La modernisation logicielle est un domaine en plein essor du fait du vieillissement des technologies datant des années 1960, tel que COBOL ou FORTRAN. Aujourd'hui, bon nombre de ces applications sont toujours en service du fait de leur criticité. Ce sont elles qui continuent d'assurer le bon fonctionnement du système d'information de l'entreprise. Face à l'arrivée en force de technologies récentes de type Cloud ou Mobile, il devient urgent de les moderniser. Toutefois, près de deux projets de modernisation sur trois échouent à cause du manque ou la perte de documentation de l'application, la complexité du logiciel (qui lui n'a cessé de grandir avec le temps), la perte des connaissances techniques (les jeunes développeurs n'ont jamais étudié ces anciennes technologies), tout cela couplé à un manque d'outils et de méthodes industrielles. Face à ce constat implacable, cette thèse suggère une solution pour l'industrialisation du processus de modernisation à travers l'utilisation de l'Ingénierie Dirigée par les Modèles (IDM), notamment en prenant pour base les métamodèles issus de l'initiative Architecture Driven Modernisation de l'OMG. Nous montrons qu'il est ainsi possible d'automatiser une majorité du processus du fait de l'utilisation de métamodèles pivots technologiquement neutres, ainsi que l'utilisation d'une série d'outils permettant d'obtenir une automatisation des tâches de migration de l'application. Nous proposons également de rendre cette solution adaptable à n'importe quel langage légataire au travers d'une chaîne de modernisation paramétrable via l'utilisation d'un moteur de transformation dédié offrant des mécanismes d'extensibilité et de réutilisabilité. Pour terminer, nous illustrons l'ensemble des idées développées dans cette thèse sur l'exemple concret d'une application codée en COBOL. / Software modernization is a fast-growing sector due to the aging technologies from the 1960s, such as COBOL or FORTRAN. Nowadays, many of these applications are still being used because of their criticality. These applications allow companies to maintain their information system. But to face the arrival of newer technologies and platforms, like Cloud or Mobile, companies must modernize their systems to deliver the same services and also develop new ones in order to compete. However, nearly two out of three modernization projects fail because of the lack or the loss of applications documentation, the complexity of the software (that has continued to grow over time), the loss of technical knowledge (the junior developers have never studied these ancient technologies), plus a lack of tools and industrial methods. Faced with this relentless observation, this thesis suggests a solution to the industrialization process of modernization through the use of Model Driven (IDM) Engineering, especially by using metamodels base from Driven Architecture Modernization of the OMG. We therefore demonstrate that it is possible to automate the process thanks to technologically neutral metamodels and by using a set of tools to get the automation of application migration tasks. We also propose to have this solution adapted to any languages through a configurable chain of modernization by using a dedicated processing engine that provides scalability and reusability mechanisms. Finally, we illustrate the ideas developed in this thesis in the concrete case of a coded COBOL application.

La fiabilité du logiciel, c’est-à-dire sa capacité à produire les fonctionnalités attendues, est essentielle au succès des projets de développement logiciel. Afin de garantir cette fiabilité, les développeurs ont pour objectif de réduire le nombre de bogues présents dans le code source du logiciel.Une des techniques ayant pour but d’aider les développeurs dans cette tâche est l’utilisation de métriques logicielles, et notamment de métriques liées au procédé de développement.L’objectif général de cette thèse est de contribuer à la validation de métriques de procédé en étudiant leur relation avec la fiabilité. Ces métriques, une fois validées, pourront être utilisées dans des modèles de prédiction de bogues ayant pour but de mieux orienter les efforts de maintenance des développeurs ou pourront permettre de mettre en place des lignes de conduite relatives au procédé de développement. Devant l’étendue de ce domaine, nous avons centré nos contributions sur un aspect du procédé de développement qui est l’organisation des développeurs et avons observé cette organisation dans des projets open-source.En parallèle de la validation de ces métriques, nous avons contribué à l’amélioration de la méthodologie permettant l’extraction et l’analyse de métriques, grâce aux informations contenues dans les dépôts logiciels. / Reliability of a software, i.e. its capacity to produce the expected behaviour, is essential to the success of software projects. To ensure such reliability, developers need to reduce the amount of bugs in the source code of the software. One of the techniques available to help developers in this task is the use of software metrics, and especially metrics related to the development process.The general objective of this thesis is to contribute to the validation of process metrics, by studying their relationship with software reliability. These metrics, once validated, can be used in bug predictionmodels with the goal to guide maintenance efforts or can be used to create development guidelines. Given the extent of this domain, we chose to focus on one particular aspect of the development process, which is developers organisation, and we studied this organisation in open-source software projects.In parallel to the validation of process metrics, we contributed to the improvement of the methodology used to extract and analyse metrics, thanks to information available in software repositories.

This thesis undertakes an empirical study of software evolution by analyzing open source software (OSS) systems. The main purpose is to aid in understanding OSS evolution. The work centers on collecting large quantities of structural data cost-effectively and analyzing such data to understand software evolution <em>dynamics</em> (the mechanisms and causes of change or growth). <br /><br /> We propose a multipurpose systematic approach to extracting program facts (<em>e. g. </em>, function calls). This approach is supported by a suite of C and C++ program extractors, which cover different steps in the program build process and handle both source and binary code. We present several heuristics to link facts extracted from individual files into a combined system model of reasonable accuracy. We extract historical sequences of system models to aid software evolution analysis. <br /><br /> We propose that software evolution can be viewed as <em>Punctuated Equilibrium</em> (<em>i. e. </em>, long periods of small changes interrupted occasionally by large avalanche changes). We develop two approaches to study such dynamical behavior. One approach uses the evolution spectrograph to visualize file level changes to the implemented system structure. The other approach relies on automated software clustering techniques to recover system design changes. We discuss lessons learned from using these approaches. <br /><br /> We present a new perspective on software evolution dynamics. From this perspective, an evolving software system responds to external events (<em>e. g. </em>, new functional requirements) according to <em>Self-Organized Criticality</em> (SOC). The SOC dynamics is characterized by the following: (1) the probability distribution of change sizes is a power law; and (2) the time series of change exhibits long range correlations with power law behavior. We present empirical evidence that SOC occurs in open source software systems.

An object-oriented software system is composed of a collection of communicating objects that co-operate with one another to achieve some desired goals. The object is the basic unit of abstraction in an OO program; objects may model real-world entities or internal abstractions of the system. Similar objects forms classes, which encapsulate the data and operations performed on the data. Therefore, extracting, analyzing, and modelling classes/objects and their relationships is of key importance in understanding and maintaining object-oriented software systems. However, when dealing with large and complex object-oriented systems, maintainers can easily be overwhelmed by the vast number of classes/objects and the high degree of interdependencies among them. In this thesis, we propose a new model, which we call the Hybrid Model, to represent object-oriented systems at a coarse-grained level of abstraction. To promote the comprehensibility of objects as independent units, we group the complete static description of software objects into aggregate components. Each aggregate component logically represents a set of objects, and the components interact with one other through explicitly defined ports. We present and discuss several applications of the Hybrid Model in reverse engineering and software evolution. The Hybrid Model can be used to support a divide-and-conquer comprehension strategy for program comprehension. At a low level of abstraction, maintainers can focus on one aggregate-component at a time, while at a higher level, each aggregate component can be understood as a whole and be mapped to coarse-grained design abstractions, such as subsystems. Based on the new model, we further propose a set of dependency analysis methods. The analysis results reveal the external properties of aggregate components, and lead to better understand the nature of their interdependencies. In addition, we apply the new model in software evolution analysis. We identify a collection of change patterns in terms of changes in aggregate components and their interrelationships. These patterns help to interpret how an evolving system changes at the architectural level, and provides valuable information to understand why the system is designed as the way it is.

This thesis undertakes an empirical study of software evolution by analyzing open source software (OSS) systems. The main purpose is to aid in understanding OSS evolution. The work centers on collecting large quantities of structural data cost-effectively and analyzing such data to understand software evolution <em>dynamics</em> (the mechanisms and causes of change or growth). <br /><br /> We propose a multipurpose systematic approach to extracting program facts (<em>e. g. </em>, function calls). This approach is supported by a suite of C and C++ program extractors, which cover different steps in the program build process and handle both source and binary code. We present several heuristics to link facts extracted from individual files into a combined system model of reasonable accuracy. We extract historical sequences of system models to aid software evolution analysis. <br /><br /> We propose that software evolution can be viewed as <em>Punctuated Equilibrium</em> (<em>i. e. </em>, long periods of small changes interrupted occasionally by large avalanche changes). We develop two approaches to study such dynamical behavior. One approach uses the evolution spectrograph to visualize file level changes to the implemented system structure. The other approach relies on automated software clustering techniques to recover system design changes. We discuss lessons learned from using these approaches. <br /><br /> We present a new perspective on software evolution dynamics. From this perspective, an evolving software system responds to external events (<em>e. g. </em>, new functional requirements) according to <em>Self-Organized Criticality</em> (SOC). The SOC dynamics is characterized by the following: (1) the probability distribution of change sizes is a power law; and (2) the time series of change exhibits long range correlations with power law behavior. We present empirical evidence that SOC occurs in open source software systems.

An object-oriented software system is composed of a collection of communicating objects that co-operate with one another to achieve some desired goals. The object is the basic unit of abstraction in an OO program; objects may model real-world entities or internal abstractions of the system. Similar objects forms classes, which encapsulate the data and operations performed on the data. Therefore, extracting, analyzing, and modelling classes/objects and their relationships is of key importance in understanding and maintaining object-oriented software systems. However, when dealing with large and complex object-oriented systems, maintainers can easily be overwhelmed by the vast number of classes/objects and the high degree of interdependencies among them. In this thesis, we propose a new model, which we call the Hybrid Model, to represent object-oriented systems at a coarse-grained level of abstraction. To promote the comprehensibility of objects as independent units, we group the complete static description of software objects into aggregate components. Each aggregate component logically represents a set of objects, and the components interact with one other through explicitly defined ports. We present and discuss several applications of the Hybrid Model in reverse engineering and software evolution. The Hybrid Model can be used to support a divide-and-conquer comprehension strategy for program comprehension. At a low level of abstraction, maintainers can focus on one aggregate-component at a time, while at a higher level, each aggregate component can be understood as a whole and be mapped to coarse-grained design abstractions, such as subsystems. Based on the new model, we further propose a set of dependency analysis methods. The analysis results reveal the external properties of aggregate components, and lead to better understand the nature of their interdependencies. In addition, we apply the new model in software evolution analysis. We identify a collection of change patterns in terms of changes in aggregate components and their interrelationships. These patterns help to interpret how an evolving system changes at the architectural level, and provides valuable information to understand why the system is designed as the way it is.

Software evolution refers to continuous change that a software system endures from inception to retirement. Each change must be efficiently and tractably propagated across models representing the system at different levels of abstraction. Model synchronization activities needed to support the systematic specification and analysis of evolution activities are still not adequately identified and formally defined. In our research, we first introduce a formal notation for the representation of domain models and model instances to form the theoretical basis for the proposed model synchronization framework. Besides conforming to a generic MOF metamodel, we consider that each software model also relates to an application domain context (e.g., operating systems, web services). Therefore, we are addressing the problems of model synchronization by focusing on domain-specific contexts. Secondly, we identify and formally define model dependencies that are needed to trace and propagate changes across system models at different levels of abstraction, such as from design to source code. The approach for extraction of these dependencies is based on Formal Concept Analysis (FCA) algorithms. We further model identified dependencies using Unified Modeling Language (UML) profiles and constraints, and utilize the extracted dependency relations in the context of coarse-grained model synchronization. Thirdly, we introduce modeling semantics that allow for more complex profile-based dependencies using Triple Graph Grammar (TGG) rules with corresponding Object Constraint Language (OCL) constraints. The TGG semantics provide for fine-grained model synchronization, and enable compliance with the Query/View/Transformation (QVT) standards. The introduced framework is assessed on a large, industrial case study of the IBM Commerce system. The dependency extraction framework is applied to repositories of business process models and related source code. The extracted dependencies were evaluated by IBM developers, and the corresponding precision and recall values calculated with results that match the scope and goals of the research. The grammar-based model synchronization and dependency modelling using profiles has also been applied to the IBM Commerce system, and evaluated by the developers and architects involved in development of the system. The results of this experiment have been found to be valuable by stakeholders, and a patent codifying the results has been filed by the IBM organization and has been granted. Finally, the results of this experiment have been formalized as TGG rules, and used in the context of fine-grained model synchronization.

There is a strong synergy between the concepts of evolution and adaptation in software engineering: software adaptation refers to both the current software being adapted and to the evolution process that leads to the new adapted software. Evolution changes for the purpose of adaptation are usually made at development or compile time, and are meant to handle predictable situations in the form of software change requests. On the other hand, software may also change and adapt itself based on the changes in its environment. Such adaptive changes are usually dynamic, and are suitable for dealing with unpredictable or temporary changes in the software's operating environment. A promising solution for software adaptation is to develop self-adaptive software systems that can manage changes dynamically at runtime in a rapid and reliable way. One of the main advantages of self-adaptive software is its ability to manage the complexity that stems from highly dynamic and nondeterministic operating environments. If a self-adaptive software system has been engineered and used properly, it can greatly improve the cost-effectiveness of software change through its lifespan. However, in practice, many of the existing approaches towards self-adaptive software are rather expensive and may increase the overall system complexity, as well as subsequent future maintenance costs. This means that in many cases, self-adaptive software is not a good solution, because its development and maintenance costs are not paid off. The situation is even worse in the case of making current (legacy) systems adaptive. There are several factors that have an impact on the cost-effectiveness and usability of self-adaptive software; however the main objective of this thesis is to make a software system adaptive in a cost-effective way, while keeping the target adaptive software generic, usable, and evolvable, so as to support future changes. In order to effectively engineer and use self-adaptive software systems, in this thesis we propose a new conceptual model for identifying and specifying problem spaces in the context of self-adaptive software systems. Based on the foundations of this conceptual model, we propose a model-centric approach for engineering self-adaptive software by designing a generic adaptation framework and a supporting evolution process. This approach is particularly tailored to facilitate and simplify the process of evolving and adapting current (legacy) software towards runtime adaptivity. The conducted case studies reveal the applicability and effectiveness of this approach in bringing self-adaptive behaviour into non-adaptive applications that essentially demand adaptive behaviour to sustain.

Software has been studied at a variety of granularities. Code, classes, groups of classes, programs and finally large scale applications have been examined in detail. What lies beyond is the study of software collections that group together many individual applications. Collecting software and distributing it via a central repository has been popular for a while in the open source world, and only recently caught on commercially with Apple’s Mac app store and Microsoft’s Windows store. In many of these software collections, there is normally a complex process that must be followed in order to fully integrate new applications into the system. Moreover, in the case of open source software collections, applications frequently rely on each other for functionality and their interactions can be complex. We know that there are thousands of applications in these software collections that people depend on worldwide, but research in this area has been limited compared to other areas and granularities of software. In this thesis, we explore the evolution and architecture of a large open source software collections by using Debian as a case study. Debian is a software collection based off the Linux kernel with a large number of packages spread over multiple hardware platforms. Each package provides a particular service or application and is actively maintained by one or more developers. This thesis investigates how these packages evolve through time and their interactions with one another. The first half of the thesis describes the life cycle of a package from inception to end by carrying out a longitudinal study using the Ultimate Debian Database (UDD). The birth of packages is examined to see how Debian is growing. Conversely, package death is also analyzed to determine the lifespan of these packages. Moreover, four different package attributes are examined. They are package age, package bugs, package maintainers and package popularity. These four attributes combine to give us the overall biography of Debian packages. Debian’s architecture is explored in the second part of the thesis, where we analyze how packages interact with each other by examining the package dependencies in detail. The dependencies within Debian are extensive, which makes for an interesting architecture, but they are complex to analyze. This thesis provides a close look at the layered pattern. This pattern categorizes each package into one of five layers based on how they are used. These layers may also be visualized to give a concise view of how an application is structured. Using these views, we define five architectural subpatterns and anti-subpatterns which can aid developers in creating and maintaining packages.

The last few years have seen a resurgence of interest in the use of symbolic execution--program analysis technique developed more than three decades ago to analyze program execution paths. However, symbolic execution remains an expensive technique and scaling it remains a key technical challenge. There are two key factors that contribute to its cost: (1) the number of paths that need to be explored and (2) the cost of constraint solving, which is typically required for each path explored. Our insight is that the cost of symbolic execution can be reduced by an incremental approach, which uses static analysis and dynamic analysis to focus on relevant parts of code and reuse previous analysis results, thereby addressing both the key cost factors of symbolic execution. This dissertation presents Memoized Incremental Symbolic Execution, a novel approach that embodies our insight. Using symbolic execution in practice often requires several successive runs of the technique on largely similar underlying problems where successive problems differ due to some change, which may be to code, e.g., to fix a bug, to analysis parameters, e.g., to increase the path exploration depth, or to correctness properties, e.g., to check against stronger specifications that are written as assertions in code. Memoized Incremental Symbolic Execution, a three-fold approach, leverages the similarities in the successive problems to reduce the total cost of applying the technique. Our prototype tool-set is based on the Symbolic PathFinder. Experimental results show that Memoized Incremental Symbolic Execution enhances the efficacy of symbolic execution. / text