Knowledge evolution in autonomic software product lines
Proceedings of the 15th International Software Product Line Conference, Volume 2
| Hi-index | 0.01 |
In a recent study of a product line, we found that the defect reports both (1) captured new requirements information and (2) implicated undocumented, tacit requirements information in the occurrence of the defects. We report four types of requirements knowledge revealed by software defect reports from integration and system testing for two products in this highdependability product line. We argue that storeandretrievebased requirements management is insufficient to avoid recurrence of these types of defects on upcoming members of the product line. We then propose the use of two mechanisms not traditionally associated with requirements management, one formal and one informal, to improve communication of these types of requirements knowledge to developers of future products in the product line. We show how the two proposed mechanisms, namely feature models extended with assumption specifications (formal) and structured anecdotes of paradigmatic productline defects (informal), can together improve propagation of the requirements knowledge exposed by these defects to future products in the product line.