Re: Round 3: moving HTTP 1.0 to informational

>And with MIME specifically forbidding UCS-2, and HTTP/1.0
>silent about it, the result would be that it would effectively be
>forbidden, perhaps for good, and for no good technical reasons.  This
>is unacceptable, and the language should remain in HTTP/1.0.

While I agree that the language should not be sent out, I should note
that MIME only disallows UCS-2 for text/* types. If you send that data
as application/*, I think it's acceptable (most client will not
recognise it though...).
 
>The bottom line is that in current practice there is no default, and
>IMHO recommended practice should be the same: no default, charset
>compulsory.

Quite. It's a pity we have to standardise behaviour that is
incorrect.