Large engineering processes need to be monitored in detail regarding when what was done in order to prove compliance with rules and regulations. A typical problem of these processes is the lack of con- trol that a central process engine provides, such that it is difficult to track the actual course of work even if data is stored in version control systems (VCS). In this paper, we address this problem by defining a mining technique that helps to generate models that visualize the work history as GANTT charts. To this end, we formally define the notion of a project-oriented business process and a corresponding mining algorithm. Our evaluation based on a prototypical implementation demonstrates the benefits in comparison to existing process mining approaches for this specific class of processes.

This thesis discusses the extraction of information from version control systems. Its goal is to describe the implementation of a software application that facilitates this type of extraction, focusing on the version control system Git. The theoretical part of the thesis identifies and analyses data stored in repositories. It also evaluates the tools available on the market using specific criteria. The practical part specifies development requirements, describes the resulting software application, and contains a how-to manual for extending the application and implementing one's own metrics that lead to gaining information. The application is developed in the CoffeeScript programming language and Node.js engine. It contains several example metrics. The output is a graphical user interface with interactive graphs served by a built-in HTTP server. Another output option is a machine-readable export to a file.

Dependency problems for developers are like sneezing for people with pollen allergies during the spring, an everyday problem. This is especially true when working in multi-repositories. The dependency problems that occur do so as a byproduct of enabling developers to work on different components of a project in smaller teams, where everything is version controlled.Nearly all developers use version control systems, such as Git, Mercurial, or Subversion. While version control systems have helped developers for nearly 40 years and are constantly getting updated, there are still functionalities that do not exist. One example of that is having a good way of managing dependencies and allowing developers to download projects without having to handle dependency problems manually. The solutions that version control systems offer to help manage dependencies (e.g., Git’s submodules or Mercurial’s subrepositories), do not enable developers a fail-safe download or build the project if it contains dependency problems.In this study, a case study was conducted at Saab Training and Simulation to explore methods for dependency management as well as discuss and highlight some of the problems that emerge when working with dependencies in multi-repositories.An argument can be made that the functionality of dependency management systems, both package managers and version control systems’ solutions are not up to date on how dependencies are used in the development, during this time.In this paper, a novel approach to dependency management is introduced with the possibility to describe the dependencies dynamically by providing the utility to describes usages of a repository (such as simulation of hardware or the main project). As well as discussing the necessary functionalities that are required to handle such a system.By re-opening the dialog about dependency management as well as describing problems that arise in such environments, the goal is to inspire further research within these areas.

This thesis proposes an implementation for a decentralized version of the Git version controlsystem. This is achieved using a simple distributed DAG ledger. The thesis analyzeshow the decentralization of Git affects security. Use and misuse cases are used to compareand evaluate conventional Git web services and a decentralized version of Git. Theproposed method for managing the state of the Git project is described as a voting systemwhere participants in a Git project vote on changes to be made. The security evaluationfound that the removal of privileged roles in the Git version control system, mitigated thepossibility of malicious maintainers taking over the project. However, with the introductionof the DAG ledger and the decentralization, the possibility of a malicious actor takingover the network using Sybil attack arises, which in turn could cause the same issues as amalicious maintainer.

No description available.

Submitted by Irene Nascimento (irene.kessia@ufpe.br) on 2016-09-27T18:16:18Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) dissertação_gjcc.pdf: 1929523 bytes, checksum: 59a910a15e3537942754d106de378d19 (MD5) / Made available in DSpace on 2016-09-27T18:16:18Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) dissertação_gjcc.pdf: 1929523 bytes, checksum: 59a910a15e3537942754d106de378d19 (MD5) Previous issue date: 16-02-29 / FACEPE / During the integration of code contributions resulting from development tasks, one likely has to deal with conflicting changes and dedicate substantial effort to resolve conflicts. While unstructured merge tools try to automatically resolve part of the conflicts via textual similarity, semistructured tools try to go further by exploiting the syntactic structure of part of the artefacts involved. To understand the impact of the unstructured and semistructured merge approaches on integration effort (Productivity) and correctness of the merging process (Quality), we conduct two empirical studies. In the first one, aiming at increasing the existing body of evidence and assessing results for systems developed under an alternative version control paradigm, we replicate an experiment to compare the unstructured and semistructured approaches with respect to the number of conflicts reported by both merge approaches. We used both semistructured and unstructured merge in a sample 2.5 times bigger than the original study regarding the number of projects and 18 times bigger regarding the number of performed merges, and we compared the occurrence of conflicts. Similar to the original study, we observed that semistructured merge reduces the number of conflicts in 55% of the performed merges of the new sample. Besides that, the observed average conflict reduction of 62% in these merges is far superior than what has been observed before. We also bring new evidence that the use of semistructured merge can reduce the occurrence of conflicting merges by half. In order to verify the frequency of false positives and false negatives arising from the use of these merge approaches, we move forward and we conduct a second empirical study. We compare the unstructured and semistructured merge approaches by reproducing more than 30,000 merges from 50 projects, and collecting evidence about reported conflicts that do not represent interferences between development tasks (false positives), and interferences not reported as conflicts (false negatives). In particular, our assumption is that false positives amount to unnecessary integration effort because developers have to resolve conflicts that actually do not represent interferences. Besides that, false negatives amount to build issues or bugs, negatively impacting software quality and correctness of the merging process. By analyzing such critical factors we hope to guide developers on deciding which approach should be used in practice. Finally, our results show that semistructured merge eliminates a significant part of the false positives reported by unstructured merge, but brings false positives of its own. The overall number of false positives is reduced with semistructured merge, and we argue that the conflicts associated to its false positives are easier to resolve when comparing to the false positives reported by unstructured merge. We also observe that more interferences were missed by unstructured merge and reported by semistructured merge, but we argue that the semistructured merge ones are harder to detect and resolve than the other way around. Finally, our study suggests how a semistructured merge tool could be improved to eliminate the extra false positives and negatives it has in relation to unstructured merge. / Durante a integração de contribuições de código resultantes das tarefas de desenvolvimento, frequentemente desenvolvedores têm que lidar com alterações conflitantes e dedicar considerável esforço para resolver conflitos. Enquanto as ferramentas de integração não-estruturadas tentam resolver automaticamente parte dos conflitos através de similaridade textual, ferramentas semiestruturadas tentam ir mais longe, explorando a estrutura sintática de parte dos artefatos envolvidos. Para entender o impacto das abordagens de integração não-estruturada e semiestruturada sobre esforço de integração (Produtividade) e corretude do processo de integração (Qualidade), nós realizamos dois estudos empíricos. No primeiro, com o objetivo de aumentar o atual corpo de evidência e avaliar resultados para sistemas desenvolvidos usando um paradigma de controle de versão alternativo, nós replicamos um experimento para comparar a abordagem não-estruturada e semiestruturada de acordo com o número de conflitos reportados por ambas as abordagens. Nós usamos tanto a integração semiestruturada quanto a não-estruturada em uma amostra 2,5 vezes maior do que a do estudo original em relação ao número de projetos e 18 vezes maior em relação ao número de integrações realizadas, e comparamos a ocorrência de conflitos. Semelhante ao estudo original, observamos que a integração semiestruturada reduz o número de conflitos em 55% das integrações da nova amostra. Além disso, a redução de conflitos média observada de 62% nestas integrações é muito superior à observada anteriormente. Nós também trazemos nova evidência de que o uso da abordagem semiestruturada pode reduzir a ocorrência de integrações com conflitos pela metade. Com o intuito de verificar a frequência de falsos positivos e falsos negativos originados do uso dessas abordagens, nós seguimos adiante e conduzimos um segundo estudo empírico. Nós comparamos as abordagens reproduzindo mais de 30.000 integrações de 50 projetos, coletando evidência sobre os conflitos reportados que não representam interferências entre as tarefas de desenvolvimento (falsos positivos), e interferências não reportadas como conflitos (falsos negativos). Em particular, a nossa suposição é de que falsos positivos denotam esforço desnecessário de integração porque os desenvolvedores têm que resolver conflitos que, na realidade, não representam interferências. Além disso, falsos negativos denotam problemas de build ou bugs, impactando negativamente a qualidade do software e corretude do processo de integração. Ao analisar esses fatores críticos, esperamos orientar os desenvolvedores em decidir qual abordagem deve ser usada na prática. Finalmente, nossos resultados mostram que a abordagem semiestruturada elimina uma parte significativa dos falsos positivos reportados pela abordagem não-estruturada, mas traz falsos positivos próprios. O número global de falsos positivos é reduzido com a integração semiestruturada, e nós argumentamos que os conflitos associados aos seus falsos positivos são mais fáceis de resolver quando comparados aos falsos positivos reportados pela abordagem não-estruturada. Observamos, também, que mais interferências deixaram de ser detectadas pela abordagem não-estruturada, mas foram detectadas pela semiestruturada. No entanto, nós acreditamos que as interferências não detectadas pela abordagem semiestruturada são mais difíceis de detectar e resolver. Por fim, nosso estudo sugere como uma ferramenta de integração semiestruturada poderia ser melhorada para eliminar os falsos positivos e falsos negativos adicionados que possui em relação à não-estruturada.

Text describes some of the most common tools used for project management and bug/issue tracking. Describes advantages and possibilities of connecting them to more complex systems of administration and version control. It also describe analysis and design of new FITkit specific system, and ways to catch errors in QDevKit application. The rest of text describes whole implementation including examples of applied constructions and logical layout.

En génie logiciel collaboratif, les systèmes de contrôle de version (SCV) jouent un rôle crucial dans la gestion des changements de code, la promotion de la collaboration et la garantie de l'intégrité des projets partagés. Cette importance s'étend à l'ingénierie dirigée par les modèles (IDM), où les experts du domaine conçoivent des modèles spécifiques au domaine (MSD). Dans ce contexte, la collaboration avec les SCV permet de coordonner les changements de modèles et de préserver l'intégrité des MSD. Cependant, les solutions existantes se concentrent principalement sur des approches génériques, considérant les modèles comme du texte générique. Ces SCV rapportent les différences entre les versions des modèles d'une manière abstraite et non intuitive pour les experts du domaine. Cela pose également des défis lors de la résolution des conflits et de la fusion des modèles, ce qui ajoute de la complexité au flux de travail des experts du domaine. L'objectif de cette thèse est de fournir des SCV spécifiques à un domaine donné en se concentrant sur les deux principaux composants des SCV, à savoir la différenciation et la fusion. Nous présentons DSMCompare, un outil de comparaison de modèles spécifique au domaine, intégré avec des capacités de détection, de résolution et de fusion de conflits de triplets de versions. DSMCompare fournit des représentations concises des différences et conflits à différents niveaux de granularité, tout en utilisant la syntaxe graphique des MSD originaux. Dans nos évaluations, DSMCompare a démontré des améliorations notables par rapport aux solutions génériques de différenciation et de fusion, notamment une réduction de la verbosité des différences rapportée, des différences exprimée en utilisant la sémantique du domaine, une détection précise des différences sémantiques et des conflits entre différentes versions d’un modèle, une résolution correcte des conflits, une diminution des interactions manuelles requises et une amélioration globale de l'efficacité pour les experts du domaine. / In the context of collaborative software engineering, version control systems (VCS) play a crucial role in managing code changes, promoting collaboration, and ensuring the integrity of shared projects. This significance extends to model-driven engineering (MDE), where domain experts design domain-specific models (DSM). In this context, collaborating with VCS aids in coordinating model changes and preserving the integrity of DSMs. However, existing solutions primarily focus on generic approaches, considering models as generic text. VCS report the differences between model versions in an abstract and unintuitive way for domain experts. This also poses challenges when resolving conflicts and merging models, adding complexity to the workflow of domain experts. The goal of this thesis is to provide domain-specific VCS for domain experts, focusing on the two main components of VCS, namely differencing and merging. We introduce DSMCompare, a domain-specific model comparison tool integrated with three-way conflict detection, resolution, and merging capabilities. DSMCompare provides concise representations of differences and conflicts at different levels of granularity, while using the graphical syntax of the original DSMs. In our evaluations, DSMCompare demonstrated significant improvements over generic differencing and merging solutions, including a reduction in reported difference verbosity, differences expressed using the semantics of the domain, accurate detection of semantic differences and conflicts between different versions of a model, correct conflict resolution, a reduction in manual interactions needed, and an overall improvement in efficiency for domain experts.