Understanding Registration-Based Abstractions: A Quantitative User Study.

The adoption of programming language innovation is impeded because all program processing tools in the tool chain must support any new or altered language features. Registration-based abstractions (RBAs) were proposed to address this difficulty by allowing the editor to transiently superimpose new language abstractions on existing code. Individual programmers can choose where and when to see a new language abstraction, while at all times the underlying code remains written in the original language. Prior work demonstrated the feasibility of RBAs, but left important questions unanswered regarding how users would interact with such an approach. We asked 50 undergraduate students to answer basic program comprehension questions with and without RBAs. Our results show that participants can quickly and easily understand new abstractions without additional training, and suggest that this will extend to the general programmer community. The results also support a comparison across RBAs, and an initial discussion of specific features that facilitate or confound understanding.