From: Ian Bell (ianbell@turnpike.com)
Date: Thu Sep 26 2002 - 07:43:53 CDT
On Wed, 25 Sep 2002, Charles Lindsey <chl@clw.cs.man.ac.uk> wrote:
>In <VyukqmClgZi9IAk4@pillar.turnpike.com> Ian Bell
><ianbell@turnpike.com> writes:
>>ESMTP servers may encounter non-8BITMIME servers on the next hop. But
>>that's OK because they know how to convert 8bit MIME parts into 7bit
>>MIME parts: leave headers alone as they're 7bit and encode leaf parts
>>with qp or base64.
>
>Indeed so, but hardly any servers do that conversion, even though they
>claim to be 8BITMIME compliant.
How do you know that? You say that it is exceedingly rare to encounter a
server that doesn't offer 8BITMIME, so how have you tested
down-conversion?
Or are you confusing "8bit clean" with 8BITMIME support? The former may
just not bother testing for 8bit content before relaying or delivery:
the latter should be applying the rules.
>But the whole point is that it is exceedinly rare to encounter a server
>that does not support 8BITMIME in some form.
How have you determined that?
>Hence my claim that,
>practically speaking, 8bit stuff in the headers of multiparts will get
>through unscathed so often that it is hardly worth the trouble for a
>gateway to worry about it
If you happen to be at the end of an RFC-compliant chain of MTAs that do
not transfer 8bit content cleanly, such messages will get through
unscathed precisely 0% of the time. Your view of whether conforming to
the MIME RFCs is "worth the trouble" may be coloured as to where you
are.
-- Ian Bell T U R N P I K E