2) Transports/user-agents that cannot pass NUL are "broken".
They are, at least according to RFCs 821 (section 4.5.2), 822 (sections
3.3 and 4.1), and 1652 (section 3). RFC-1521 does not explicitly state
that NUL characters are permitted in 8bit body parts, but I don't see
how you can read it to state that they are forbidden.
You and Keith are crazy.
Do you think the character 'A' may be dropped, because you can't read
it to explicitely state that dropping code for 'A' (65) is forbidden?
It is explicitely stated that 0-255 is allowed, which is more than
enough for users to use them and expect them preserved.
There seems to be consensus that the specs need updating to reflect
reality.
No, and, moreover, the reality is 8BITMIME SMTP.
Otherwise, why 0-127 of RFC 822 is not amended?
Masataka Ohta