Lincoln S. Rocha

Peter Hamfelt, Ricardo Britto, Lincoln Rocha et al.

Machine learning (ML) has rapidly grown in popularity, becoming vital to many industries. Currently, the research on code smells in ML applications lacks tools and studies that address the identification and validity of ML-specific code smells. This work investigates suitable methods and tools to design and develop a static code analysis tool (MLpylint) based on code smell criteria. This research employed the Design Science Methodology. In the problem identification phase, a literature review was conducted to identify ML-specific code smells. In solution design, a secondary literature review and consultations with experts were performed to select methods and tools for implementing the tool. We evaluated the tool on data from 160 open-source ML applications sourced from GitHub. We also conducted a static validation through an expert survey involving 15 ML professionals. The results indicate the effectiveness and usefulness of the MLpylint. We aim to extend our current approach by investigating ways to introduce MLpylint seamlessly into development workflows, fostering a more productive and innovative developer environment.

Luan P. Lima, Lincoln S. Rocha, Carla I. M. Bezerra et al.

Modern programming languages (e.g., Java and C#) provide features to separate error-handling code from regular code, seeking to enhance software comprehensibility and maintainability. Nevertheless, the way exception handling (EH) code is structured in such languages may lead to multiple, different, and complex control flows, which may affect the software testability. Previous studies have reported that EH code is typically neglected, not well tested, and its misuse can lead to reliability degradation and catastrophic failures. However, little is known about the relationship between testing practices and EH testing effectiveness. In this exploratory study, we (i) measured the adequacy degree of EH testing concerning code coverage (instruction, branch, and method) criteria; and (ii) evaluated the effectiveness of the EH testing by measuring its capability to detect artificially injected faults (i.e., mutants) using 7 EH mutation operators. Our study was performed using test suites of 27 long-lived Java libraries from open-source ecosystems. Our results show that instructions and branches within $\mathtt{catch}$ blocks and $\mathtt{throw}$ instructions are less covered, with statistical significance than the overall instructions and branches. Nevertheless, most of the studied libraries presented test suites capable of detecting more than 70% of the injected faults. From a total of 12,331 mutants created in this study, the test suites were able to detect 68% of them.