[pt] Injeção de dependências, uma forma de inversão de controle, é uma forma de estruturar a configuração e composição de componentes de software que traz vários benefícios como um acoplamento reduzido entre componentes. No entanto, um framework genérico de injeção de dependências requer instrospecção em tempo de execução, o que explica por que injeção de dependências é popular em Java mas praticamente inexistente em C Mais Mais. Neste trabalho apresentamos um sistema de introspecção para C Mais Mais 11 e mostramos como ele pode ser usado para melhorar uma implementação de Service Component Architecture (SCA) para C Mais Mais. Usamos vários novas funcionalidades de C Mais Mais 11 como perfect forwarding, variadic templates e lvalue references para melhorar a usabilidade da API de reflexão e minimizar o overhead de execução. / [en] Dependency injection, a form of inversion of control, is a way of structuring the configuration and composition of software components that brings many benefits such as a loose coupling of components. However, a generic dependency injection framework requires runtime type introspection and this is why dependency injection is popular in Java and almost non-existent in C plus plus. In this work we present a introspection system for C plus plus 11 and show how to use it to improve an implementation of the Service Component Architecture (SCA) for C plus plus. It uses several features of C plus plus 11 such as perfect forwarding, variadic templates and lvalue references to improve usability and minimize overhead.

Model Driven Development is used to improve software quality and efficiency by automatically transforming abstract and formal models into software implementations. This is particularly sensible if the model’s integrity can be proven formally and is preserved during the model’s transformation. A standard to specify software model integrity is the Object Constraint Language (OCL). Another topic of research is the dynamic development of software components, enabling software system composition at component run-time. As a consequence, the system’s verification must be realized during system run-time (and not during transformation or compile time). Many established verification techniques cannot be used for run-time verification. A method to enable model-based run-time verification will be developed during this work. How OCL constraints can be transformed into executable software artifacts and how they can be used in the component-based system Treaty will be the major task of this diploma thesis. / Modellgetriebene Entwicklung dient der Verbesserung von Qualität und Effizienz in der Software-Entwicklung durch Automatisierung der notwendigen Transformationen von abstrakten bzw. formalen Modellen bis zur Implementierung. Dies ist insbesondere dann sinnvoll, wenn die Integrität der ursprünglichen Modelle formal bewiesen werden kann und durch die Transformation gewährleistet wird. Ein Standard zur Spezifikation der Integrität von Softwaremodellen ist die Object Constraint Language (OCL). Eine weitere Forschungsrichtung im Software-Engineering ist die Entwicklung von dynamischen Komponenten-Modellen, die die Komposition von Softwaresystemen im laufenden Betrieb ermöglichen. Dies bedeutet, dass die Systemverifikation im laufenden Betrieb realisiert werden muss. Die meisten der etablierten Verifikationstechniken sind dazu nicht geeignet. In der Diplomarbeit soll ausgehend von diesem Stand der Technik eine Methode zur modellbasierten Verifikation zur Laufzeit entwickelt werden. Insbesondere soll untersucht werden, wie OCL-Constraints zur Laufzeit in ausführbare Software-Artefakte übersetzt und in dem komponentenbasierten System Treaty verwendet werden können.

El objetivo de la tesis es definir una metodología de desarrollo de aplicaciones de tiempo real basadas en componentes, orientada a aplicaciones cuyos requisitos temporales se especifican utilizando un modelo reactivo de comportamiento temporal. La metodología se construye en base a extensiones que incorporan a las especificaciones, modelos de referencia y procesos estándares propios de la ingeniería de componentes convencionales, esto es, sin requisitos temporales, los datos y los procesos necesarios para la especificación, diseño y análisis de los aspectos relativos al comportamiento temporal. La metodología se sustenta en cuatro contribuciones principales:- Se propone la metodología de modelado modular del comportamiento temporal Mod-MAST, que permite construir el modelo de una aplicación basada en componentes por composición de los modelos de los componentes que la forman. - Se propone la extensión RT-D&C de la especificación Deployment and Configuration of Component-based Distributed Applications de OMG, que permite incluir metadatos relativos a comportamiento temporal en los descriptores de componentes, plataformas de ejecución y aplicaciones. - Se especifica la tecnología de componentes RT-CCM como una extensión de la especificación estándar Lightweight CCM de OMG, que añade los mecanismos necesarios para desarrollar aplicaciones con comportamiento temporal predecible.- Se propone la tecnología de componentes Ada-CCM como implementación concreta de RT-CCM basada en el lenguaje de programación Ada 2005.Todos estos elementos se integran en un proceso completo de diseño de tiempo real de aplicaciones basadas en componentes. / The objective of this work is to define a methodology for the development of real-time component-based applications, focused on applications whose timing requirements are specified according to a reactive model of the timing behaviour. The methodology is built through a set of extensions that incorporate to the standard specifications, reference models and processes typical from the conventional components engineering, i.e. components without timing requirements, the data structures and the processes required for the specification, design and analysis of the aspects related to timing behaviour. The methodology relies on four main contributions:- The Mod-MAST modular modelling methodology, which allows building the real-time model of a component-based application by composing the models of the components that form it.- The RT-D&C extension of the Deployment and Configuration of Component-based Distributed Applications Specification of the OMG, which allows including metadata related to timing behaviour in the descriptors of components, execution platforms and applications.- The RT-CCM components technology, which is an extension of the standard Lightweight CCM Specification of the OMG that incorporates mechanisms to develop applications with predictable timing behaviour.- The Ada-CCM components technology has been developed. It is an implementation of the RT-CCM technology based on the Ada 2005 programming language.All these elements have been integrated in a complete real-time design process for component-based applications.

Cette thèse aborde les formalismes pour la modélisation multi-physique en support au cycle en V deconception. Ce travail a été réalisé dans le cadre du projet ANR–MoCoSyMec, selon la méthodologie duprototypage virtuel fonctionnel (PVF) et illustré sur des systèmes électromagnétiques.Nous nous sommes principalement intéressés au langage VHDL-AMS, en tant que support aux différentsniveaux de modélisation apparaissant dans le cycle en V de conception. Cela nous a conduits à traiter laportabilité et l’interopérabilité en VHDL-AMS de diverses méthodes et outils de modélisation. Nous avonsproposé et validé, via le formalisme des composants logiciels ICAr, des solutions aux limites de l’utilisation deVHDL-AMS pour modéliser certains phénomènes physiques reposants sur des calculs numériques.Nous avons étendu la norme ICAr pour supporter des modèles dynamiques décrits par des équationsdifférentielles algébriques (DAE) ; et pour des besoins de co-simulation, nous pouvons également y associer unsolveur. Ces développements sont désormais capitalisés dans le framework CADES.Enfin, nous avons proposé une architecture pour le portage de modèles d’un formalisme à un autre. Elle a étédéfinie et mise en oeuvre plus particulièrement pour des modèles magnétiques réluctants (Reluctool) et desMEMS magnétiques (MacMMems) vers le VHDL-AMS.Ces formalismes et méthodologies sont mis en oeuvre autour du PVF d’un contacteur électromagnétique. / This PhD report deals with modeling formalisms for multi-physical systems in the design V- cycle. Thiswork was carried out within the French ANR-MoCoSyMec project, according to the methodology of functionalvirtual prototyping (PVF) and illustrated with electromagnetical systems.The work focuses on the VHDL-AMS modeling language, as a support for several modeling levels appearingin the design V-cycle. In this work, the portability and interoperability problems have been studied, usingVHDL-AMS, for various modeling methods and tools. Solutions have been proposed and validated for use limitsof VHDL-AMS language, specifically for the modeling of some physical phenomena using numericalcomputations, through the software component formalism called ICAr.The ICAr software component standard has been extended to support dynamic models described throughdifferential algebraic equations (DAE). It has also been extended for co-simulation purposes in which a solver isassociated to the dynamic model inside the ICAr component. These developed solutions are now available in theframework CADES.Finally, architecture has been proposed for the transforming of models from a professional formalism intoanother, specifically into VHDL-AMS. It has been designed and implemented for reluctant magnetic models(RelucTool) and magnetic MEMS (MacMMems).These formalisms and methodologies are implemented around the functional virtual prototyping (PVF) of anelectromagnetic contactor.