The use of test cases as "API documentation" that doesn't show how to use the module for common bog-standard use cases, and instead brings in other unrelated CSV modules to conflate the example is particularly *chef kiss*
Why demonstrate the CSV parser parsing a string literal or filesystem stream when you can generate the CSV using your fancy (otherwise useless) CSV generator and hook it up in a way that's incompatible for the common use case?
Why demonstrate the CSV parser parsing a string literal or filesystem stream when you can generate the CSV using your fancy (otherwise useless) CSV generator and hook it up in a way that's incompatible for the common use case?