DocumentCode :
175540
Title :
Studying Fine-Grained Co-evolution Patterns of Production and Test Code
Author :
Marsavina, Cosmin ; Romano, Daniela ; Zaidman, Andy
Author_Institution :
Delft Univ. of Technol., Delft, Netherlands
fYear :
2014
fDate :
28-29 Sept. 2014
Firstpage :
195
Lastpage :
204
Abstract :
Numerous software development practices suggest updating the test code whenever the production code is changed. However, previous studies have shown that co-evolving test and production code is generally a difficult task that needs to be thoroughly investigated. In this paper we perform a study that, following a mixed methods approach, investigates fine-grained co-evolution patterns of production and test code. First, we mine fine-grained changes from the evolution of 5 open-source systems. Then, we use an association rule mining algorithm to generate the co-evolution patterns. Finally, we interpret the obtained patterns by performing a qualitative analysis. The results show 6 co-evolution patterns and provide insights into their appearance along the history of the analyzed software systems. Besides providing a better understanding of how test code evolves, these findings also help identify gaps in the test code thereby assisting both researchers and developers.
Keywords :
data mining; program testing; public domain software; software engineering; association rule mining algorithm; coevolving test; fine-grained coevolution patterns; mixed methods approach; open-source systems; production code; software development practices; test code; Association rules; Educational institutions; Production; Software systems; Statistical analysis; Testing;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Source Code Analysis and Manipulation (SCAM), 2014 IEEE 14th International Working Conference on
Conference_Location :
Victoria, BC
Type :
conf
DOI :
10.1109/SCAM.2014.28
Filename :
6975653
Link To Document :
بازگشت