Incorrect charset decoding in random articles...

[Daniel Pittman <daniel@rimspace.net>]

[-- Attachment #1: Type: text/plain, Size: 1148 bytes --]

A while ago on the list various people including myself encountered a
problem where Gnus would mis-detect the character set of various
messages.

Anything forwarded as a message/rfc822 attachment or inline seems to
trigger this, as do various messages at random.

The problem is that the article is decoded as some far east character
set, presenting as a mixed collection of ideograms[1] and '?'
characters.

Some but not all of the articles seemed to have broken[2] charset
specification, such as 'charset=us-ascii:iso-8859-1:utf8'

Others, though, had no MIME detail at all or they had a correct charset
specification.


I can't remember what the solution at the time was, though I believe a
code change was involved. Maybe Simon Josefsson did it? My memories are
hazy as it was over three months ago.


Anyway, I tried searching the list archives and my local one but didn't
find the details I was looking for.

I am running "21.5  (beta3) \"asparagus\" XEmacs Lucid"
on Linux inanna 2.4.15-pre5 #1 Sat Nov 17 01:38:23 EST 2001 i686 unknown


I can forward a selection of the mis-displayed messages in addition to
the attached one if you wish.


[-- Attachment #2: broken.mail.bz2 --]
[-- Type: application/octet-stream, Size: 3681 bytes --]

[-- Attachment #3: Type: text/plain, Size: 422 bytes --]


        Daniel

Footnotes: 
[1]  Is this the right term? My brain has seized up this morning...

[2]  At the very least, unfamiliar to me from reading the MIME specs.

-- 
It's important for us to explain to our nation that life is important. It's
not only life of babies, but it's life of children living in, you know, the
dark dungeons of the Internet.
        -- George W. Bush, Arlington Heights, Ill., Oct. 24, 2000