[Top] [All Lists]

Re: [ietf-smtp] Survival of arbitrary MIME Content-foo header field?

2018-07-30 12:17:53
On 7/30/2018 10:04 AM, Steve Atkins wrote:

On Jul 30, 2018, at 9:53 AM, Dave Crocker <dhc(_at_)dcrocker(_dot_)net> wrote:
I'd expect it was very likely to survive, particularly if the MIME was 
well-formed, as intermediate hosts are unlikely to randomly crack open a MIME 
body and rewrite it. That DKIM works remarkably well is a data point there.


The ones I've noticed that rewrite MIME have all been Exchange servers, so 
someone with more Exchange clue might be able to give some insight there.

A few hosts add additional MIME content (advertising, footers), but the ones 
I've paid any attention to seem to do so by fairly crude textual insertion, so 
probably wouldn't corrupt existing content too much.

To the extent that the unknown MIME header field depends on the associated content to be unmodified, that's an obvious problem.

However my impression is that such additions by mediators, such as mailing lists or anti-abuse enterprise front-ends, are done to the primary body-part -- the cover note -- and not to any attachments.

So a MIME content-foo for a subordinate body-part ought to survive... mumble.

What if it is at the message header level?  Any better/worse?

That I'm less sure about. I'd expect it to be at best the same or more likely 
worse, as lots of intermediate hosts do look at and modify the headers.

I can't see a reason it should be better and for some reason I'd expect it to be somewhat (or very) worse.


Dave Crocker
Brandenburg InternetWorking

ietf-smtp mailing list