I think that was Patrik. Perhaps others too. Personally,
I avoid QP (quoted-printable) whenever possible. It seems to be best
suited to files that are "mostly text" or "mostly 7-bit clean".
If you want to send something like a GIF or an executable, or even a
spreadsheet, I would strongly recommend Base64 instead.
QP is not really useful, as the environments where I think it was
intended for, Europena languages with at large portion of the
letters in ASCII, it is still unreadable. We got a number of complaints
from our customers until people stopped sending out QP. It is
only quite occasional now that we QP messages here in my Danish
environment.
If you think Q-P is unusable, then use Base64, but never, ever use
8bit without using the 8BITMIME ESMTP extension.
Whatever you might think something never be used, it was, has been
and is being used by some.
What we should do is NOT to simply deny defact standards but to
provide a sensible (sensible to real users, NOT to ASCII-using MIME
designers) way of migration.
As long as we have
one server which is not 8-bit clean, then you can never assume that
you can just-send-8.
Sure. At the same time, as long as someone has 8-bit clean servers
on all the mail pathes, then he can assume that he can just-send-8.
So, I'm afraid your statement is completely meaningless.
In you environment, Keld, I know that you are using MNEMONICS according
to RFC-1345, but that is a different way of solving the problem. When
actually having 8-bit characters to send, I strongly argument
against a just-send-8 solution.
The important point is that real users did not choose 8BITMIME nor QP
but mnemonics.
That is, hypothetical virtue of backward compatibility of MIME is
denied by the practice.
Masataka Ohta