I did a logical analyze of what can cause this bug, wich leads to the
following conclusion.

- Pan does not expect some people will send raw 8 bit data in headers.

- Pan is copying the content of the "From" header to an internal UTF-8
buffer without testing if there is raw 8 bit inside it. This is done
when pan tries to create the "%s has written in message:" line in
front of the answer.

- As a result there is some invalid ISO-8859-1 data inside this UTF-8
buffer

- Later a conversion fonction is called to convert the date to
iso-8859-1 that stops at the first invalid character (this is the
behavious of iconv for example).

- This error is not detected, and Pan goes on sending as output most
of the buffer content as UTF-8 instead of the expected ISO-8859-1

This analyze is confirmed that raw UTF-8 data in headers will not
cause the bug, everything is correct in the output, and the name in
the "%s wrote" line is correct and is encoded in iso-8859-1.

Recommended solution for correction :
- test if the header is valid UTF-8 (iconv from/to UTF-8 is a stupid
but simple way to do it).
- if not, convert it from the encoding used for the body to UTF-8.
- if this fails again ?? Maybe remove all non-7 bit data ?

Jean-Marc:  thanks for that excellent analysis! 

Note that this only happens for %a (which doesn't convert to UTF-8),
not for %n (which does).

Fixed in CVS for 0.13.4:

http://cvs.gnome.org/bonsai/cvsview2.cgi?diff_mode=context&whitespace_mode=show&subdir=pan/pan&command=DIFF_FRAMESET&file=message-window.c&rev1=1.343&rev2=1.344&root=/cvs/gnome

The bug seems to persist in the latest beta version 0.13.3.91

e.g, an original post with some headers:

=================================================
[...]
Reply-To: "Mahaléo" <mahaleo@wanadoo.fr>
From: "Mahaléo" <quidam@pour_le_bot.org.antispam>
Newsgroups: fr.test
Subject: zzz test ignore
Date: Sat, 1 Feb 2003 01:18:01 +0100
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1106
Message-ID: <3e3b11e9$0$247$626a54ce@news.free.fr>
NNTP-Posting-Date: 01 Feb 2003 01:16:41 MET
MIME-Version: 1.0
Content-Type: text/plain

Etre ou ne pas être, c'est ça la question.
=================================================

And the follow-up message:

=================================================
[...]
From: "Mahaleo" <quidam@pour_le_bot.org.antispam>
Subject: Re: zzz test ignore
Date: Sat, 01 Feb 2003 01:26:27 +0100
User-Agent: Pan/0.13.3.91 (How did the starling get into the bar?)
Message-ID: <pan.2003.02.01.00.26.24.214810@mahaleowanadoofr>
Newsgroups: fr.test
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-15
Content-Transfer-Encoding: 8bit

Sat, 01 Feb 2003 01:18:01 +0100, dans
<3e3b11e9$0$247$626a54ce@news.free.fr>, Mahaléo a écrit:

[...]
> Reply-To: "Mahaléo" <mahaleo@wanadoo.fr>
> From: "Mahaléo" <quidam@pour_le_bot.org.antispam>
> Newsgroups: fr.test
> Subject: zzz test ignore
> Date: Sat, 1 Feb 2003 01:18:01 +0100
> X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1106
> Message-ID: <3e3b11e9$0$247$626a54ce@news.free.fr>
> [...]
> MIME-Version: 1.0
> Content-Type: text/plain
> 
> Etre ou ne pas être, c'est ça la question.

Bouh!
=================================================

Mahaleo

Punting for Christophe's return

/me scratches his head ...

Mahaleo : could you let me know the charset of the group you're
posting to (group properties) and the locale you're using ('locale'
from the command line) ?  

Could you also attach the full message Pan creates to this bugreport
(http://bugzilla.gnome.org/createattachment.cgi?id=104479) ?

Created attachment 14880 [details]
This is the full message Pan creates to the bugreport 104479

The charset of the group is: ISO-8859-1 (but the same bug appears 
with ISO-8859-15). 
 
My 'locale': 
 
[mahaleo@localhost mahaleo]$ locale 
LANG=fr_FR.UTF-8 
LC_CTYPE="fr_FR.UTF-8" 
LC_NUMERIC="fr_FR.UTF-8" 
LC_TIME="fr_FR.UTF-8" 
LC_COLLATE="fr_FR.UTF-8" 
LC_MONETARY="fr_FR.UTF-8" 
LC_MESSAGES="fr_FR.UTF-8" 
LC_PAPER="fr_FR.UTF-8" 
LC_NAME="fr_FR.UTF-8" 
LC_ADDRESS="fr_FR.UTF-8" 
LC_TELEPHONE="fr_FR.UTF-8" 
LC_MEASUREMENT="fr_FR.UTF-8" 
LC_IDENTIFICATION="fr_FR.UTF-8" 
LC_ALL= 
 

*** Bug 104459 has been marked as a duplicate of this bug. ***

Hmmm, I can only explain this if either your group's charset is UTF-8,
or if you're switching between profiles that have '%n' in the
attribution (definite bug there, brought out if your locale is in
UTF-8, which is the case for you).

Looking further ...

The bug does not appear if I change my 'locale' by editing
'/etc/sysconfig/i18n' and by changing line
LANG="fr_FR.UTF-8"
by
LANG="fr_FR@euro"

My 'locale' becomes:
[mahaleo@localhost mahaleo]$ locale
LANG=fr_FR@euro
LC_CTYPE="fr_FR@euro"
LC_NUMERIC="fr_FR@euro"
LC_TIME="fr_FR@euro"
LC_COLLATE="fr_FR@euro"
LC_MONETARY="fr_FR@euro"
LC_MESSAGES="fr_FR@euro"
LC_PAPER="fr_FR@euro"
LC_NAME="fr_FR@euro"
LC_ADDRESS="fr_FR@euro"
LC_TELEPHONE="fr_FR@euro"
LC_MEASUREMENT="fr_FR@euro"
LC_IDENTIFICATION="fr_FR@euro"
LC_ALL=

But I am not sure that it is a good thing to change Utf-8...

I have just installed 0.13.91 (on Red Hat 8, with 'locale':
LANG=fr_FR.UTF-8), with the same attribution which involved the bug
(%n a écrit), and the bug does not appear any more.

Thanks for that good job.

Marking as 'fixed' based on user feedback.

Thanks, Mahaleo!