This study investigates the potential of Large Language Models (LLMs) to automate and enhance software maintenance tasks, focusing on bug detection and code refactoring. Traditional software maintenance, which includes debugging and code optimization, is time-consuming and prone to human error. With advancements in artificial intelligence, LLMs like ChatGPT and Copilot offer promising capabilities for automating these tasks. Through a series of quasi-experiments, we evaluate the effectiveness of ChatGPT 3.5, ChatGPT 4 (Grimoire GPT), and GitHub Copilot. Each model was tested on various code snippets to measure their ability to identify and correct bugs and refactor code while maintaining its original functionality. The results indicatethat ChatGPT 4 (Grimoire GPT) outperforms the other models, demonstrating superior accuracy and effectiveness, with success percentages of 87.5% and 75% in bug detection and code refactoring respectively. This research highlights the potential of advanced LLMs to significantly reduce the time and cost associated with software maintenance, though human oversight is still necessary to ensure code integrity. The findings contribute to the understanding of LLM capabilities in real-world software engineering tasks and pave the way for more intelligent and efficient software maintenance practices.

Rapporten belyser arbetet kring det kandidatarbete som utfördes av åtta studenter i kursen TDDD96 - Kandidatprojekt i programvaruutveckling på Linköpings universitet undervårterminen 2020. Uppgiften som utfördes var att utveckla en webbapplikation för att filtrera övervakningsvideor för Polismyndigheten. Resultatet av arbetet blev ett fungerandekoncepttest som släppts som öppen källkod under namnet Mycroft samt en användarmanual. Rapporten innehåller en bakgrund till projektet och projektgruppen, en teoridel sombeskriver de verktyg och utvecklingsmetoder som projektgruppen har använt samt en delsom redovisar gruppens utvecklingsmetod och andra administrativa metoder. Resultatetbeskriver den slutgiltiga produkten samt resultatet från gruppens arbetsprocesser. Rapporten avslutas med en diskussion gällande resultatet, metoden och framtiden för projektgruppen och produkten. Rapporten innehåller även åtta individuella fördjupningsarbetenfrån vardera gruppmedlem.

Sofiaängen är en psykoterapeutisk dagverksamhet och skola som ligger på Södermalm i Stockholm. Sofiaängen riktar sig till ungdomar mellan 13–20 år med psykiska och sociala problem. Sofiaängen var i behov av ett fullständigt journalsystem för att kunna underlätta deras arbetsrutiner med bokföring av patientbehandlingar. Då deras nuvarande system inte är färdigställd och behövs vidareutvecklas. Detta ledde till att arbetet delades upp i två delar, först att ta fram ett förslag på en generell processmodell för hantering av en icke färdigställda IT-system till akademin och därefter att leverera ett fullt fungerande journalsystem åt Sofiaängen. Examensarbetet har resulterat till att man har tagit fram ett förslag på generell processmodell för hantering av icke färdigställda projekt samt ett fullt fungerande journalsystem åt Sofiaängen som kan sättas direkt i drift. / Sofiaängen is a psychotherapeutic day school and school located at Södermalm in Stockholm. Sofiaängen is aimed at young people between 13-20 years with mental and social problems. Sofiaängen needed a complete journal system to facilitate their work routines with records of patient treatments. Their current system is not completed and needed further development. Thus, the work was divided into two parts, first to develop a proposal for a general process model for managing an unfinished IT system to the academy and then to deliver a fully functioning journal system for Sofiaängen. The thesis has resulted in a proposal for a general process model for handling unfinished projects as well as a fully functioning journal system for Sofiaängen, which can be put into operation immediately.

There is a great demand for fast deliveries of improved and updated software in different software development areas, like Internet of Things, web, and cloud, in today’s digitalized world. Software developers and organizations must adapt to be able to deliver according to customers’ wishes, to be able to retain them, and remain competitive with other organizations. Continuous integration and continuous delivery (CI/CD) are methods used within the software development world, allowing developers to automate parts of their work to develop and deliver software faster and with better quality. Tools used for CI/CD come with different benefits and performances making it difficult for developers to choose a tool. There are numerous tools to choose from, and there is a lack of performance comparisons of them. This thesis aims to give developers a performance comparison between the two well-known CI/CD tools, GitHub Actions and Jenkins, to facilitate their choice of a CI/CD tool. The research was qualitative, inductive, and comparative. A literature study and practical tests were conducted to study the performance differences between the two wellknown CI/CD tools, GitHub Actions and Jenkins. The literature study was conducted f irst and gave the necessary knowledge to perform the practical tests, and the practical tests gave the actual results. The practical tests were performed on two different software projects ,and two different tests per projec, per server were conducted. The results from both projects indicated apparent differences in performance between GitHub Actions and Jenkins, as Jenkins ran faster than GitHub Actions while running on a Windows server, and GitHub Actions ran faster than Jenkins while running on an Ubuntu server. These findings indicate that the two well-known CI/CD tools perform differently depending on the server the developers would use these tools. It can not be concluded that one of the tools has better performance than the other; instead, one tool has better performance depending on the operating system the tool is running on. If the developers were to use the tools on an Ubuntu server, GitHub Actions would be the preferred tool, and if they were to use the tool on a Windows server, Jenkins would be the preferred tool. / Det finns en stor efterfrågan på snabba leveranser av förbättrad och uppdaterad mjukvara i olika mjukvaruutvecklings områden så som Sakernas Internet, webb och moln i dagens digitaliserade värld. Mjukvaruutvecklare och organisationer måste anpassa sig för att kunna leverera till kundernas önskemål för att kunna behålla dom och förbli konkurrenskraftiga med andra organisationer. Kontinuerlig integration och kontinuerlig leverans (CI/CD) är metoder som används inom mjukvaruutvecklings världen, så att utvecklare kan automatisera delar av sitt arbete för att utveckla och leverera mjukvara snabbare och med bättre kvalité. Verktyg som används för CI/CD kommer med olika fördelar och prestanda som gör det svårt för utvecklare att välja ett verktyg. Det finns många verktyg att välja mellan och det finns en brist på prestandajämförelser av dem. Detta examensarbete syftar till att ge utvecklare en prestandajämförelse mellan de två välkända CI/CD-verktygen GitHub Actions och Jenkins, för att underlätta utvecklarens val av ett CI/CD-verktyg. En kvalitativ, induktiv och komparativ forskningsmetod användes för att genomföra denna studie. En litteraturstudie och praktiska tester genomfördes för att studera prestandaskillnader mellan de två välkända CI/CD-verktygen GitHub Actions och Jenkins. Litteraturstudien genomfördes först och gav författarna nödvändiga kunskap för att utföra dem praktiska testerna, dem praktiska testerna gav de faktiska resultaten. Praktiska testerna utfördes på två olika mjukvaruprojekt och två olika tester per projekt, en per server genomfördes. Resultaten från båda projekten visade på uppenbara skillnader i prestanda mellan GitHub Actions och Jenkins. Då Jenkins kördes snabbare än GitHub Actions när körningen kördes på en Windows server och GitHub Actions kördes snabbare än Jenkins när de kördes på en Ubuntu server. Dessa resultat tyder på att de två välkända CI/CD-verktygen fungerar olika beroende på vilken server utvecklarna skulle använda dessa verktyg på. Det går inte att dra slutsatsen att ett verktyg är bättre över det andra, i stället har ett verktyg bättre prestanda beroende på vilket operativ system verktyget körs på. Om utvecklarna skulle använda verktygen på en Ubuntu server skulle GitHub Actions vara det föredragna verktyget och om utvecklarna skulle använda verktyget på en Window server skulle Jenkins vara det föredragna verktyget.

Scratch is a block programming language which introduces beginners to programming. Instead of code the user has access to a set of blocks with text and icons, explaining how the block will affect the program that is written. The connection between one block and the corresponding code can be hard to understand for the beginner. The goal of this project is therefore to develop a user-friendly, flashcard-based mobile application to show the target group of 8- to 16-year-olds that behind every block there is code. The application is developed in TypeScript, using React Native as framework and the developer tool Expo for setting up and publishing of the application. The final application consists of 6 different screens; a starting screen, an information screen, a menu, a submenu, an "under development"-screen and a flashcard view. The user can navigate between the screens and by choosing a specific block the flashcard view displays a flashcard with the block and the corresponding translation in Python. The goal of the project is fulfilled, and with a testing group it is also confirmed that the application is user-friendly. Although the goal is achieved, the conclusion that the step between block programming and syntax is hard can be drawn, with difficulties in translating the blocks appearing along the way.

Software systems are becoming increasingly complex, with developers frequentlyutilizing numerous dependencies. In this landscape, accurate tracking and understanding of dependencies within JavaScript and TypeScript codebases are vital formaintaining software security and quality. However, there exists a gap in how existing vulnerability scanning tools, such as Dependabot, convey information aboutthe usage of these dependencies. This study addresses the problem of providing amore comprehensive dependency usage overview, a topic critical to aiding developers in securing their software systems. To bridge this gap, a custom pipeline wasimplemented to supplement Dependabot, extracting the dependencies identified asvulnerable and providing specific information about their usage within a repository.The results highlight the pros and cons of this approach, showing an improvement inthe understanding of dependency usage. The effort opens a pathway towards moresecure software systems.

This thesis presents an in-depth investigation on the subject of how natural language processing and machine learning techniques can be utilized in order to perform a comprehensive analysis of programming issues found in different open source project repositories hosted on GitHub. The research is focused on examining issues gathered from a number of JavaScript repositories based on their user generated textual description. The primary goal of the study is to explore how natural language processing and machine learning methods can facilitate the process of identifying and categorizing distinct issue types. Furthermore, the research goes one step further and investigates how these same techniques can support users in searching for potential solutions to these issues. For this purpose, an initial proof-of-concept implementation is developed, which collects over 30 000 JavaScript issues from over 100 GitHub repositories. Then, the system extracts the titles of the issues, cleans and processes the data, before supplying it to an unsupervised clustering model which tries to uncover any discernible similarities and patterns within the examined dataset. What is more, the main system is supplemented by a dedicated web application prototype, which enables users to utilize the underlying machine learning model in order to find solutions to their programming related issues. Furthermore, the developed implementation is meticulously evaluated through a number of measures. First of all, the trained clustering model is assessed by two independent groups of external reviewers - one group of fellow researchers and another group of practitioners in the software industry, so as to determine whether the resulting categories contain distinct types of issues. Moreover, in order to find out if the system can facilitate the search for issue solutions, the web application prototype is tested in a series of user sessions with participants who are not only representative of the main target group which can benefit most from such a system, but who also have a mixture of both practical and theoretical backgrounds. The results of this research demonstrate that the proposed solution can effectively categorize issues according to their type, solely based on the user generated free-text title. This provides strong evidence that natural language processing and machine learning techniques can be utilized for analyzing issues and automating the overall learning process. However, the study was unable to conclusively determine whether these same methods can aid the search for issue solutions. Nevertheless, the thesis provides a detailed account of how this problem was addressed and can therefore serve as the basis for future research.

Cílem této práce je vytvořit systém, který je schopný automaticky určit množinu testů, které mají být spuštěny, když dojde v ComplianceAsCode projektu ke změně. Navržená metoda vybírá množinu testů na základě statické analýzy změněných zdrojových souborů, přičemž bere v úvahu vnitřní strukturu ComplianceAsCode. Vytvořený systém je rozdělen do čtyř částí - získání změn s využitím verzovacího systému, statická analýza různých typů souborů, zjištění souborů, které jsou ovlivněny těmi změnami, a výpočet množiny testů, které musí být spuštěny pro danou změnu. Naimplementovali jsme analýzu několika různých typů souborů a náš systém je navržen tak, aby byl jednoduše rozšiřitelný o analýzy dalších typů souborů. Vytvořená implementace je nasazena na serveru, kde automaticky analyzuje nové příspěvky do ComplianceAsCode projektu. Automatické spouštění informuje přispěvatelé a vývojáře o nalezených změnách a doporučuje, které testy by pro danou změnu měly být spuštěny. Tím je ušetřen čas strávený při kontrole správnosti příspěvků a čas strávený spouštěním testů.

Táto diplomová práca sa zaoberá prieskumom a taxonómiou sieťových forenzných nástrojov. Popisuje základné informácie o sieťovej forenznej analýze, vrátane procesných modelov, techník a zdrojov dát používaných pri forenznej analýze. Ďalej práca obsahuje prieskum existujúcich taxonómií sieťových forenzných nástrojov vrátane ich porovnania, na ktorý naväzuje prieskum sieťových forenzných nástrojov. Diskutované sieťové nástroje obsahujú okrem nástrojov spomenutých v prieskume taxonómií aj niektoré ďalšie sieťové nástroje. Následne sú v práci detailne popísané a porovnané datasety, ktoré sú podkladom pre analýzu jednotlivými sieťovými nástrojmi. Podľa získaných informácií z vykonaných prieskumov sú navrhnuté časté prípady použitia a nástroje sú demonštrované v rámci popisu jednotlivých prípadov použitia. Na demonštrovanie nástrojov sú okrem verejne dostupných datasetov použité aj novo vytvorené datasety, ktoré sú detailne popísane vo vlastnej kapitole. Na základe získaných informácií je navrhnutá nová taxonómia, ktorá je založená na prípadoch použitia nástrojov na rozdiel od ostatných taxonómií založených na NFAT a NSM nástrojoch, uživateľskom rozhraní, zachytávaní dát, analýze, či type forenznej analýzy.

The development of generative AI has made great strides. More and more organizations are looking into implementing this new technology to increase productivity and efficiency. One of these new AI system- development tools is GitHub Copilot. The tool has shown great promise by offering functions such as automatic code generation, but this does not come without faults, as the generated code may be lacking in quality. How system developers within organizations experience this new technology is unknown, nor is it a worthwhile investment for the organizations in question. A qualitative study with semi-structured interviews has been carried out to capture the experiences of system developers concerning GitHub Copilot. The study was based on the theoretical framework Technology Acceptance Model 2 (TAM 2), in which some selected factors were used to describe the intention to use the system. A study was conducted to identify factors that cause an increase and/or a decrease in user acceptance.  We believe this would provide insights into what context GitHub Copilot would lead to increased productivity and efficiency. Based on the four factors studied, perceived usefulness, perceived ease of use, job relevance, and output quality, the study concludes with factors that affect a user’s intention to use GitHub Copilot. The study reveals that system developers perceive the usage of GitHub Copilot as positive. They believe that it has the potential to increase both productivity and efficiency. They perceive the tool as easy to get started with and easy to use. The quality of the generated code is perceived as somewhat lacking, but this did not affect their acceptance of the system.