On Mon, Nov 30, 2015 at 10:47:08PM +0000, admin@??? wrote:
> I propose a new setting (configurable per filter) with three values:
>
> fail_body_mime_type = ascii (default, current behavior)
> fail_body_mime_type = utf8
> fail_body_mime_type = latin1
Two comments:
* I agree with John Klensin that at this point the only sensible
choices are us-ascii and utf-8. Use of other charsets is to
be avoided:
https://tools.ietf.org/html/rfc6657#section-3
Thus, new subtypes of the "text" media type SHOULD NOT define a
default "charset" value. If there is a strong reason to do so
despite this advice, they SHOULD use the "UTF-8" [RFC3629] charset as
the default.
* The proposed parameter name is not well thought out. Don't
confuse MIME "type" (really Content-Type), with the "charset"
attribute.
One choice is to just always use UTF-8, so that any non-ASCII
content in the bounce body is required to be UTF-8 encoded. This
will work well when some day EAI addresses are included in the
error message. If the code simply replaces "us-ascii" with "utf-8",
then no configuration is required to use non-ascii strings, just
make sure they are UTF-8 encoded.
--
Viktor.
