Re: bug in testing of message id's

On Tue, Apr 20, 2004 at 02:58:40PM +0300, Patrik R?dman wrote:
> Good catch. I've applied a shorter patch which should fix this, see below.
> 
> (As you may notice from the patch format, I'm trying out Subversion. :) )

I wouldn't have noticed.. I've been playing with gnuarch, that's still
puzzling my brain though.

> -            if (p == buf || *p == p[-1] || !p[1])
> +            if (p == buf || p[-1] == '.' || !p[1])

Without applying the patch, or reading it in context, this keeps the
order of the '-' and '.' the same, so [email protected] would fail? I looked
at the NNTP-rfc and they're very, very liberal in the message id's.
Okay, they just point at other RFC's, but almost anything seems to be
allright. 

I haven't looked at how what the id's are used for in the code, but
wouldn't a <anything>@<anything> be enough of a test on incoming
messages? The whole function seems overly strict, with a special test
for @[num.num.num.num].. If the message-id arrives, a lot of other
newsservers have already considered the message-id to be acceptable.

Chris

-- 
    I've been down so long, if I'd cheer up, I'd still be depressed.
            - Lisa Simpson, Moanin' Lisa Blues.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFAhRmBgQ3S2XWlk0sRAhksAJ9Pat0a8sxgF5XdlVqT94EtyqZD8QCgiZA2
mZoGvIxLd5wVsGidHT9hT80=
=HQ8M
-----END PGP SIGNATURE-----