DepMiner: Automatic Recommendation of Transformation Rules for Method Deprecation

Oleksandr Zaitsev (1, 2, 3) , Stéphane Ducasse (3) , Nicolas Anquetil (3) , Arnaud Thiefaine (1)

1 Arolla
2 CRIStAL - Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189
3 RMOD - Analyses and Languages Constructs for Object-Oriented Application Evolution

Software applications often depend on external libraries and must be updated when one of those libraries releases a new version. To make this process easier, library developers try to reduce the negative effect of breaking changes by deprecating the API elements before removing them and suggesting replacements to the clients. Modern programming languages and IDEs provide powerful tools for deprecations that can reference the replacement or incorporate the rules written by library developers and use them to automatically update the client code. However, in practice library developers often miss the deprecation opportunities and fail to document the deprecations. In this work, we propose to help library developers support their clients with better deprecations. We rely on the transforming deprecations offered by Pharo and use data mining to detect the missing deprecation opportunities and generate the transformation rules. We implemented our approach for Pharo in a prototype tool called DepMiner. We have applied our tool to five open-source projects and proposed the generated deprecations to core developers of those projects. 63 recommended deprecations were accepted as pull requests.

Domaines: Génie logiciel [cs.SE]; Intelligence artificielle [cs.AI]; Langage de programmation [cs.PL]

Oleksandr Zaitsev, Stéphane Ducasse, Nicolas Anquetil, Arnaud Thiefaine. DepMiner: Automatic Recommendation of Transformation Rules for Method Deprecation. ICSR 2022 - 20th International Conference on Software and System Reuse, Jun 2022, Montpellier, France. ⟨hal-03647706⟩