Scholarlink

Measurement is crucial in software engineering. Based on the results of measurement, critical software business decisions are made. Therefore, software metrics should have precise operational definitions describing how the measurement should be made. We examine this issue in the context of two well-known metrics, LOC and number of branches, as well as related test coverage metrics. We also show significant differences in how measurement tools understand the same metrics. We analyze the reasons for these differences and propose a unified approach to defining software metrics based on formal grammars.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

“To Measure Is to Know”, but Not in Software Engineering. A Call for Operational Definitions of Code Metrics

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Adam Roman,
  • Michał Mnich,
  • Jarosław Hryszko

摘要

Measurement is crucial in software engineering. Based on the results of measurement, critical software business decisions are made. Therefore, software metrics should have precise operational definitions describing how the measurement should be made. We examine this issue in the context of two well-known metrics, LOC and number of branches, as well as related test coverage metrics. We also show significant differences in how measurement tools understand the same metrics. We analyze the reasons for these differences and propose a unified approach to defining software metrics based on formal grammars.