperational model including platform specific aspects» Example: the UML model of the GPS system on .NETExample: the UML model of the GPS system on .NET

Possibly expressed with a UML profile (.NET profile for UML)Possibly expressed with a UML profile (.NET profile for UML)

– Not so clear about platform modelsNot so clear about platform models » Reusable model at various levels of abstractionReusable model at various levels of abstraction

CCM, C#, EJB, EDOC, …CCM, C#, EJB, EDOC, …

Olivier Barais 63

A PSM for our GPSA PSM for our GPS Here, the platform is .NETHere, the platform is .NET

Olivier Barais 64

How to go From PIM to PSM?How to go From PIM to PSM?

"just" weave the platform aspect !"just" weave the platform aspect ! How can I do that?How can I do that?

– Through Model transformationsThrough Model transformations– Now hot topic at OMG with RFP Q/V/TNow hot topic at OMG with RFP Q/V/T

» Query/View/TransformationQuery/View/Transformation

Olivier Barais 65

Weaving aspects into UML Weaving aspects into UML Models?Models?

It’s what Model Driven Architecture is about!It’s what Model Driven Architecture is about!

Requirements Analysis ArchitecturalDesign

DetailedDesign

Implementation Validation

Lifecycle

Modelingpoint of views

Proofs,QoS

Analysis,Simulation

Technical Aspects

BusinessAspects

Text(e.g. XML)

PIMPIM

Doc Doc Doc Doc Doc Doc

PIMPIM

PIMPIM

PSMPSM

PIMPIM

PSMPSM

CodeCode

TestsTests

Formal Models Formal Models

Endomorphic TransformationsExomorphic TransformationsOutside UML scope

PIM=Platform Independent ModelPSM= Platform Specific Model

Olivier Barais 66

But many more dimensions in But many more dimensions in modeling!modeling!

Beyond Design Model Beyond Design Model – where UML is arguably good…where UML is arguably good…

Business modelBusiness model GUI modelGUI model Development process modelDevelopment process model Performance & Resource modelPerformance & Resource model Deployment modelDeployment model Test modelTest model Etc.Etc.

Olivier Barais 67

How to take these dimensions How to take these dimensions into account?into account?

Expression with eitherExpression with either– UML, usUML, usinging built-in extension built-in extension

mechanisms to link with other mechanisms to link with other semantic domainssemantic domains

– DS(M)L, based on MOFDS(M)L, based on MOF ExploitationExploitation

– Weave all these aspects Weave all these aspects into a design modelinto a design model

– Define Define mappingsmappings between these between these aspects (as in e.g. aspects (as in e.g. federationsfederations))

Servicemodel

Domainmodel

Resourcemodel

DesignModel

Testmodel

Installationmodel

Olivier Barais 68

Element stereotype

Design pattern application(parametric collaboration)

Historycmd_executed : string

last_command() : string

<<persistent>>

ServiceProvider

<<command>> action_1()<<command>> action_2()<<command>> action_3()

Interpreter

execute()

0..*

1

0..*

11..*1..* 1..*

+invoker

1..*

…and alsoTagged values& Contracts

Command pattern

receiver

invoker

Embedding implicit semantics Embedding implicit semantics into a modelinto a model

Olivier Barais 69

……and the resultand the result we we wanwantt......

Historycmd_executed : string

last_command() : string

<<persistent>>

History_StorageProxy

load_last_command() : stringstore_last_command(cmd : string)

action_1_cmd

do()

action_2_cmd

do()

action_3_cmd

do()

ServiceProvider

<<command>> action_1()<<command>> action_2()<<command>> action_3()

+proxy

Interpreter

execute()

0..*

1

0..*

1

ServiceProvider_Command

do() 10..*

+cmdTarget

10..*1..*0..* 1..*

+commands

0..*

Olivier Barais 70

0..*

1

Historycmd_executed : string

last_command() : string

<<persistent>>

Interpreter

execute()

ServiceProvider

<<command>> action_1()<<command>> action_2()<<command>> action_3()

History_StorageProxy

load_last_command() : stringstore_last_command(cmd : string)

Persitence implementation

1..*0..*

ServiceProvider_Command

do() 10..*

action_3_cmd

do()

action_2_cmd

do()

action_1_cmd

do()

Command pattern implementation

How To: Automatic Model How To: Automatic Model TransformationsTransformations

In some domains (e.g.; RT systems)transformations can get more complexthan initial model!=> must be managed with sound SE principles

Olivier Barais 71

Why complex transformations?Why complex transformations? Example: Air Traffic ManagementExample: Air Traffic Management

– ““business model” quite stable & not that complexbusiness model” quite stable & not that complex Various modeling languages used beyond UMLVarious modeling languages used beyond UML

– As many points of views as stakeholdersAs many points of views as stakeholders Deliver software for (many) variants of a platformDeliver software for (many) variants of a platform

– Heterogeneity is the ruleHeterogeneity is the rule Reuse technical solutions across large product Reuse technical solutions across large product

lines (e.g. fault tolerance, security…)lines (e.g. fault tolerance, security…) Customize generic transformationsCustomize generic transformations Compose reusable transformationsCompose reusable transformations Evolve & maintain transformations for 15+ years!Evolve & maintain transformations for 15+ years!

Olivier Barais 72

The 3 ages of TransformationsThe 3 ages of Transformations Awk-like (inc. Awk-like (inc. sed, perlsed, perl…) …)

XSLTXSLT– W3C standard for transforming XMLW3C standard for transforming XML– Operates on tree structuresOperates on tree structures– syntactical & inefficientsyntactical & inefficient

QVT-likeQVT-like– Now hot topic at OMG with RFP Q/V/TNow hot topic at OMG with RFP Q/V/T

» Query/View/TransformationQuery/View/Transformation– Operates on graphsOperates on graphs

BEGIN {action)pattern #1 {action #1}…pattern #n {action #n}END {action)

SE Limit: ~100 LOC

SE Limit: ~1000 LOC

SE Limit: ?Standard?Which/When?

Olivier Barais 73

Transformations are AssetsTransformations are Assets=> apply sound SE principles=> apply sound SE principles

Must be ModeledMust be Modeled– with the UML, using the power of OOwith the UML, using the power of OO

Must be DesignedMust be Designed– Design by Contract, using OCLDesign by Contract, using OCL

Must be ImplementedMust be Implemented– Made available through libraries of components, frameworks…Made available through libraries of components, frameworks…

Must be TestedMust be Tested– test casestest cases

» input: a UML Modelinput: a UML Model» output: a UML Model, + contract checkingoutput: a UML Model, + contract checking

Must be Evolved Must be Evolved – Items of Configuration ManagementItems of Configuration Management– Transformations of transformationsTransformations of transformations

Olivier Barais 74

PrinciplesPrinciples

1.1. Everything relevant to the development Everything relevant to the development process is a modelprocess is a model

2.2. All the meta-models All the meta-models can becan be written in a written in a language of a unique meta-meta-modellanguage of a unique meta-meta-model

» Same M3 helps Interoperability of tools defined at M2Same M3 helps Interoperability of tools defined at M2

3.3. A development process can be modelled A development process can be modelled as a partially ordered set of model as a partially ordered set of model transformations, that take models as input transformations, that take models as input and produce models as outputand produce models as output

Olivier Barais 75

ConsequencesConsequences

1.1. Models are aspect oriented. Conversely: Models are aspect oriented. Conversely: Aspects are modelsAspects are models

2.2. Transformations are Transformations are aspects weavers. aspects weavers. They can be modeled.They can be modeled.

3.3. Every meta-model defines a domain Every meta-model defines a domain specific languagespecific language

4.4. Software development has two Software development has two dimensions: M1-model development and dimensions: M1-model development and M2-transformation developmentM2-transformation development

Olivier Barais 76

ChallengesChallenges Language definition problems (Q/V/T)Language definition problems (Q/V/T)

– Expressive, easy to use language(s) for transformationsExpressive, easy to use language(s) for transformations Technological IssuesTechnological Issues

– Tool set, interoperability…Tool set, interoperability… Software Engineering IssuesSoftware Engineering Issues

– From requirements to tests and SCMFrom requirements to tests and SCM Theoretical issuesTheoretical issues

– A transformation is a A transformation is a mappingmapping between semantic domains between semantic domains» Beyond Code Generation: Proof, Performance Evaluation, Test Beyond Code Generation: Proof, Performance Evaluation, Test

Synthesis …Synthesis …– Compositional weaving, cf. Aspect Oriented Software DCompositional weaving, cf. Aspect Oriented Software Dvpvp

Olivier Barais 77

UML & UML & MModelodel D Driven riven Architecture: SummaryArchitecture: Summary

Modeling to master complexityModeling to master complexity– Multi-dimensional and aspect oriented by definitionMulti-dimensional and aspect oriented by definition

Models: from contemplative to productiveModels: from contemplative to productive– Meta-modeling toolsMeta-modeling tools

Model Driven EngineeringModel Driven Engineering– Weaving aspects into a design modelWeaving aspects into a design model

» E.g. Platform SpecificitiesE.g. Platform Specificities

Model Driven Architecture (PIM / PSM): just a Model Driven Architecture (PIM / PSM): just a special case of Aspect Oriented Designspecial case of Aspect Oriented Design

Related: Generative Prog, Software FactoriesRelated: Generative Prog, Software Factories

Olivier Barais 78

Preview of the rest of this coursePreview of the rest of this course Meta-ModelingMeta-Modeling

– Meta-models & abstract syntaxesMeta-models & abstract syntaxes– Static Semantics with OCLStatic Semantics with OCL– Dynamic Semantics with KermetaDynamic Semantics with Kermeta

Model Driven Specification LanguageModel Driven Specification Language– The logo case studyThe logo case study

Model Transformations & code generationModel Transformations & code generation Applications of MDEApplications of MDE

– Design Pattern ApplicationDesign Pattern Application– Model RefactoringModel Refactoring– Product Line Modeling and DerivationProduct Line Modeling and Derivation