Macro Studies of FOSS Ecology

Abstract

The ecology of Free and Open Source Software (FOSS) is dotted by projects of every kind ranging from small desktop applications to large mission critical systems. To enable maximum visibility among the developer community, these projects are often hosted in community project management portals. The current work studies one such portal, sourceforge.net by analyzing the data of 200,000 projects and 2 million developers for the period Feb 2005 to Aug 2009. The scope of the present study includes the analysis of developer contribution. The slow growth rate of developer community and high number of single developer projects are the major findings of the present work.
Chapter Preview
Top

Procedure

Sourceforge.net maintains the data in a relational database. To extract the developer count from this dataset the relation USER GROUP is selected. The structure of the relation is given in Table 1.

The attribute USER ID is used to count the number of developers for each month and GROUP ID is used for counting the projects. The process for counting developers and projects for all months is given below

S ←{S1,S2...S52} original datasetsAi ← φ(X), ∀X ∈S select unique developersA ← {A1, A2...A52}Bi ← ψ(X), ∀X ∈ S select unique projectsB ← {B1, B2...B52}xi ← |C|, ∀C ∈ A count the developersyi ← |C|, ∀C ∈ B count the projectsD ← {x1, x2, ..., x52} list of developer countP ← {y1, y2, ..., y52} list of project count

Using the relation USER GROUP we can also extract the number of developers subscribing to each project. The procedure for doing this activity is mentioned below

S ← {S1, S2...S52} original datasetsA ← φ(X), X ∈ S select unique developersB ← ψ(X), X ∈ S select unique projectsC ← {x, |x ∈ A, y ∈ B, x → y}Top

Results And Discussion

The growth pattern of developers and projects in Sourceforge.net is summarized in Figure 1. The number of developers and projects are significant metrics since it can be used to gauge the interest among the general public to participate in FOSS development. The number of developers (Developer count) for all 54 datasets is given in Figure 2. The number of projects (Project count) in these time period is in Figure 3.

The statistical summary of the Developer count is given in Table 2. The statistical summary of the Project count is given in Table 3

Complete Chapter List

Search this Book:
Reset