antirez 用 GPT-6 Astra 检验 RedLock 安全性论证

antirez @antirez原推发布于 2026.10.06

在 X 上打开原推

GPT 6 Astra is able to provide interesting safety arguments of the original RedLock algorithm, given the assumption (that was always part of the specification) that clients finish work under the lock auto-expire time and clocks are imprecise but error-bounded. More work needed but interesting start.