NSF Award
2216594
The broader impact/commercial potential of this I-Corps project is the development of an intelligent software maintenance technology that focuses on improvement of the software and systems that underpin our national infrastructure and drive our economy. It is estimated corporations are spending over $300 billion per year globally, paying down “technical debt” on addressing issues related to maintaining legacy systems or dealing with bad software. By adopting the proposed technology, software development teams will avoid critical bugs that, in turn, will avoid financial loss. The proposed technology will help chief technology officers (CTOs) and chief information officers (CIOs) - not only developers - carry out several critical activities including: gaining actionable insight into the state of their code; accurately estimating the scope and cost of code cleanup; and assessing code quality trends and how to repair these code quality issues. By carrying out these activities, the proposed technology will decrease the time, cost, and risk of code cleanup while improving team performance.<br/><br/>This I-Corps project is based on the development of an intelligent refactoring bot that may be easily integrated into any project repository. The bot may be customized to monitor the quality in the repository after a number of pull-requests. The bot analyzes the changed files to identify refactoring opportunities using a set of quality attributes. It will then find the best sequence of refactorings to fix the quality issues, if any. The developer is able to review the recommendations in the generated pull-request. The initial proof-of-concept of the proposed interactive refactoring software was tested on 12 large-scale, open-source, and industrial systems with a total of 37 developers from industry. The technology reduced the amount of time that developers spent to understand existing large-scale source code by an average of 47%. Existing refactoring products, when tested on the same systems, were able to reduce that time by less than 24% on average. In addition, the technology was able to detect and fix an average of 92% of existing design defects in these systems. In contrast, existing refactoring products were able to only detect an average of less than 76% of these defects.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
