I, too, contend that uuencode is the correct solution. The install
base is enormous. Uuencode is one of the most widely implemented
encoding schemes around. Granted there are problems with trailing
blanks being dropped, however this has been addressed in recent
years. I believe the version shipped with 4.3BSD (and later) systems
avoids the trailing blank problem. If we were to specify that specific
encoding, we can work around the problem.
I don't think that the translation problems (ASCII/EBCDIC) should take
away from uuencode (or any other encoding). The character translation
problems applies equally to "plain old text" in the message body. I
really think this is a transport layer issue.
--lyndon