[Reportbug-maint] Bug#496781: Bug#496781: reportbug: Creates bad signatures when signing utf8 text
morph at debian.org
Sat Apr 18 22:53:36 UTC 2009
On Tue, Apr 14, 2009 at 13:59, Bjørn Mork <bjorn at mork.no> wrote:
> Sandro Tosi <morph at debian.org> writes:
>> Are you still facing this problem? even with the latest 4.0 version?
>> If you go to http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=496781
>> you can see that your message is correctly identified as signed, and
>> it's claimed a valid. What other input you received the mail is sent
>> out as not correctly signed?
> I still face the same problem.
> But looking closer at it, I'm becoming unsure whether the bug is really
> in sendmail (or my sendmail config) or in Gnus (my MUA).
> Yes the signature can be verified as it appears on the BTS web page or
> in any MUA which decode the Quoted-Printable encoding before verifying
> the signature.
> But if you look at the raw message:
> you'll note the header lines
> Content-Transfer-Encoding: quoted-printable
> X-MIME-Autoconverted: from 8bit to quoted-printable by canardo.mork.no id m7RBkqIR015319
> This is my local sendmail converting the original 8bit message,
> including signature, to quoted-printable. I know this conversion can be
> questioned, but such conversions will happen somewhere on the Internet
> whether I change my configuration or not...
> I'm more uncertain about Gnus as a MUA in this case. When I'm reading
> such a message in Gnus, it will first verify the signature (i.e. fail)
> and then decode the content. It should probably have done it the other
> way round since there is no way a PGP signature could be embedded in a
> valid QP-encoded message (as the signature itself will contain the
> character =)
In both the proposed situations, it doesn't seem a bug in reportbug
itself, but in the mua/mta used.
> But the problem would also disappear if reportbug used OpenPGP/MIME
I don't think it's reportbug to set the mime type, but either mua or
mta does it.
> Feel free to downgrade this bug to wishlist, or close it
> with wontfix if you find that most appropriate.
Well, I'd rather come to an agreement with out about what to do. From
the analysis you did, it seems this bug is in the external programs
used by reportbug, and not in reportbug itself, so I'm kind of
thinking about closing it, what do you think?
Sandro Tosi (aka morph, morpheus, matrixhasu)
My website: http://matrixhasu.altervista.org/
Me at Debian: http://wiki.debian.org/SandroTosi
More information about the Reportbug-maint