* s. keeling [Mon, 22 Dec 2003 23:52:30 -0700]:
> With help from one of the list recipients, this is now verified and
> reproducible. Something between me and those people whose keys are
> determined by my copy of gpg to be "Bad signature", is mangling mail.
> Specifically, that something is fixing line breaks:
> -> gpg: There is no indication that the signature belongs to the ow-ner.
> +> gpg: There is no indication that the signature belongs to the owner.
More exactly, that something is removing "=\n" sequences from
Quoted-Printable encoded mails, so the diff would read:
> gpg: WARNING: This key is not certified with a trusted signature!
-> gpg: There is no indication that the signature belongs to the ow=
-ner.
+> gpg: There is no indication that the signature belongs to the owner.
> Mail sent to me containing lines less than (ca.) eighty chars results
> in "Good signature". Mail sent to me which includes lines longer than
> (ca.) eighty chars results in "Bad signature".
So, long lines remain long lines in the decoded message, but the Q-P
codification gets mangled and modified messages no longer have
72-or-less-chars lines (and the signature fails, as it gets applied to
the Q-P message rather than to the decoded message).
> Debian Woody 3.0r2
> Mutt/1.3.28i
> Exim 3.35 #1 (Debian)
> GNU Emacs 20.7.2
> GnuPG 1.0.6-4
> Procmail v3.22
At this point, I just can figure out it *might* be a broken MTA as
somebody else pointed out; but no more I can say...
All this Q-P mangling must sound familiar to someone out there. Any
hints?
Cheers.
--
Adeodato Simó (a.k.a. thibaut)
EM: asp16 [ykwim] alu.ua.es | IM: my_dato [jabber.org] | PK: DA6AE621
Truth is the most valuable thing we have, so let's economize it.
-- Mark Twain
Attachment:
signature.asc
Description: Digital signature