09-20-07 06:13 AM
Thomas J. Boschloo wrote:
<much snippage>
[vbcol=seagreen]
> The problem here is however that it is impossible to write Italian with
> just the 7-bit ASCII character set. And it is impossible to encode
> characters in plain text signature messages.
>=20
> That some form of mapping takes place seems inevitable and I suspect
> that it is this what is breaking the signatures.
>=20
> I looked up the latest OpenPGP draft this evening and it seems to
> suggest OpenPGP implementations to always assume UTF-8, but they are
> allowed to use other encoding schemes when needed. A clear case of YMMV
> I would say :-)
>=20
> =E2=82=AC
> Thomas
First of all, your sig verified fine even with the "high ASCII"
character in it. UTF-8 encoding to the rescue. 
You have to remember that we're talking about *clear signed* text here,
and the OpenPGP spec also deals with MIME messages. What works in a
MIME/Multipart message doesn't always work in a flat-file pure text
message. The two are different things for the purposes at hand.
Actually, the OpenPGP spec is a good read if you want to understand why
some people are having problems with unarmored signatures on text
containing nonstandard characters. It gives mention of the 8-bit versus
7-bit problem, and some insight into why it's necessary to stick to
US-ASCII or UTF-8 in this situation.
That's RFC 2440 - http://tools.ietf.org/html/rfc2440
[ Post a follow-up to this message ]
| 
