The primary reason ascii armor was chosen was for the case of encrypted
messages. The way RFC2015 is currently worded, a non-MIME mail user can
simply pipe the whole message to pgp and everything will work just fine
without the need for any additional software. It also makes even
MIME-compliant software easier to write.
I have to agree here, so long as there is a *large* userbase of 2.6.x
users Ascii Armor is far from outdated. I see very little advantage here
of switching from Ascii Armor to Base64 other than adding one more PITA
for backward compatibility.
The advantage is that using standard MIME is more easily specified and
implemented than using ASCII armor. That is, a developer who has a MIME
toolkit must modify that toolkit in order to use ASCII armor. Forcing ASCII
armor when it shows no noticable value over standard MIME is one more
impediment to new developers adopting OpenPGP.
Another is that the spec does not fully specify how to do ASCII armor,
specifically how to create the checksum. Not only does the draft not
specify how to do this, it references a book that is "out of stock
indefinitely" according to the publisher.
I strongly favor backward compatibility with the installed base, but not at
the expense of losing developers over unneeded features. Since a
description of ASCII armor has already been published (well, minus how to
do th checksums) in RFC 1991, any developer who cares about backward
compatibilty can become backward compatible.
--Paul Hoffman, Director
--Internet Mail Consortium