• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 82
  • 18
  • 4
  • 3
  • 2
  • 2
  • 1
  • 1
  • 1
  • Tagged with
  • 137
  • 137
  • 60
  • 49
  • 29
  • 21
  • 19
  • 18
  • 15
  • 15
  • 15
  • 14
  • 13
  • 11
  • 11
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
121

[pt] COMPLETUDE DE REFATORAÇÕES COMPOSTAS DE CÓDIGO-FONTE PARA A REMOÇÃO BENÉFICA DE ANOMALIAS DE CÓDIGO / [en] ON THE COMPLETENESS OF COMPOSITE CODE REFACTORINGS FOR BENEFICIAL SMELL REMOVA

ANA CARLA GOMES BIBIANO 22 June 2023 (has links)
[pt] A refatoração de código é uma transformação de código que visa aprimorar a estrutura interna do código. Uma refatoração isolada raramente é suficiente para remover completamente uma estrutura de código ruim, como uma anomalia de código. Os desenvolvedores então aplicam refatorações compostas para remover totalmente uma anomalia de código. Uma refatoração composta consiste em duas ou mais refatorações inter-relacionadas. Um refatoração composta é considerada completa quando elimina totalmente a anomalia de código alvo. Estudos relatam que os desenvolvedores geralmente falham em remover completamente as anomalias de código alvo por meio de refatorações compostas. Refatorações compostas concluídas podem não ser totalmente benéficas para a estrutura do código. Pois, estas podem induzir efeitos colaterais, como a introdução de anomalias de código ou a propagação de anomalias existentes. Há uma compreensão limitada sobre a completude das refatorações compostas e seus possíveis efeitos colaterais. Esta tese investiga como as refatorações compostas removem totalmente as anomalias de código sem induzir efeitos colaterais. Descobrimos que 64 por cento das refatorações compostas completas são formadas por tipos de refatoração não recomendados anteriormente. Dessa forma, derivamos um catálogo de recomendações para apoiar os desenvolvedores na aplicação de refatorações compostas. Na avaliação do catálogo, 85 por cento de 21 desenvolvedores relataram que usariam as recomendações do catálogo e que suas próprias soluções de refatoração teriam induzido efeitos colaterais. Também avaliamos qualitativamente três abordagens existentes para recomendar automaticamente refatorações compostas. Nesse estudo, a maioria (80 por cento) dos 10 desenvolvedores relatou que as abordagens existentes frequentemente induzem efeitos colaterais. No geral, as descobertas e o catálogo proposto podem ajudar os desenvolvedores a realizar refatorações compostas completas. / [en] Code refactoring is a code transformation that aims to enhance the internal code structure. A single refactoring is rarely sufficient to achieve the full removal of a poor code structure, such as a code smell. Developers then apply composite refactorings to fully remove a code smell. A composite refactoring (or, simply, composite) consists of two or more interrelated single refactorings. A composite is considered complete when it fully eliminates the target smell. However, studies report that developers often fail in completely removing target code smells through composites. Even when composite refactorings are complete they may still not be entirely beneficial to the code structure. They may induce side effects, such as the introduction of new smells or the propagation of existing ones. There is a limited understanding of the completeness of composite refactorings and their possible effects on structural quality. This thesis investigates whether and how composite refactorings fully remove smells without inducing side effects. We found that 64 per cent of complete composites in several software projects are formed of refactoring types not previously recommended in the literature. Based on this study, we derived a catalog of recommendations for supporting developers in applying composite refactorings. Out of twenty one developers evaluating our catalog, 85 per cent reported that they would use the catalog recommendations and that their own refactoring solutions would have induced side effects. We also qualitatively evaluated three existing approaches to automatically recommend composite refactorings. In our study with ten developers, most (80 per cent) developers reported that existing approaches frequently induce side effects. Overall, the findings and the proposed catalog can help developers to perform complete composite refactorings with better awareness of possible side effects.
122

Analysis Techniques for Software Maintenance

Pérez Rubio, Sergio 05 May 2023 (has links)
[ES] Vivimos en una sociedad donde la digitalización está presente en nuestro día a día. Nos despertamos con la alarma de nuestro teléfono móvil, apuntamos nuestras reuniones en nuestro calendario digital, guardamos nuestros archivos en el almacenamiento en la nube, y entramos a las redes sociales prácticamente a diario. Cada una de estas acciones se ejecuta sobre un sistema software que asegura su correcto funcionamiento. Esta digitalización masiva ha hecho que el desarrollo de software se dispare en los últimos años. Durante el ciclo de vida de un sistema software, la etapa de mantenimiento supone un gasto de billones de dólares anuales. La razón detrás de este gasto es la aparición de bugs o errores que no fueron detectados durante la fase de producción del software, y que se traducen en un mal funcionamiento del sistema. Por este motivo, las técnicas de detección y localización de errores como el testeo, la verificación o la depuración son un factor clave para asegurar la calidad del software. Aunque son muchas las técnicas que se utilizan para la depuración, testeo y verificación de sistemas software, esta tesis se centra solo en algunas de ellas. En concreto, esta tesis presenta mejoras en la técnica de fragmentación de programas (depuración), una nueva metodología para hacer testeo de regresión (testeo) y una nueva implementación del modelo de verificación de diseño-por-contrato para el lenguaje de programación Erlang (verificación). Las mejoras propuestas para la técnica de fragmentación de programas incluyen diversas propuestas aplicables a diferentes escenarios: (i) mejoras en la representación y fragmentación de programas orientados a objetos, (ii) mejoras en la representación y fragmentación de estructuras de datos complejas (objetos, vectores, listas, tuplas, registros, etc.), (iii) un nuevo modelo de grafo basado en una representación más detallada de programas, aumentando la expresividad del grafo y generando fragmentos de código más precisos como resultado, y (iv) una nueva técnica para calcular fragmentos mínimos de un programa dado un conjunto específico de posibles valores de entrada. Por otro lado, la nueva metodología para hacer testeo de regresión se denomina testeo de punto de interés, e introduce la posibilidad de comparar automáticamente el comportamiento de un punto cualquiera del código dadas dos versiones del mismo sistema software. Por último, la tesis contiene la nueva implementación del modelo de verificación de programas diseño-por-contrato para el lenguaje de programación Erlang, donde se explican en detalle los nuevos tipos de contratos diseñados para las partes secuencial y concurrente de Erlang. Todos los análisis presentados en esta tesis han sido formalmente definidos y su corrección ha sido probada, asegurando de esta manera que los resultados tendrán el grado de fiabilidad necesario para ser aplicados a sistemas reales. Otra contribución de esta tesis es la implementación de dos herramientas de fragmentación de programas para dos lenguajes de programación diferentes (Java y Erlang), una herramienta para realizar testeo de punto de interés para el lenguaje de programación Erlang y un sistema para ejecutar verificación de diseño-por-contrato en Erlang. Es de destacar que todas las herramientas implementadas a lo largo del desarrollo de esta tesis son herramientas de código abierto y públicamente accesibles, de manera que pueden ser usadas o extendidas por cualquier investigador interesado en este area. / [CA] Vivim en una societat on la digitalització està present al nostre dia a dia. Ens alcem amb l'alarma del nostre telèfon mòbil, apuntem les nostres reunions al nostre calendari digital, guardem els nostres arxius al emmagatzematge al núvol, i entrem a las xarxes socials pràcticament a diari. Cadascuna d'aquestes accions s'executa sobre un sistema programari que assegura el seu correcte funcionament. Aquesta digitalizació massiva ha fet que el desenvolupament de programari es dispare en els últims anys. Durant el cicle de vida de un sistema programari, l'etapa de manteniment suposa una despesa de bilions de dòlars anuals. La raó darrere d'aquesta despesa és l'aparició de bugs o errors que no van ser detectats durant la fase de producció del programari, i que es traduïxen en un mal funcionament del sistema Per este motiu, les tècniques de detecció i localització d'errors com el testeig, la verificació o la depuració són un factor clau per a assegurar la qualitat del programari. Encara que són moltes les tècniques utilitzades per a la depuració, testeig i verificació de sistemes programari, esta tesi es centra només en algunes d'elles. En concret, esta tesi presenta millores en la tècnica de fragmentació de programes (depuració), una nova metodologia per a fer testeig de regressió (testeig) i una nova implementació del model de verificació de disseny-per-contracte per al llenguatge de programació Erlang (verificació). Les millores proposades per a la tècnica de fragmentació de programes inclouen diverses propostes aplicables a diferents escenaris: (i) millores en la representació i fragmentació de programes orientats a objectes, (ii) millores en la representació i fragmentació d'estructures de dades complexes (objectes, vectors, llistes, tuples, registres, etc.), (iii) un nou model de graf basat en una representació més detallada de programes, augmentant l'expressivitat del graf i generant fragments de codi més precisos com a resultat, i (iv) una nova tècnica per a calcular fragments mínims d'un programa donat un conjunt específic de possibles valors d'entrada. D'altra banda, la nova metodologia per a fer testeig de regressió es denomina testeig de punt d'interés, i introduïx la possibilitat de comparar automàticament el comportament d'un punt qualsevol del codi donades dos versions del mateix sistema programari. Finalment, la tesi conté la nova implementació del model de verificació de programes disseny-per-contracte per al llenguatge de programació Erlang, on s'expliquen en detall els nous tipus de contractes dissenyats per a les parts seqüencial i concurrent d'Erlang. Totes les anàlisis presentades en aquesta tesi han sigut formalment definides i la seua correcció ha sigut provada, assegurant d'aquesta manera que els resultats tindran el grau de fiabilitat necessari per a ser aplicats a sistemes reals. Una altra contribució d'aquesta tesi és la implementació de dos ferramentes de fragmentació de programes per a dos llenguatges de programació diferents (Java i Erlang), una ferramenta per a realitzar testeig the punt d'interés per al llenguatge de programació Erlang i un sistema per a executar verificació de disseny-per-contracte a Erlang. Cal destacar que totes les ferramentes implementades al llarg del desenvolupament d'aquesta tesi són ferramentes de codi obert i públicament accessibles, de manera que poden ser usades o esteses per qualsevol investigador interessat en el tema. / [EN] We live in a society where digitalisation is present in our everyday life. We wake up with the alarm of our mobile phone, book our meetings in our digital calendar, save all our media in the cloud storage, and spend time in social networks almost daily. Every one of these tasks is run over a software system that ensures its correct functionality. This massive digitalisation has made software development to shoot up in the last years. During the lifetime of software systems, the maintenance process entails a waste of billions of dollars every year. The cause of this waste is the occurrence of bugs or errors undetected during the software production, which result in a malfunction of the system. For this reason, error detection and localisation techniques, such as testing, verification, or debugging, are a key factor to ensure software quality. Although many different techniques are used for the debugging, testing, and verification of software systems, this thesis focus on only some of them. In particular, this thesis presents improvements in the program slicing technique (debugging field), a new approach for regression testing (testing field), and a new implementation of the design-by-contract verification model for the Erlang programming language (verification field). The improvements proposed for the program slicing technique include several enhancements applicable to different scenarios: (i) improvements in the representation and slicing of object-oriented programs, (ii) enhancements in the representation and slicing of (possibly recursive) complex data structures (objects, arrays, lists, tuples, records, etc.), (iii) a new graph model based on a fine-grained representation of programs that augments the expressivity of the graph and provides more accurate slicing results, and (iv) a new technique to compute minimal slices for a program given a set of specific program inputs. On the other side, the new approach for regression testing is called point of interest testing, and it introduces the possibility of automatically comparing the behaviour of any arbitrary point in the code given two versions of the same software system. Finally, the thesis presents a new implementation of the design-by-contract verification model for the Erlang programming language, where new types of contracts are explained in detail for both the sequential and concurrent parts of Erlang. All the analyses presented here have been formally defined and their correctness have been proved, ensuring that the results will have the reliability degree needed for real-life systems. Another contribution of this thesis is the implementation of two program slicers for two different programming languages (Java and Erlang), a tool to perform point of interest testing for the Erlang programming language, and a system to run design-by-contract verification in Erlang. It is worth mentioning that all the tools implemented in this thesis are open source and publicly available, so they can be used or extended by any interested researcher. / Pérez Rubio, S. (2023). Analysis Techniques for Software Maintenance [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/193146
123

Supporting Source Code Comprehension During Software Evolution and Maintenance

Alhindawi, Nouh Talal 30 July 2013 (has links)
No description available.
124

Part-of-Speech Tagging of Source Code Identifiers using Programming Language Context Versus Natural Language Context

AlSuhaibani, Reem Saleh 03 December 2015 (has links)
No description available.
125

[pt] ENTENDENDO CARACTERÍSTICAS E EFEITOS ESTRUTURAIS DE REFATORAÇÃO EM LOTES NA PRÁTICA / [en] UNDERSTANDING CHARACTERISTICS AND STRUCTURAL EFFECTS OF BATCH REFACTORINGS IN PRACTICE

ANA CARLA GOMES BIBIANO 04 November 2019 (has links)
[pt] Refatorar código-fonte consiste em aplicar transformações sobre a estrutura de código-fonte de projetos de software. Refatoração é bastante usada para remover estruturas pobres que dificultam a manutenção de sistemas de software. Poucas transformações isoladas são capazes de remover por completo estruturas pobres, mesmo as mais simples. Por exemplo, encurtar um método longo usualmente requer a extração de vários métodos. Até 60 por cento das transformações são inter-relacionadas e aplicadas em lotes, durante a dita refatoração em lote, ao invés de aplicadas isoladamente. Embora lotes serão frequentes na prática, o conhecimento sobre as características que constituem lotes está fragmentado na literatura. Qual o tamanho usual de lotes? As transformações internas a lotes costumam variar? Não há uma sumarização de conhecimento que responda tais questões. Ademais, são poucas as evidências sobre o efeito de lotes sobre a manutenção de sistemas. Lotes tendem a introduzir ou remover estruturas pobres, especialmente aquelas indicadas por anomalias de código-fonte? A resposta a perguntas como essa é insuficiente para apoiar a aplicação de lotes. Esta dissertação de mestrado apresenta dois estudos experimentais complementares visando resolver as limitações supracitadas. A dissertação começa com uma revisão da literatura sobre refatoração em lote baseada em 29 estudos. Nós identificamos sete características de lotes tais como o escopo de código-fonte afetado pela aplicação de um lote, mais sete tipos de efeito de lotes sobre a manutenção de sistemas, tais como a remoção de anomalias. As características e tipos de efeito identificadas foram sumarizadas por um mapa conceitual. A dissertação encerra-se com uma análise quantitativa de 57 projetos de sistemas abertos e fechados. Ao computar 4.607 lotes com uma heurística, nós descobrimos que a maioria dos lotes leva um único commit para ser aplicada (93 por cento) mas afeta mais do que um só método (90 por cento). Surpreendentemente, a maioria dos lotes introduz (51 por cento) ou não remove (38 por cento) anomalias. Revelamos também lotes até então desconhecidos mas capazes de remover por completo certas anomalias. Esta dissertação sugere trabalhos futuros com base em conflitos identificados na literatura quanto a características e tipos de efeito de lotes. / [en] Code refactoring means applying transformations on the code structure of a software project. Refactoring usually intends to remove poor code structures that harm the software maintenance. Each single transformation rarely suffices to fully remove poor code structures, even the simplest ones. For instance, shortening a long method often requires many method extractions. Up to 60 percent of the refactorings in software projects are constituted of a set of interrelated transformations, the so-called batches, rather than single transformations applied in isolation. Although batches are frequent in practice, the knowledge of batch characteristics is fragmented across studies. What is the usual size of batches? How do transformations vary within a batch? There is no summary that helps to address these questions. More critically, there is little empirical evidence of the batch effect on maintenance. Are batches more likely to introduce or remove poor code structures, especially those spotted by code smells? The current answer to questions like this is insufficient to support the batch application in practice. This Master s dissertation presents two complementary empirical studies that address both aforementioned literature gaps. The dissertation starts with a literature review of batch refactoring with 29 studies. We identified seven batch characteristics such as the scope in which batches are applied to code structures, plus seven types of batch effect on software maintenance, including code smell removal. All batch characteristics and types of effect were summarized in a conceptual map. The dissertation ends with the quantitative analysis of 57 open and closed software projects. From 4,607 heuristic-computed batches, we found that most batches occur entirely within one commit (93 percent) but affect more than just one method (90 percent). Surprisingly, batches mostly end up introducing (51 percent) or not removing (38 percent) code smells. Our results enabled us to reveal certain forms of batches, not documented by previous studies, that are useful to fully remove certain types of code smells.
126

Visualisering av mjukvarusystem och systemberoenden

Buchberger, Daniel, Sozinov, Konstantin January 2016 (has links)
Visualization of software systems is mainly used for understanding and modeling software architecture. Examples of what can be visualized are source code, systems during execution together with manipulated data or system architecture on either a detailed or a general level. The connection between all of these types of visualization is that they help computer engineers maintain or develop software systems. The largest challenge with visualizing software systems is when the systems in question are exceedingly large and containing an abundance of relations. A problem of this kind arise at the company DGC, where this study is performed. The company manages at present more than a couple hundred systems with over ten thousand relations between each other. The problem with the company’s existing visualization is that they are very difficult to follow and understand, since they mainly present raw data in the form of tables that do not show all relations optimally, and which can also easily distract the company stakeholders from their intended task. The aim of this study is to find a visualization method where a system and its underlying parts are shown together with their relations to other systems. The created visualization will be able to be used by several different stakeholders and will as clearly as possible present relevant information to them. Furthermore, it should also aim to contribute to a better understanding of the complex software systems and their relations. After the implementation of the chosen visualization method, a time-based study is performed to determine whether it poses an improvement of perception among the stakeholders or not. An interview is also conducted where these stakeholders provide their opinion on the created visualization and if they think it presents the necessary information in a pedagogical way. Analysis of the result of the timekeeping and the interview answers indicate that the aim of the study is met. The majority of the resulting times show that the created visualization is more efficient than the original ones, and the interview answers furthermore show how different stakeholders all together agree that the visualization can be used in both future development and maintenance. / Visualisering av mjukvarusystem används i stor utsträckning för att förstå och modellera mjukvaruarkitektur. Ett par exempel på vad som kan visualiseras är: källkod, system under exekvering tillsammans med manipulerat data eller systemarkitektur på detaljerad eller övergripande nivå. Sambandet mellan alla dessa typer av visualiseringar är att de hjälper dataingenjörer att underhålla och utveckla mjukvarusystem. Den största utmaningen i att visualisera mjukvarusystem är när systemen i fråga är väldigt stora och har många beroenden. Ett sådant problem uppstår på företaget DGC där denna undersökning utförs. Företaget hanterar i dagsläge mer än hundratals mjukvarusystem med över tiotusentals beroenden sinsemellan. Problemet med företagets befintliga visualiseringar är att dessa är väldigt svåra att följa eftersom de oftast presenterar rådata i form av tabeller som inte visar alla relationer på ett optimalt sätt och distraherar lätt intressenter på företaget från sin avsedda uppgift. Undersökningen syftar på att finna en visualiseringsmetod där ett system och dess underliggande systemdelar visas tillsammans med deras beroenden till andra. Den framtagna visualiseringen skall kunna användas av flera olika intressenter, och i högsta möjliga mån presentera relevant information för dem på ett tydligt sätt. Vidare ska den skapade visualiseringen också bidra till bättre förståelse för komplexa mjukvarusystem och deras beroenden. Efter framtagandet av visualiseringen utförs en tidsundersökning, för att avgöra huruvida den skapade visualiseringen stimulerar förståelse hos de olika intressenterna på företaget. Det utförs även en intervju där intressenterna på företaget lämnar sina synpunkter angående den skapade visualiseringen och om den kan presentera nödvändig information på ett pedagogiskt sätt. Analys av resultaten av tidtagningarna och svaren på intervjuerna visar tydligt att undersökningens syften nåtts. Majoriteten av tidtagningarna visade att den skapade visualiseringen är effektivare än de ursprungliga och intervjusvaren uppvisar hur olika intressenter ser att den kan användas vid vidareutvecklingen eller underhåll, vilket alla intressenter höll med om.
127

A unified framework for the comprehension of software's time dimension

Benomar, Omar 02 1900 (has links)
Les logiciels sont de plus en plus complexes et leur développement est souvent fait par des équipes dispersées et changeantes. Par ailleurs, de nos jours, la majorité des logiciels sont recyclés au lieu d’être développés à partir de zéro. La tâche de compréhension, inhérente aux tâches de maintenance, consiste à analyser plusieurs dimensions du logiciel en parallèle. La dimension temps intervient à deux niveaux dans le logiciel : il change durant son évolution et durant son exécution. Ces changements prennent un sens particulier quand ils sont analysés avec d’autres dimensions du logiciel. L’analyse de données multidimensionnelles est un problème difficile à résoudre. Cependant, certaines méthodes permettent de contourner cette difficulté. Ainsi, les approches semi-automatiques, comme la visualisation du logiciel, permettent à l’usager d’intervenir durant l’analyse pour explorer et guider la recherche d’informations. Dans une première étape de la thèse, nous appliquons des techniques de visualisation pour mieux comprendre la dynamique des logiciels pendant l’évolution et l’exécution. Les changements dans le temps sont représentés par des heat maps. Ainsi, nous utilisons la même représentation graphique pour visualiser les changements pendant l’évolution et ceux pendant l’exécution. Une autre catégorie d’approches, qui permettent de comprendre certains aspects dynamiques du logiciel, concerne l’utilisation d’heuristiques. Dans une seconde étape de la thèse, nous nous intéressons à l’identification des phases pendant l’évolution ou pendant l’exécution en utilisant la même approche. Dans ce contexte, la prémisse est qu’il existe une cohérence inhérente dans les évènements, qui permet d’isoler des sous-ensembles comme des phases. Cette hypothèse de cohérence est ensuite définie spécifiquement pour les évènements de changements de code (évolution) ou de changements d’état (exécution). L’objectif de la thèse est d’étudier l’unification de ces deux dimensions du temps que sont l’évolution et l’exécution. Ceci s’inscrit dans notre volonté de rapprocher les deux domaines de recherche qui s’intéressent à une même catégorie de problèmes, mais selon deux perspectives différentes. / Software systems are getting more and more complex and are developed by teams that are constantly changing and not necessarily working in the same location. Moreover, most software systems, nowadays, are recycled rather than being developed from scratch. A comprehension task is crucial when performing maintenance tasks; it consists of analyzing multiple software dimensions concurrently. Time is one of these dimensions, as software changes its state with time in two manners: during their execution and during their evolution. These changes make sense only when analyzed within the context of other software dimensions, such as structure or bug information. Multidimensional analysis is a difficult problem to solve. However, there are certain methods that bypass this difficulty, such as semi-automatic approaches. Software visualization is one of them, as it allows being part of the analysis by exploring and guiding information search. The first stage of the thesis consists of applying visualization techniques to better understand software dynamicity during execution and evolution. Changes over time are represented by heat maps. Hence, we utilize the same graphical representation to visualize both change types over time. Other approaches that permit the analysis of a program’s dynamic behavior over time involve the use of heuristics. In the thesis’ second stage, we are interested in the identification of the programs’ execution phases and evolution patterns using the same approach, i.e. search-based optimisation. In this context, the premise is the existence of internal cohesion between change events that allow the clustering in phases. This hypothesis of cohesion is defined specifically for change events in the code during software evolution and state changes during program execution. This thesis’ main objective is to study the unification of these two time dimensions, evolution and execution, in an attempt to bring together two research domains that work on the same set of problems, but from two different perspectives.
128

Context-based code quality assessment / Avaliação de qualidade de código baseada em contexto

Aniche, Mauricio Finavaro 15 July 2016 (has links)
Two tasks that software engineers constantly perform are writing code that is easy to evolve and maintain, and detecting poorly written pieces of code. For the former, software engineers commonly rely on well-known software architecture styles, such as Model-View-Controller (MVC). To the latter, they rely on code metrics and code smell detection approaches. However, up to now, these code metrics and code smell approaches do not take into account underlying architectureall classes are assessed as if they were the same. In practice, software developers know that classes differ in terms of responsibilities and implementation, and thus, we expect these classes to present different levels of coupling, cohesion, and complexity. As an example, in an MVC system, Controllers are responsible for the flow between the Model and the View, and Models are responsible for representing the systems business concepts. Thus, in this thesis, we evaluate the impact of architectural roles within a system architecture on code metrics and code smells. We performed an empirical analysis in 120 open source systems, and interviewed and surveyed more than 50 software developers. Our findings show that each architectural role has a different code metric values distribution, which is a likely consequence of their specific responsibilities. Thus, we propose SATT, an approach that provides specific thresholds for architectural roles that are significantly different from others in terms of code smells. We also show that classes that play a specific architectural role contain specific code smells, which developers perceive as problems, and can impact class\' change- and defect-proneness. Based on our findings, we suggest that developers understand the responsibilities of each architectural role in their system architecture, so that code metrics and code smells techniques can provide more accurate feedback. / Duas tarefas que desenvolvedores de software constantemente fazem são escrever código fácil de ser mantido e evoluído, e detectar pedaços de código problemáticos. Para a primeira tarefa, desenvolvedores comumente fazem uso de conhecidos padrões arquiteturais, como Model-View-Controller (MVC). Para a segunda tarefa, desenvolvedores fazem uso de métricas de código e estratégias de detecção de maus cheiros de código (code smells). No entanto, até o momento, métricas de código e estratégias de detecção de maus cheiros de código não levam em conta a arquitetura do software em análise. Isso significa que todas classes são avaliadas como se umas fossem iguais às outras. Na prática, sabemos que classes são diferentes em suas responsibilidades e implementação, e portanto, esperamos que elas variem em termos de acoplamento, coesão e complexidade. Por exemplo, em um sistema MVC, Controladores são responsáveis pelo fluxo entre a camada de Modelo e a camada de Visão, e Modelos representam a visão de negócios do sistema. Nesta tese, nós avaliamos o impacto dos papéis arquiteturais em técnicas de medição de métricas de código e de detecção de maus cheiros de código. Nós realizamos um estudo empírico em 120 sistemas de código aberto, e entrevistamos e realizamos questionários com mais de 50 desenvolvedores. Nossos resultados mostram que cada papel arquitetural possui distribuições diferentes de valores de métrica de código, consequência das diferentes responsabilidades de cada papel. Como consequência, propomos SATT, uma abordagem que provê thresholds específicos para papéis arquiteturais que são significantemente diferentes de outros em termos de métricas de código. Mostramos também que classes que cumprem um papel arquitetural específico também contêm maus cheiros de código específicos. Esses maus cheiros são percebidos por desenvolvedores como problemas reais e podem fazer com que essas classes sejam mais modificadas e apresentem mais defeitos do que classes limpas. Sugerimos então que desenvolvedores entendam a arquitetura dos seus sistemas, bem como as responsabilidades de cada papel arquitetural que as classes desempenham, para que tanto métricas de código quanto estratégias de detecção de maus cheiros de código possam prover um melhor retorno.
129

Identificação e visualização de dependências em sistemas de software orientados a objetos / Identification and Visualization of Dependencies in Object-Oriented Software Systems

Oliva, Gustavo Ansaldi 22 September 2011 (has links)
Degradação do design é um problema central investigado na área de evolução de software. A densa rede de interdependências que emerge entre classes e módulos ao longo do tempo resulta em código difícil de mudar, não reutilizável e que não comunica por si só sua intenção. Dentre outros motivos, designs degradam porque requisitos mudam de maneiras não antecipadas pelo design inicial, ou seja, as modificações no código introduzem dependências novas e não planejadas entre classes e módulos do sistema. A gerência de dependências visa reduzir a degradação do design por meio de uma série de mecanismos que auxiliam na administração da complexidade estrutural inerente de sistemas orientados a objetos. Neste trabalho, investigamos as técnicas de identificação de dependências estruturais e lógicas. Em particular, por meio de um estudo de larga escala, comparamos os conjuntos desses dois tipos de dependências. Em seguida, conduzimos um estudo de caso a fim de identificar as origens de dependências lógicas. Por fim, fazemos um levantamento das técnicas de visualização de dependências e mostramos a ferramenta XFlow. / Design degradation is a central problem investigated in the area of software evolution. The dense web of interdependencies that emerges among classes and modules over time results in code that is hard to change, not reusable and that does not communicate its intention. Among other reasons, designs degrade because requirements changes in ways that were not anticipated by the initial design, i.e. the changes in code introduce new and unplanned dependencies among classes and modules of the system. Dependency management aims to reduce design degradation by means of a series of mechanisms that helps in the management of the inherent structural complexity of object oriented systems. In this work, we investigate structural and logical dependencies identification techniques. In particular, by means of a large scale study, we compare the sets of these two kinds of dependencies. Afterwards, we conduct a case study in order to uncover the origins of logical dependencies. Finally, we survey dependency visualization techniques and present the XFlow tool.
130

Change-effects analysis for effective testing and validation of evolving software

Santelices, Raul A. 17 May 2012 (has links)
The constant modification of software during its life cycle poses many challenges for developers and testers because changes might not behave as expected or may introduce erroneous side effects. For those reasons, it is of critical importance to analyze, test, and validate software every time it changes. The most common method for validating modified software is regression testing, which identifies differences in the behavior of software caused by changes and determines the correctness of those differences. Most research to this date has focused on the efficiency of regression testing by selecting and prioritizing existing test cases affected by changes. However, little attention has been given to finding whether the test suite adequately tests the effects of changes (i.e., behavior differences in the modified software) and which of those effects are missed during testing. In practice, it is necessary to augment the test suite to exercise the untested effects. The thesis of this research is that the effects of changes on software behavior can be computed with enough precision to help testers analyze the consequences of changes and augment test suites effectively. To demonstrate this thesis, this dissertation uses novel insights to develop a fundamental understanding of how changes affect the behavior of software. Based on these foundations, the dissertation defines and studies new techniques that detect these effects in cost-effective ways. These techniques support test-suite augmentation by (1) identifying the effects of individual changes that should be tested, (2) identifying the combined effects of multiple changes that occur during testing, and (3) optimizing the computation of these effects.

Page generated in 0.1359 seconds