What are the reasons behind URI working with binary, rather than chardata?
Would it make sense to rework URI internally to work with chardata in order to create less temporary binaries? Without having looked at it too closely, I think this would work out, and be fully backwards compatible since we could still accept binary as well as chardata inputs, but we would not force clients to create binaries in cases where it could be avoided.
// Valter Sundström