How can feature usage be tracked across product variants? Implicit Feedback in Software Product Lines

TL;DR

This paper explores a novel method for tracking implicit feedback across software product line variants without embedding trackers into individual code, using a platform-centric approach tested with an extension to a variant management tool.

Contribution

It introduces a second transformation step for feedback collection in SPLs, avoiding direct code modifications, and demonstrates its feasibility with a practical extension to an existing tool.

Findings

Platform-level feedback is perceived as useful by practitioners.

The FEACKER extension integrates seamlessly with existing tools.

Stakeholders favor platform-centric feedback analysis.

Abstract

Implicit feedback is collecting information about software usage to understand how and when the software is used. This research tackles implicit feedback in Software Product Lines (SPLs). The need for platform-centric feedback makes SPL feedback depart from one-off-application feedback in both the artefact to be tracked (the platform vs the variant) as well as the tracking approach (indirect coding vs direct coding). Traditionally, product feedback is achieved by embedding `usage trackers' into the software's code. Yet, products are now members of the SPL portfolio, and hence, this approach conflicts with one of the main SPL tenants: reducing, if not eliminating, coding directly into the variant's code. Thus, we advocate for Product Derivation to be subject to a second transformation that precedes the construction of the variant based on the configuration model. This approach is tested through FEACKER, an extension to pure::variants. We resorted to a TAM evaluation on pure-systems GmbH employees(n=8). Observed divergences were next tackled through a focus group (n=3). The results reveal agreement in the interest in conducting feedback analysis at the platform level (perceived usefulness) while regarding FEACKER as a seamless