From: Erland Sommarskog (sommar@algonet.se)
Date: Sat Jan 11 2003 - 10:48:24 CST
=?ISO-8859-1?Q?Claus_F=E4rber?= (list-ietf-wg-apps-usefor@faerber.muc.de) wrote:
> Russ Allbery <rra@stanford.edu> schrieb/wrote:
> > It's really not particularly hard to keep track of what strings were
> > originally encoded and to see if they were actually changed by the user
> > before re-encoding them instead of just using the original strings.
>
> Further, not all user agents do present the full message, including
> headers, to the user for editing.
> Most non-textmode user agents have seperate input methods for that data.
My reaction to Charles's statement was the same as yours.
However, the real world is not like that. Xnews, the newsreader I use,
decodes the RFC2047 if it can (it can only handle iso-8859-1), and then
spits out 8-bit. MACsoup RFC2047-encodes an 8-bit subject, despite it
has not the faintest idea of what charset the header really is in.
(Well, of course it makes a reasonable assumption, but the fact that
that reasonable assumption works in practice, also shows that it works
in practice.)
Had authors of newsreaders understood that it is important not to
change the subject line, except when the user explicitly and kept
existing RFC2047 or 8-bit, then RFC2047 would have had as bad name
as it has in the Usenet community.
-- Erland Sommarskog, Stockholm, sommar@algonet.se