René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"reach: \"a heartbeat named a busy mutant\" depends on wall-clock timing and is reached 1 to 2 times per shard","articleBody":"## Summary\n\nThe reach state \"a heartbeat named a busy mutant\" is entered only when a slow test crosses the 2 s heartbeat interval. Its count per shard depends on the machine's timing, not only on the constructed history.\n\n## Evidence\n\nFixed seed 20261105, main `35422e3`, Linux fixed-seed shards 1 to 3:\n\n- GitHub CI (run 43578388114): shard 1 = 2, shard 2 = 2, shard 3 = 2 (total 6).\n- The same commit on the Arca Arclet pool (tags `r61-main{1,2,3}-35422`): shard 1 = 1, shard 2 = 2, shard 3 = 2 (total 5).\n\nThe same Arclet pool gave head `346c8a8` shard 1 = 1 (tag `r61-fs1-346`), so the difference is between machines, not between main and the PR.\n\n## Why it matters\n\nUnder \"determinism under load\" (decisions register, 2026-11-13), a reach state must not depend on wall-clock timing. A state reached only by a slow test crossing a 2 s beat is that kind of dependency. The state still has to be reached on every shard that declares it. It is, but only by timing.\n\n## Proposed fix\n\nGate the state with a constructed history that holds a mutant for a deterministic time at a named test-only pause point (the same pattern as PAUSE_AT in the daemon witness), so the heartbeat fires regardless of machine speed. Not started here.\n\n## Related\n\n#61 (fix(oracle-gaps): retire a batch's copies from Vitest's state). #80, #81 and #82 are separate.","author":{"url":"https://github.com/KyleAMathews","@type":"Person","name":"KyleAMathews"},"datePublished":"2026-11-14T22:17:12Z","url":"https://github.com/TanStack/oracles/issues/84"}