Niklas Hambüchen writes:

 > I would therefore vote to expose internals of *all* data types exposed.
 > Not doing so is assuming the ability to foretell everything a user will
 > ever do with it, which is never true.

I agree. I had plenty of trouble because libraries hid types and/or
functions that I needed access to, but I never had any trouble because
a library exposed a type that I didn't need access to.