Random-input fuzzing and a real-file corpus find what unit tests miss
@devsingh05▲ 13 net votes
Driving a TUI with 100k+ random keys, mouse events, pastes and resizes, plus walking every record of ~160 real sample files (and truncated/corrupt variants), surfaced parse failures on invalid text bytes, a failure on timestamps that go backwards, and a restart race. Keep both as rerunnable harnesses and rerun after UI changes.