This thesis introduces enhancements over an existing Component Oriented CASE Tool (CoseCase). Constraint checking facility is implemented for the connections provided in the tool: A user programmable set of rules governing the allowed connections among different modeling elements is added as a capability. The previous implementation of the tool did not consider the semantics behind the elements and their interconnection. Also related connection types are tested against cycle formations. Other aspects of the tool have been enhanced such as the dynamic graphical presentation of connection handles and connectors. Deleting a sub-tree from the design diagram is made operational besides the correction of faulty operating routines especially related to attaching new elements to the diagram.

Component Oriented Software Engineering (COSE) seems to be the future of software engineering. Currently, COSEML is the only modeling language that completely supports the COSE approach. Abstract decomposition of the system and their representing components are shown in a hierarchy diagram to support the COSE process model. In COSEML, only static modeling is supported through this single diagram. However, software is about behavior and static modeling is not sufficient to describe the system. The aim of this thesis is providing the benefits of dynamic modeling to COSEML by adopting collaboration diagrams. For this purpose, first, specification of modified collaboration diagrams is made for COSEML. Then software is developed for supporting collaboration diagrams in COSECASE. Also, an e-store application is modeled with COSEML using the collaboration diagrams. With this work, modeling the dynamic behavior of the system in both abstract and component levels is made possible. Furthermore, use case realization is enabled in the COSE modeling. More important, modeling the sequential interactions among components is made possible. Consequently, a suitable environment is provided for automated testing and application generation from the model.

A component-oriented tool for hierarchical modeling of instructional designs is developed. The motivation is to show that hierarchical representation of instructional designs is easier, better and more effective for modeling. Additionally a modeling language is developed to provide an effective, flexible and easy to use integration model in which all teaching components are discovered, defined and connected. In order to fulfill the above purposes an abstract notation is developed that is sufficiently general and adapting top-down hierarchic approach to represent Units of Learning (UoL), Operational Knowledge Units (OKU), Learning Objects (LO), and Learning Components (LC) with respect to the common structures found in different instructional models. COSEML, a top-down hierarchic, and component oriented modeling language has been used as a reference and the core concept in developing the Educational Component Oriented Modeling Language (ECOML). The high-level architecture of ECOML provides the means for designing instructional structures. It describes how LOs, UoLs, OKUs and LCs are sequenced in a certain context or knowledge domain. The resulting model can be reused in different contexts and across different educational platforms.

A Component Oriented Software Engineering (COSE) modeling environment is enhanced with the capability to import executable components and deliver applications through their composition. For this purpose, an interface layer that utilizes JAR libraries as components has been developed. Also, Domain Engineering process has been applied to Geographical Information Systems (GIS) domain and utilized towards converting the environment to a development framework. The interface layer imports JAR libraries into the COSECASE tool, which is a graphical tool supporting COSE approach and COSE Modeling Language (COSEML). As a result, systems can be designed using abstractions and then implemented by corresponding deployed components. Imported code is made available to the COSECASE environment through this interface layer. Also, Domain Analysis, Domain Design, and Domain Implementation phases of Domain Engineering process have been applied to the GIS domain. Components developed in this Domain Implementation phase have been imported into COSECASE. A simple GIS application has been designed and generated through the interface layer of COSECASE for demonstration purposes.

Nos últimos anos, computação em grade tem emergido como uma forma de utilização de recursos geograficamente distribuídos em múltiplas organizações. Devido ao fato de grids serem altamente distribuídos e compostos por recursos heterogêneos, a computação em grade tem dado importância a requisitos específicos, como escalabilidade, desempenho e a necessidade de um modelo de programação adequado. Vários modelos de programação já foram propostos para a computação em grade. Entretanto, ate agora, nenhum deles supriu todos os requisitos. Diferentemente, na área de alto desempenho em clusters, o modelo de passagem de mensagens se tornou um verdadeiro padrão com um grande número de bibliotecas e aplicações legadas. Este trabalho propõe um framework híbrido que combina os altos desempenho e aceitação do padrão MPI, melhorado com extensões intuitivas para permitir aos desenvolvedores o projeto e desenvolvimento de aplicações em grade ou a gridi-ficação de aplicações já existentes, com a flexibilidade de um runtime baseado em componentes, modelando uma hierarquia de recursos e suportando a comunicação entre clusters. A solução proposta se baseia na adição de comunicadores MPI e uma API relacionada, a qual oferece um suporte ao desenvolvimento de aplicações que levam em conta a topologia hierárquica de grades computacionais, adequado a desenvolvedores habituados a MPI. características (Simula_c~ao Baseada no Algoritmo de Monte Carlo, Mergesort e um solver Poisson3D) mostraram que a gridificação pode melhorar consideravelmente o desempenho dessas aplicações em ambientes de grade. Ainda que o objetivo deste trabalho não seja competir com distribuições MPI existentes, o desempenho da solução proposta _e comparável ao desempenho de MPI, sendo melhor em alguns casos. A partir dos resultados obtidos com o protótipo apresentado, é possível concluir que o custo adicionado pela utilização de componentes não é desprezível, mas dentro do esperado. Entretanto, espera-se que os benefícios para aplicações de grade devem superar os custos adicionais. Além disso, as extensões a interface MPI oferecem a usuários as abstrações necessárias ao projeto de algoritmos paralelos de forma hierárquica, visando ambientes de grade. / In the past several years, grid computing has emerged as a way to harness computing resources geographically distributed across multiple organizations. Due to its inherently largely distributed and heterogeneous nature, grid computing has enlarged the importance of specific requirements, such as scalability, performance and the need of an adequate programming model. Several programming models have been proposed for grid programming. Nonetheless, so far, none of them met all the requirements. Differently, in the field of high performance cluster computing, the message passing model became a true standard with a large number of libraries and legacy applications. This work proposes a hybrid framework that combines the high performance and high acceptability of the MPI standard boosted with intuitive extensions to enable developers to design grid applications or "gridify" existing ones with the flexibility of a component-based runtime modeling resources hierarchy and offering support to inter-cluster communication. The proposed solution relies on the addition of new MPI communicators and a related API, which may offer a support well-suited to programmers used to MPI in order to reflect a hierarchical topology within the deployed application. Carlo Simulation, a Mergesort and a Poissond3D solver) have shown that the "gridification" of applications improve application performance on grid environments. Even if the goal is not to compete against existing MPI distributions, the performance of the solution is comparable with MPI performance, even better in some cases. From the results obtained in the evaluation of this prototype, we conclude that the overhead introduced by the components is not negligible, but inside of the expected. However, we can expect the benefits to grid applications to bypass the generated overhead. Besides, the extended interface may offer users the adequate abstractions to design parallel algorithms in a hierarchical way addressing grid environments.

It is widely believed and empirically shown that component reuse improves both the quality and productivity of software development. This brings the necessity of a graphical editor to model the projects by using components. A graphical editor was implemented for the development of Component Oriented software development. The editor facilitates modeling efforts through application of the graphical modeling language COSEML. Both design patterns and software components have come to play important roles in software development. The correlation between software components and design patterns is apparent. In the design phase of the projects design patterns are used widely both in component and object oriented projects. Design patterns can be used as super components in component-based development . Software reuse, software components, design patterns, use of design patterns in component-based development, and component architectures are studied in details to address the need for the approach. COSE modeling activity starts with the decomposition of the target system into building blocks in a top-down order. Next, interfaces between these blocks are defined. If required design patterns can be added to model as super components.

In order to use components that are developed for a domain, domain knowledge is required. If the default relations in a domain are offered by a framework, this can be a starting point for the application engineer as an important kind of domain knowledge. A generic design for creating and saving a domain is implemented in this thesis. This approach starts with creating a domain from components and relations among these components. The relations and components are saved once and used several times. In addition, this generic design helps for code generation by using components. A framework for this design is implemented and applied for GIS domain. A basic code generation approach is also implemented in this framework for demonstration purposes. This framework can be used by domain engineers in order to create a domain and by application engineers to develop custom applications. It has the ability to offer default relations and helps creating new relations between components. Parameters and sequence of function calls can be defined by using a GUI. All the relations including default and userdefined ones can be used for code generation. COSECASE, which offers a tool for component-oriented design is extended with domain operations such as creating domain, saving domain, loading domain, and generating domain code. As the starting point, domain analysis for GIS domain is completed to define the domain. Then the components that have been implemented for GIS domain and relations between these components are saved within the framework. Moreover, some basic applications are generated by using this framework in the GIS domain. Also a sample domain is created to prove that our approach can be applied to any domain. The relations in this sample domain are saved in the framework and same basic applications are generated.

Nos últimos anos, computação em grade tem emergido como uma forma de utilização de recursos geograficamente distribuídos em múltiplas organizações. Devido ao fato de grids serem altamente distribuídos e compostos por recursos heterogêneos, a computação em grade tem dado importância a requisitos específicos, como escalabilidade, desempenho e a necessidade de um modelo de programação adequado. Vários modelos de programação já foram propostos para a computação em grade. Entretanto, ate agora, nenhum deles supriu todos os requisitos. Diferentemente, na área de alto desempenho em clusters, o modelo de passagem de mensagens se tornou um verdadeiro padrão com um grande número de bibliotecas e aplicações legadas. Este trabalho propõe um framework híbrido que combina os altos desempenho e aceitação do padrão MPI, melhorado com extensões intuitivas para permitir aos desenvolvedores o projeto e desenvolvimento de aplicações em grade ou a gridi-ficação de aplicações já existentes, com a flexibilidade de um runtime baseado em componentes, modelando uma hierarquia de recursos e suportando a comunicação entre clusters. A solução proposta se baseia na adição de comunicadores MPI e uma API relacionada, a qual oferece um suporte ao desenvolvimento de aplicações que levam em conta a topologia hierárquica de grades computacionais, adequado a desenvolvedores habituados a MPI. características (Simula_c~ao Baseada no Algoritmo de Monte Carlo, Mergesort e um solver Poisson3D) mostraram que a gridificação pode melhorar consideravelmente o desempenho dessas aplicações em ambientes de grade. Ainda que o objetivo deste trabalho não seja competir com distribuições MPI existentes, o desempenho da solução proposta _e comparável ao desempenho de MPI, sendo melhor em alguns casos. A partir dos resultados obtidos com o protótipo apresentado, é possível concluir que o custo adicionado pela utilização de componentes não é desprezível, mas dentro do esperado. Entretanto, espera-se que os benefícios para aplicações de grade devem superar os custos adicionais. Além disso, as extensões a interface MPI oferecem a usuários as abstrações necessárias ao projeto de algoritmos paralelos de forma hierárquica, visando ambientes de grade. / In the past several years, grid computing has emerged as a way to harness computing resources geographically distributed across multiple organizations. Due to its inherently largely distributed and heterogeneous nature, grid computing has enlarged the importance of specific requirements, such as scalability, performance and the need of an adequate programming model. Several programming models have been proposed for grid programming. Nonetheless, so far, none of them met all the requirements. Differently, in the field of high performance cluster computing, the message passing model became a true standard with a large number of libraries and legacy applications. This work proposes a hybrid framework that combines the high performance and high acceptability of the MPI standard boosted with intuitive extensions to enable developers to design grid applications or "gridify" existing ones with the flexibility of a component-based runtime modeling resources hierarchy and offering support to inter-cluster communication. The proposed solution relies on the addition of new MPI communicators and a related API, which may offer a support well-suited to programmers used to MPI in order to reflect a hierarchical topology within the deployed application. Carlo Simulation, a Mergesort and a Poissond3D solver) have shown that the "gridification" of applications improve application performance on grid environments. Even if the goal is not to compete against existing MPI distributions, the performance of the solution is comparable with MPI performance, even better in some cases. From the results obtained in the evaluation of this prototype, we conclude that the overhead introduced by the components is not negligible, but inside of the expected. However, we can expect the benefits to grid applications to bypass the generated overhead. Besides, the extended interface may offer users the adequate abstractions to design parallel algorithms in a hierarchical way addressing grid environments.

Nos últimos anos, computação em grade tem emergido como uma forma de utilização de recursos geograficamente distribuídos em múltiplas organizações. Devido ao fato de grids serem altamente distribuídos e compostos por recursos heterogêneos, a computação em grade tem dado importância a requisitos específicos, como escalabilidade, desempenho e a necessidade de um modelo de programação adequado. Vários modelos de programação já foram propostos para a computação em grade. Entretanto, ate agora, nenhum deles supriu todos os requisitos. Diferentemente, na área de alto desempenho em clusters, o modelo de passagem de mensagens se tornou um verdadeiro padrão com um grande número de bibliotecas e aplicações legadas. Este trabalho propõe um framework híbrido que combina os altos desempenho e aceitação do padrão MPI, melhorado com extensões intuitivas para permitir aos desenvolvedores o projeto e desenvolvimento de aplicações em grade ou a gridi-ficação de aplicações já existentes, com a flexibilidade de um runtime baseado em componentes, modelando uma hierarquia de recursos e suportando a comunicação entre clusters. A solução proposta se baseia na adição de comunicadores MPI e uma API relacionada, a qual oferece um suporte ao desenvolvimento de aplicações que levam em conta a topologia hierárquica de grades computacionais, adequado a desenvolvedores habituados a MPI. características (Simula_c~ao Baseada no Algoritmo de Monte Carlo, Mergesort e um solver Poisson3D) mostraram que a gridificação pode melhorar consideravelmente o desempenho dessas aplicações em ambientes de grade. Ainda que o objetivo deste trabalho não seja competir com distribuições MPI existentes, o desempenho da solução proposta _e comparável ao desempenho de MPI, sendo melhor em alguns casos. A partir dos resultados obtidos com o protótipo apresentado, é possível concluir que o custo adicionado pela utilização de componentes não é desprezível, mas dentro do esperado. Entretanto, espera-se que os benefícios para aplicações de grade devem superar os custos adicionais. Além disso, as extensões a interface MPI oferecem a usuários as abstrações necessárias ao projeto de algoritmos paralelos de forma hierárquica, visando ambientes de grade. / In the past several years, grid computing has emerged as a way to harness computing resources geographically distributed across multiple organizations. Due to its inherently largely distributed and heterogeneous nature, grid computing has enlarged the importance of specific requirements, such as scalability, performance and the need of an adequate programming model. Several programming models have been proposed for grid programming. Nonetheless, so far, none of them met all the requirements. Differently, in the field of high performance cluster computing, the message passing model became a true standard with a large number of libraries and legacy applications. This work proposes a hybrid framework that combines the high performance and high acceptability of the MPI standard boosted with intuitive extensions to enable developers to design grid applications or "gridify" existing ones with the flexibility of a component-based runtime modeling resources hierarchy and offering support to inter-cluster communication. The proposed solution relies on the addition of new MPI communicators and a related API, which may offer a support well-suited to programmers used to MPI in order to reflect a hierarchical topology within the deployed application. Carlo Simulation, a Mergesort and a Poissond3D solver) have shown that the "gridification" of applications improve application performance on grid environments. Even if the goal is not to compete against existing MPI distributions, the performance of the solution is comparable with MPI performance, even better in some cases. From the results obtained in the evaluation of this prototype, we conclude that the overhead introduced by the components is not negligible, but inside of the expected. However, we can expect the benefits to grid applications to bypass the generated overhead. Besides, the extended interface may offer users the adequate abstractions to design parallel algorithms in a hierarchical way addressing grid environments.

No description available.