works just like the Unicode one. The C APIs match the ones in the Unicode
implementation, but were extended to be able to reuse the existing
Unicode codecs for string purposes too.

Conversions from string to Unicode and back are done using the
default encoding.