Software Engineering (cs.SE)

Abstract: Despite the increasing interest in quantum computing, the aspect of development to achieve cost-effective and reliable quantum software applications has been slow. One barrier is the software engineering of quantum programs, which can be approached from two directions. On the one hand, many software engineering practices, debugging in particular, are bound to classical computing. On the other hand, quantum programming is closely associated with the phenomena of quantum physics, and consequently, the way we express programs resembles the early days of programming. Moreover, much of the software engineering research today focuses on agile development, where computing cycles are cheap and new software can be rapidly deployed and tested, whereas in the quantum context, executions may consume lots of energy, and test runs may require lots of work to interpret. In this paper, we aim at bridging this gap by starting with the quantum computing workflow and by mapping existing software engineering research to this workflow. Based on the mapping, we then identify directions for software engineering research for quantum computing.

Abstract: Code cloning, the duplication of code fragments, is common in software development. While some reuse aids productivity, excessive cloning hurts maintainability and introduces bugs. Hence, automatic code clone detection is vital. Meanwhile, large language models (LLMs) possess diverse code-related knowledge, making them versatile for various software engineering challenges. However, LLMs' performance in code clone detection is unclear and needs more study for accurate assessment.In this paper, we provide the first comprehensive evaluation of LLMs for clone detection, covering different clone types, languages, and prompts. We find advanced LLMs excel in detecting complex semantic clones, surpassing existing methods. Adding intermediate reasoning steps via chain-of-thought prompts noticeably enhances performance. Additionally, representing code as vector embeddings, especially with text encoders, effectively aids clone detection.Lastly, the ability of LLMs to detect code clones differs among various programming languages. Our study suggests that LLMs have potential for clone detection due to their language capabilities, offering insights for developing robust LLM-based methods to enhance software engineering.

Abstract: Enterprises in their journey to the cloud, want to decompose their monolith applications into microservices to maximize cloud benefits. Current research focuses a lot on how to partition the monolith into smaller clusters that perform well across standard metrics like coupling, cohesion, etc. However, there is little research done on taking the partitions, identifying their dependencies between the microservices, exploring ways to further reduce the dependencies, and making appropriate code changes to enable robust communication without modifying the application behaviour. In this work, we discuss the challenges with the conventional techniques of communication using JSON and propose an alternative way of ID-passing via APIs. We also devise an algorithm to reduce the number of APIs. For this, we construct subgraphs of methods and their associated variables in each class and relocate them to their more functionally aligned microservices. Our quantitative and qualitative studies on five public Java applications clearly demonstrate that our refactored microservices using ID have decidedly better time and memory complexities than JSON. Our automation reduces 40-60\% of the manual refactoring efforts.

Abstract: Successful deployment of Deep Neural Networks (DNNs), particularly in safety-critical systems, requires their validation with an adequate test set to ensure a sufficient degree of confidence in test outcomes. Mutation analysis, one of the main techniques for measuring test adequacy in traditional software, has been adapted to DNNs in recent years. This technique is based on generating mutants that aim to be representative of actual faults and thus can be used for test adequacy assessment. In this paper, we investigate for the first time whether mutation operators that directly modify the trained DNN model (i.e., post-training) can be used for reliably assessing the test inputs of DNNs. We propose and evaluate TEASMA, an approach based on post-training mutation for assessing the adequacy of DNN's test sets. In practice, TEASMA allows engineers to decide whether they will be able to trust test results and thus validate the DNN before its deployment. Based on a DNN model's training set, TEASMA provides a methodology to build accurate prediction models of the Fault Detection Rate (FDR) of a test set from its mutation score, thus enabling its assessment. Our large empirical evaluation, across multiple DNN models, shows that predicted FDR values have a strong linear correlation (R2 >= 0.94) with actual values. Consequently, empirical evidence suggests that TEASMA provides a reliable basis for confidently deciding whether to trust test results or improve the test set.