Kazu,
It's a SHOULD: you can be perfectly compliant and NOT implement it.
That's the point. A SHOULD is a strong recommendation to do something
that way, but it is not required to be implemented in order to be
compliant. Indeed, it's a directive to an implementor for how to
solve the problem in an interoperable way.
Just because it's not the right solution for a couple of environments
only means that those environments SHOULD ignore the SHOULD :)
-derek
PS: How does applying a CTE break interoperability? Any MIME-aware
client MUST implement CTE anyways, so it should decode the base64 well
before it gets up to the user.
Kazu Yamamoto (=?iso-2022-jp?B?GyRCOzNLXE9CSScbKEI=?=)
<kazu(_at_)iijlab(_dot_)net> writes:
I would like to request one thing to openpgp-mime-02.txt.
Section 3 says:
Additionally, if body parts to be signed contain trailing whitespace,
or lines beginning with the five characters "From ", implementations
SHOULD use either Quoted-Printable or Base64 to protect these body
parts against corruption by transport or delivery agents. Applying
this rule also ensures that trailing whitespace in the data encoded
cannot be modified without invalidating the signature.
I feel that "SHOULD" is too strong. I have two reasons.
(1) In the ISO-2022 environment (including Korean and Japanese), for
long historical reasons, applying CTE: breaks inter-operability
between MUAs. Even base64, which is considered more suitable to
ISO-2022-{JP,KR} than quoted-printable, should not be applied to
ISO-2022-{JP,KR} anyway. So, SHOULD is too strong for this kind
of environment.
(2) This rule implies that MUAs SHOULD search "From " pattern from
text in a deeply nested multipart. This is too tough for many
MUAs.
--Kazu
--
Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
Member, MIT Student Information Processing Board (SIPB)
URL: http://web.mit.edu/warlord/ PP-ASEL-IA N1NWH
warlord(_at_)MIT(_dot_)EDU PGP key available