v6.2.0

Performance

• Skip the final random() call in fisherYatesShuffle — the last iteration always returns index 0, so one RNG call per shuffle is now elided (#858, #620 @usumerican)
• Bump pcg to 3.0.0, yielding roughly 1.4x faster seeded shuffles and drop a transitive dependency on Long (#850, #857)

Changed

• Migrated to ESM-first builds with a CJS compatibility build (#758)
• Update tsconfig base to @tsconfig/node24 (#836)
• Drop Ramda dependency (#835)
• Drop Jest and use native node testing (#761)
• Declared a minimum supported Node version in engines (#841, #756 @dynst)

Notes

The skip-final-call optimization does not change the shuffled output for a given seed, but it does reduce RNG state advancement by one step per shuffle. RNG operations will observe a different RNG state after each shuffle.

v6.1.1...v6.2.0