A new framework called ReLay has been developed to address the challenge of rebuilding historical software builds, which often fail due to environmental decay. The study found that a significant majority of historical builds could not be reconstructed. ReLay employs a hybrid approach, using a large language model (LLM) to identify and validate repair rules from common failure symptoms, which are then reused without further model calls. Cases that cannot be resolved by these rules are escalated to a per-case agent. This combined strategy successfully restores more builds at a lower cost than using agents alone. AI
IMPACT This research could improve the reliability of software engineering benchmarks and historical software analysis by addressing build failures.
RANK_REASON Academic paper detailing a new framework and its evaluation. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
