Design

• No trait based dispatch (Arbitrary). Generators and shrinkers are straightforward coroutines instead
• If two tests need a different shape of the tested data type, just combine generators differently. No newtype wrappers necessary.
• No macros. Test cases are composed of function calls, ifs and loops
• Linear, top to bottom test code
• Only implement shrinker when it actually provides value, i.e. a large enough falsifier is found and it needs to be shrunk
• Panics are caught and treated as test failures subject to shrinking. An .unwrap() in tested code is also a violation of tested properties
• RNG seed value is printed on stderr. Use cargo test -- ... --nocapture to grab it
• RNG seed value is taken from FALSIFY_SEED environment variable if set. It's used for test failure reproduction
• Need a custom generator that produces only a subset of possible values of a given type? Just copy paste the generic generator code and specialize for your requirements. There's no "You wanted a banana but what you got was a gorilla holding the banana and the entire jungle."

Guidelines for implementing generators/shrinkers

• arb_*() coroutines should generate random values indefinitely and never return. Even if some data type is small enough to generate all values exhaustively. The coroutine may be used as a building block in testing a more complex data type. Terminating, then, would result in terminating the "big" generator prematurely.
• In contract to arb_*(), shrink_*() coroutines should finish once they can't shrink further. It allows the caller to attempt a different shrinking strategy if the falsifier is still unmanageably large.