ietf-openpgp
[Top] [All Lists]

Re: some requests

2001-01-23 20:15:17
Derek,

Just in case you misunderstand my opination. I'm not objecting to 7bit
restriction of Multipart/Security at this time. (Yes, I don't like it
but the point is not that.) 

I'm asking not to recommend (ie SHOULD) MIME encoding for "7bit"
character set. ISO-2022-{JP,KR} is 7bit (not 8bit) and almost all user
use it as raw (i.e. without base64) in daily life even with MIME UAs.

From: Derek Atkins <warlord(_at_)MIT(_dot_)EDU>
Subject: Re: some requests

There are still many non-MIME UA users. 

Yea, but we're defining a MIME type.  We can't be defining MIME types
for people who don't use MIME.  For example, the sender doesn't know
that the recipient does not understand MIME, and if they did they
shouldn't be sending MIME in the first place.  Sure, we can TRY to be
as compatible with non-MIME-mailers as we can, but I don't believe
that we should bend over backwards just to make it easier for them,
when quite frankly those non-MIME users wont be able to read many MIME
messages in the first place.

Multipart/Security is designed to maintain both to non-MIME UAs and
MIME-aware-but-Multipart/Security-unaware UAs. Both users can at least
read the first. Yes, it requires 7bit restriction. 

But why should we use base64 to 7bit character set?

BTW, do you know any region where base64 is actually used for
text/plain? If we use base64, readability becomes drastically poor.

Yes.  I've been receiving messages from someone in, I believe, Korea
(I don't have the mail in front of me, but I seem to recall it was a
.kr address) who has been sending me English text/plain using base-64
encoding.

I guess this is a bug. Not represents typical UAs in Korea.

For example, Chinese people uses GB-2321 character set, which is 8bit
and base64 is suitable if required. However, they tend to send GB-2321
text with CTE: 8bit (without base64) for readability. (Most SMTP
channels in Asia are 8bit clean.)

I'm currently of the mind that we should not require base-64 encoding,
but we should encourage base64 encoding.  Basically if you, as a
sender, know that you have an 8-bit clear channel to your recipient
you should be allowed to send an 8-bit message.  However, if you are
not sure you have an 8-bit channel then you should base-64 encode your
message.  I just wish there were an easy way to determine ahead of
time if there is an 8-bit clean channel.

Maybe my explanation is wrong. GB-2321 is an example that how base64
is unfriendly to the real world in Asia. Of course, base64 is required
for GB-2321 when a signature is calculated.

Encourage base64 encoding for any and all character sets seems to me a
bad idea. A right way is to encourage people to deply 8bit clean SMTP
channels.

--Kazu

<Prev in Thread] Current Thread [Next in Thread>