Code smells are indicators of potential design issues in source code that can affect the overall software quality. Detecting these smells has been the focus of extensive research efforts using both traditional metric-based techniques and, more recently, deep learning models. Among these, Large Language Models (LLMs) have emerged as powerful tools due to their strong capabilities in code understanding and classification tasks. LLMs such as CodeT5 and CodeBERT were employed to detect various smells with promising results. However the integration of these models into software engineering workflows remains challenging as current approaches often lack traceability, modularity, and reusability, making it difficult to track results, compare configurations, or systematically explore model variations. To address these limitations, we introduce in this paper DetectorsForge, a Software Product Line (SPL) that enables the systematic configuration, customization, and execution of LLM-based code smell detectors. DetectorsForge unifies models, datasets, transfer learning techniques, and evaluation methods to adapt pre-trained models to new downstream tasks. These components are integrated within a reusable architecture that supports the automated derivation of code smell detection variants. Experimental results demonstrate the applicability of our approach for generating various code smell detectors.

Modern real-time system development methodologies describe a stage in which application tasks are deployed onto an execution platform.The deployment process is divided into two steps: (i) task placement on processors and (ii) task scheduling to determine their execution order.The overall performance of the deployment model depends on the two steps, which are interdependent.In this paper, a new method based on reinforcement learning techniques, called PSRL, is proposed.PSRL explores all the feasible placements in the first step.In the second step, an optimal schedule is considered for each feasible placement.PSRL generates the optimal deployment, which corresponds to the placement and scheduling that minimize task response times.Application to case studies shows the applicability and quality of the obtained solutions when compared to related work.

International audience

This paper deals with the reconfigurable real-time systems that should be adapted to their environment under real-time constraints. The reconfiguration allows moving from one implementation to another by adding/removing/modifying parameters of real-time software tasks which should meet related deadlines. Implementing those systems as threads generates a complex system code due to the large number of threads, which may lead to a reconfiguration time overhead as well as the energy consumption and the memory allocation increase. Thus this paper proposes a multiobjective optimization approach for reconfigurable systems called MO2R2S for the development of a reconfigurable real-time system. Given a specification, the proposed approach aims to produce an optimal design while ensuring the system feasibility. We focus on three optimization criteria: 1) response time; 2) memory allocation; and 3) energy consumption. To address the portability issue, the optimal design is then transformed to an abstract code that may in turn be transformed to a concrete code which is specific to a procedural programming (i.e., POSIX) or an object-oriented language (i.e., RT-Java). The MO2R2S approach allows reducing the number of threads by minimizing the redundancy between the implementation sets. By an experimental study, such optimization permits to decrease the memory allocation by 28.89%, the energy consumption by 40.2%, and the response time by 61.32%.