Crucible is continuous integration for chess engines. It watches your engine's git history, builds every commit, plays it against its predecessor under the Sequential Probability Ratio Test, and shows you an Elo timeline so you can see which changes made the engine stronger or weaker.
Author: Sullivan Bognar
Crucible 0.1.2 what's new?
Added
Go 1.26.2 to the default Docker engine toolchain image.
Optional fixed-game canonical progression matches via testing.progression_games, while experimental and manual patch tests continue to use SPRT.
Changed
Release gates now always play their full, color-balanced configured game count instead of stopping early on an SPRT boundary.
Fixed
Preserve the daemon environment when launching engine build commands, preventing Docker builds from losing /usr/local/cargo/bin and failing with cargo: not found.
Fail a job immediately when a game cannot be launched instead of repeating the same infrastructure error for the entire game budget.
Comments
Post a Comment