The macro studies discussed the essential characteristics of the FOSS ecology. The size and nature of the projects and developers involved in FOSS development in a large project eco-system like Sourceforge.net were discussed. But the studies were limited in their scope. The important metrics like number of developers working on each project, their movement in the ecology, number of tasks which come up in projects and how effectively they are completed cannot be studied for all 150,000 projects. Therefore, it becomes necessary to study the important features of FOSS development by selecting few projects and subjecting them to rigorousanalysis.
Selection of Projects for Micro Analysis
To select few projects from a large pool of projects currently active in Sourceforge.net, the following options were available
- 1.
Randomly choose few sampleprojects
- 2.
Select one project from eachyear
- 3.
Select projects which meets certain criteria
Table 1.
Structure of table STATS GROUP_RANK_BYMONTH
| Column | Type | Modifiers |
| group id | integer | not null default 0 |
| rankdate | integer | not null default 0 |
| ranking | integer | not null default 0 |
| percentile | double precision | default 0.0 |
| score | bigint | not null default(0)::bigint |
Top ranked projects inSourceforge.net
| Project-Id | Name |
| 1 | SourceForge.net |
| 235 | Pidgin |
| 84122 | Azureus |
| 162271 | Openbravo ERP |
| 176962 | ADempiere ERP Business Suite |
| 196195 | PostBooks ERP |
Random selection was rejected because the intention of the current study is to rigorously analyse the projects to detect common pattern among them. Choosing a project from every year would be beneficial if the focus of studies was time-series analysis of the evolution of FOSS. Therefore, the appropriate method is to select those projects which meet certain identified criteria.
The requirement for such a study would be to select those projects who have been proved successful. In traditional software world, quantifying success is easy because there are metrics such as number of sales. But in FOSS ecology such a metric loses its significance. Therefore, it was decided to use a metric identified by the Sourceforge.net as the measure of success.
Sourceforge.net has the practice of ranking all the projects hosted in its sitebased on a fixed formula. This data is captured in the table STATS_GROUP_RANK_BYMONTH. The structure of this table is given in Table 1.
The projects which have remained in top position (Rank 1) for more than two consecutive months were selected as candidates for further analysis. The list of selected projects is given in Table 2