<<< Date Index >>>     <<< Thread Index >>>

Re: Bad charset conversion with gpg-signed mime-forward.



* Tamotsu Takahashi [2005-02-19]:
> > Full report is here: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=295528
> 
> You said in this report you are using UTF-8 locale.
> This is important.

I don't know how this affects forwarded messages.

> > Summary: the charset declared in the header of the forwarded email are
> > changed from an original value of iso-8859-1 to utf-8 without actually
> > converting the message body. I don't know why this charset conversion
> > is necessary, but it does not seem to occur properly anyway.
> 
> For me, the forwarded message was declared to be euc-jp,
> because my $charset was "euc-jp".

From the manual, $charset is used to set the character set of your
controlling terminal. I don't see how it's relevant for the encoding
of forwared messages.

> Once I set both $send_charset and $file_charset to "iso-8859-1",
> it was forwarded properly, though it was not 8bit but QP.
> (Note: $file_charset feature is not in the CVS.)
> I set only $send_charset to "iso-8859-1" and failed.
> So, $file_charset should help you.

Indeed, it helps. But I don't understand why it's needed in the first
place. §3 of RFC3156 tells us 8bits chars should be QP- or
base64-encoded when sending multipart/signed, but why change the
charset too?

> > I could reproduce the bug with latest CVS, but now my &eacute; letter
> > is encoded in quoted-printable by the folowing sequence: "=EF=BF=BD"
> > instead of "=E9" previously.
> 
> "=EF=BF=BD" is UTF-8 character to show that it failed to be converted.
> (Try "grep '.357.277.275' *.c".) So something went wrong.

Thanks for the hint, I'll use that as a starting point for debugging
and see what's wrong.

Attachment: signature.asc
Description: Digital signature