Exploring the Influence of Identifier Names on Code Quality: An empirical study

Full text available as:

Preview

PDF (Accepted Manuscript) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Download (178Kb) | Preview

Abstract

Given the importance of identifier names and the value of naming conventions to program comprehension, we speculated in previous work whether a connection exists between the quality of identifier names and software quality. We found that flawed identifiers in Java classes were associated with source code found to be of low quality by static analysis. This paper extends that work in three directions. First, we show that the association also holds at the finer granularity level of Java methods. This in turn makes it possible to, secondly, apply existing method-level quality and readability metrics, and see that flawed identifiers still impact on this richer notion of code quality and comprehension. Third, we check whether the association can be used in a practical way. We adopt techniques used to evaluate medical diagnostic tests in order to identify which particular identifier naming flaws could be used as a light-weight diagnostic of potentially problematic Java source code for maintenance.