Supporting software evolution through feedback on executing/skipping energy tests for proposed source code changes. Issue 4 (20th February 2019)

• Record Type: Journal Article
• Authors: Sahin, Cagri; Pollock, Lori; Clause, James
• Is Part Of: Journal of software. Volume 31:Issue 4(2019)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2019-02-20
• Subjects: android applications -- repository mining -- software evolution -- software energy testing -- source code changes; Software engineering -- Periodicals; Computer software -- Development -- Periodicals; Software maintenance -- Periodicals; 005.1
• DOI: 10.1002/smr.2155
• Languages: English
• ISSNs: 2047-7473

Abstract

Abstract: With the increasing use of battery‐powered devices comes the need to test mobile applications for energy consumption and energy issues. Unfortunately, energy testing is expensive because it is a manual, labor‐intensive process that often requires multiple, separate, energy‐measuring devices to collect energy usage data. The high costs of energy testing can negatively affect the planning process of application evolution. For example, developers might be limited in the number of changes they can include in a release because they must conservatively plan to conduct energy testing after each change. In this paper, we present a new approach to provide developers with feedback on executing/skipping energy tests for proposed code changes. Our technique leverages change impact analysis and precomputed API energy usage information. More specifically, for a proposed change, the technique predicts whether energy testing will be required, and if so, which energy tests will need to be run. Such information may allow developers to avoid spending unnecessary time for energy testing and develop an effective application evolution timeline. To investigate the feasibility of our technique, we implemented a prototype for Android applications and conducted three case studies at different granularity levels on 10 Android applications. Abstract : We present a technique, Energy Retest Umpire (ERU), that provides feedback on executing/skipping energy tests for proposed code changes. This feedback can allow developers to avoid spending unnecessary time for energy testing and develop an effective application evolution timeline. At a high level, ERU leverages change impact analysis and pre‐computed API energy usage information. To investigate the feasibility of our technique, we implemented a prototype for Android applications and conducted case studies at different granularity levels.