IMO, the vacation draft needs to say that the vacation response
generation mechanism MUST take charsets and content-types into account
even if the :mime parameter is not given.
Implementations SHOULD choose a fitting charset and recode the text
given in the script (in UTF-8) into the chosen charset and apply a
suitable CTE.
I have no problem with suggesting this, but I don't think it rises to the level
of a SHOULD. I think we need to stop short of any sort of requirement (and
SHOULD is a form of requirement) that stuff be recoded from UTF-8 into other
charsets. Sending UTF-8 without any hint of recoding needs to be an acceptable
option.
At the very minimum, implementations should add
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
MIME-version: 1.0
to the header of the generated message.
Reminding people that appropriate MIME labelling MUST be used for
whatever charset is generated is a reasonable thing to do, of course.
This might sound obvious, but we just encountered a vacation response
message that had no MIME headers whatsoever, but contained 8bit-encoded
UTF-8 text.
Sure. But we already have rules that say such messages are illegal,
regardless of whether or not they are a vacation response. It is far from
clear that anything we say will prevent this.
Ned