Section 3, second paragraph, and elsewhere: "MUST NOT be longer than 72 bytes". "bytes" is an imprecise term. Do they really mean "8-bit ASCII characters", octets, or are 9-bit bytes as implemented on older hardware architectures also acceptable? Section 3.4 uses the term "characters" to describe a line length limitation. Consistency would be good.
I agree that consistency would be good. We do specify that that header-values MUST be encoded in UTF8. So changing to allow 72 characters could mean a significant growth in the number of bytes allowed on a line (*3 for Japanese text, for example.) 8-bit ASCII characters is close, except for those UTF-8 header values. What if we specified "MUST NOT be longer than 72 8-bit bytes?" Thanks, Joseph