Return
Automatic method change suggestion to complement multi-entity edits
DOI:10.1016/j.jss.2019.110441.png)
Abstract
En 中文
When maintaining software, developers sometimes change multiple program entities (i.e., classes, methods, and fields) to fulfill one maintenance task. We call such complex changes multi-entity edits. Consistently and completely applying multi-entity edits can be challenging, because (1) the changes scatter in different entities and (2) the incorrectly edited code may not trigger any compilation or runtime error. This paper introduces CMSuggester, an approach to suggest complementary changes for multi-entity edits. Given a multi-entity edit that (i) adds a new field or method and (ii) modifies one or more methods to access the field or invoke the method, CMSuggester suggests other methods to co-change for the new field access or method invocation. The design of CMSuggester is motivated by our preliminary study, which reveals that co-changed methods usually access existing fields or invoke existing methods in common. Our evaluation shows that based on common field accesses, CMSuggester recommended method changes in 463 of 685 tasks with 70% suggestion accuracy; based on common method invocations, CMSuggester handled 557 of 692 tasks with 70% accuracy. Compared with prior work ROSE, TARMAQ, and Transitive Association Rules (TAR), CMSuggester recommended more method changes with higher accuracy. Our research can help developers correctly apply multi-entity edits. (C) 2019 Elsevier Inc. All rights reserved.
Keywords:
Multi-entity edit
Common field access
Common method invocation
Change suggestion
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
4.1
Papers:
5.5K
Citations:
8.4K

