Análisis y evaluación del desarrollo de software tradicional vs el desarrollo de software guiado por pruebas en términos de calidad de software y productividad de los desarrolladores para una Empresa del sector público

Abstract

Experimentation is a discipline that allows us to create models from which it is possible to relate processes with objects; these follow a cycle and, depending on the obtained results, knowledge is created (Wohlin, Runeson, Höst, Ohlsson, Regnell & Wesslén, 2012). TDD is a software development technique that has been sporadically used for decades but has recently emerged promoting good results; nevertheless there is little empirical evidence in the industrial context that neither supports nor denies the utility of this practice (Nagappan, Maximilien, Bhat & Williams, 2008). Nowadays there is not a knowledge body for TDD, thus the importance of performing replications. Experimental replications help us to confirm results from performed experiments and, in this manner, broaden the knowledge base available to researchers. (Mendonça, et al., 2008) According with the scientific literature, experiments performed in the industrial context that compare TDD with other software development techniques provide contradicting results and do not determine whether a method outperforms another. In March 2016, as part of the FidiPro project, ESEIL and GrlSE performed an experiment in a company in Finland where their hypothesis were similar to those of this experiment; due to confidentiality this company is referred as “Empresa F” throughout this document. This project includes the study of TDD and the performance of an experimental replication in a public company in Cuenca.