The decision to restrict our support of Byte Sequences to a single character encoding will allow us to avoid the additional complexity of either using heuristics to try to determine the intended character encoding, or else requiring that the encoding be specified with the Byte Sequence.
At least initially, we will only support sending keys and/values via response headers as either printable-ASCII-character Strings, or as Byte Sequences with UTF-8 as the intended character set.
The decision to restrict our support of Byte Sequences to a single character encoding will allow us to avoid the additional complexity of either using heuristics to try to determine the intended character encoding, or else requiring that the encoding be specified with the Byte Sequence.
We choose to support UTF-8 in particular because the standard preferences it for this purpose.