| Recognized above all as a technical task, the construction of software is nevertheless more and more acknowledged as a genuine human experience for the most part. While being an intensive cognitive task in the same time as it also constitutes "an exercise in complex interrelationships" as expressed by Fred Brooks in 1975 (Brooks, 1975), many human factors are thus involved during the development of a software. This document presents an empirical research that was undertaken in a industrial software engineering setting with the aim of examining one of the various human aspects of software development: the ad hoc collaborative activities taking place between teammates during a software development process.;To tackle this ambitious enterprise, a research methodology adapted for the study of the particular object of interest as well as the study field setting had to be designed and executed. This methodology, largely inspired from human sciences such as sociology as well as psychology, consisted of a participant observation where the observer took the role of a complete participant as a full-fledged member of the team. During the fairly extensive observation period that lasted several months, various data collection methods have been employed and included the audio-video recording of working sessions, the capture of electronic mails exchanged by the team members as well as a back up of the source code of the software developed by the observed team and all artefacts found on the file server shared by the team.;The analysis of the mass of data collected during the observation phase has been realized by the technique of the Exploratory Sequential Data Analysis (ESDA) as proposed by Fisher and Sanderson (Sanderson & Fisher, 1994; Fisher & Sanderson, 1993- ). This technique particularly well-suited for exploratory and descriptive researches as this one proposes, among others, the transformation of qualitative data into quantitative data in order to reduce the variability of the data and to facilitate further statistical calculations to obtain an array of additional derived data.;From the quite extended analysis phase that have been realized following this technique, several interesting results were obtained, results confirming firstly the importance of the phenomenon but more importantly the opportunistic nature of ad hoc collaborative activities and the role of different identified actors who are taking part. Several obtained results and their impacts have been revealed and discussed in the course of several scientific communications such as conferences and journal papers. Three of these articles constitute the body of the present document.;The motivations for this endeavour were numerous, but among others, because some researches such the one conducted by Perry, Staudenmayer and Votta (Perry et al., 1994- ) as well as another piloted by Robillard and Robillard (Robillard & Robillard, 2000) revealed the unsuspected importance of this phenomenon in term of time spent by the developers along an entire software project. Despite this really important reality, no known research has tried to characterize this type of activities as well as the communication that ensue in order to determine their impact on the whole software process. Therefore, it was legitimate to believe that a better understanding of this type of activities would help to better comprehend many pivotal aspects of software engineering, such as collaboration, coordination and, more generally, work team dynamics and help us to propose ultimately further process enhancements with the intend of making collaboration during a software engineering process more efficient between involved stakeholders. Furthermore, this better comprehension could help as well the definition of more adapted software development environment comprising tools intended to support collaboration in a team, as much localized in the same office as geographically distributed, since that it has been found that the distance, in the latter context, raises barriers to informal communications, such as those studied by this research, leading to exaggerated coordination problems (Herbsleb & Grinter, 1999).;Finally, from the obtained results of the present study, a number of hypothesis could be inducted and validated by further confirmatory researches forging a new theoretical base of knowledge in the domain. This better understanding of the phenomenon, which until today was practically unexplored, can also guide subsequent practices formalization in order to attempt, as stated above, a number of such practices were also proposed in the course of submitted and published papers. |
PDF Full Text Request