Devika Sondhi

Venkatesh Vinayakarao, Devika Sondhi, Sumit Keswani et al.

There are often multiple ways to implement the same requirement in source code. Different implementation choices can result in code snippets that are similar, and have been defined in multiple ways: code clones, examples, simions and variants. Currently, there is a lack of a consistent and unambiguous definition of such types of code snippets. Here we present a characterization study of code variants - a specific type of code snippets that differ from each other by at least one desired property, within a given code context. We distinguish code variants from other types of redundancies in source code, and demonstrate the significant role that they play: about 25% to 43% of developer discussions (in a set of nine open source projects) were about variants. We characterize different types of variants based on their code context and desired properties. As a demonstration of the possible use of our characterization of code variants, we show how search results can be ranked based on a desired property (e.g., speed of execution).

Sandeep Hans, Atul Kumar, Toshikai Yasue et al.

Recent advances in Large Language Model (LLM) based Generative AI techniques have made it feasible to translate enterprise-level code from legacy languages such as COBOL to modern languages such as Java or Python. While the results of LLM-based automatic transformation are encouraging, the resulting code cannot be trusted to correctly translate the original code, making manual validation of translated Java code from COBOL a necessary but time-consuming and labor-intensive process. In this paper, we share our experience of developing a testing framework for IBM Watsonx Code Assistant for Z (WCA4Z) [5], an industrial tool designed for COBOL to Java translation. The framework automates the process of testing the functional equivalence of the translated Java code against the original COBOL programs in an industry context. Our framework uses symbolic execution to generate unit tests for COBOL, mocking external calls and transforming them into JUnit tests to validate semantic equivalence with translated Java. The results not only help identify and repair any detected discrepancies but also provide feedback to improve the AI model.