Research that depends on historical software often hits a wall: the source code is fixed, but the surrounding ecosystem is not. Base images, package versions, and other external resources drift over time, so old builds break in ways that have nothing to do with the code itself. The new arXiv paper introduces RELAY, a system designed to address this problem by scaling the restoration of historical builds.
RELAY's approach combines two ideas: reusing repair rules across similar build failures, and applying agentic repair to handle cases that rules cannot cover. The authors argue that this combination makes it feasible to restore many historical builds without manually patching each one. The paper's abstract is truncated, so specific results and evaluation details are not available in the source.
The work speaks to a practical bottleneck in vulnerability research and software engineering studies, where the ability to rebuild old versions is often a prerequisite for meaningful analysis. If RELAY delivers on its promise, it could make historical build restoration more automated and less of a manual chore.