Differences between this run and the previous probe snapshot.
Spotted a result that's wrong — e.g. an op marked failed that actually works? Have an idea? Tell us; it goes straight to the maintainer.