still missing tests

Yeah, throw an exception in that case? Currently it just ignores the upper bits. E.g. a charcode of 256 would be 0 and 257 = 1.

I mostly added the start/end for API consistency, but if there's a consensus, I can also remove it.

This conflicts with an existing buffer set method: See https://github.com/ry/node/blob/master/lib/buffer.js#L191

Thanks elarkin, I changed it to "fill" now as xk suggested in #472

thanks. sorry it took so long