A Study of Feature Scattering in the Linux Kernel

Abstract

Feature code is often scattered across a software system. Scattering is not necessarily bad if used with care, as witnessed by systems with highly scattered features that evolved successfully. Feature scattering, often realized with a pre-processor, circumvents limitations of programming languages and software architectures. Unfortunately, little is known about the principles governing scattering in large and long-living software systems. We present a longitudinal study of feature scattering in the Linux kernel, complemented by a survey with 74, and interviews with nine Linux kernel developers. We analyzed almost eight years of the kernel's history, focusing on its largest subsystem: device drivers. We learned that the ratio of scattered features remained nearly constant and that most features were introduced without scattering. Yet, scattering easily crosses subsystem boundaries, and highly scattered outliers exist. Scattering often addresses a performance-maintenance tradeoff (alleviating complicated APIs), hardware design limitations, and avoids code duplication. While developers do not consciously enforce scattering limits, they actually improve the system design and refactor code, thereby mitigating pre-processor idiosyncrasies or reducing its use. !

REVaMP2 Project: Towards Round-Trip Engineering of Software Product Lines - Approach, Intermediate Results and Challenges

“…Of course, the project partners are also very active in the scientific community. With more than 50 publications, the project has had a significant impact on the state of the art, with notable publications at ASE [20] , Isola [21] , MODELS [22] as well as in IST [22], [23] , TSE [24] and many others. Members of the project have also organised the main scientific event of the product line engineering community, SPLC in Gothenburg in 2018, with well over 100 participants and workshops and tutorials geared directly towards the topics of the projects.…”

Seamless Variability Management with the Virtual Platform

“…To the left, the code is cloned and adapted (one line changed in the branch BC+SmallDisplay); to the right, a configuration option represents the change in a common codebase (integrated platform). The changes are usually more complex (e.g., features can be highly scattered [40], [41]), as well as the representation of variability in the integrated platform. We also need more variability concepts, among others, features [42], [43], [44], codelevel configuration [10], feature-to-asset traceability [14], [15], [45], a feature model (a hierarchical structure with features and their dependencies) [12], [13], a configurable build system [10], and a configurator tool [16], [46], [47].…”

On the relationship between design discussions and design quality: a case study of Apache projects

“…In total, we received 152 responses from 1, 012 email requests (15.01% response rate). Previous studies in Software Engineering field have have reported response rates between 5.7% [60] and 7.9% [49]. We discarded 22 partial responses, which left us with 130 responses (12.9% response rate after excluding the partial responses).…”