Re: [exim] RFC on accepting mail to return path addresses

--On 22 March 2006 13:12:08 +0100 Sander Smeenk <ssmeenk@???>
wrote:

> Quoting Ian Eiloart (iane@???):
>
>> > What is the opinion of the masses, am I right to reject mail to return
>> > paths if they don't come from <> or postmaster?
>> I don't think you should.
>
> But don't you agree that a return path is only to be used for returning
> a message if it can't be delivered, hence 'return path'? I strongly feel
> the mailinglistsetups at apache.org and mysql.com are b0rked by calling
> out to return paths with a non-null-sender.

No, technically that's not true. They're also used for delay warnings,
vacation messages, and all sorts of other automatic responses. Of course,
all of those should use the null sender, too.

AFAICT, RFC2821 says that errors must be sent to the return path address,
and error messages must use a null return path. It doesn't say anything
else about whether you can send other messages to that address, or what
return path those messages should have. It certainly doesn't say anything
at all about callouts.


>> If you do want to, you could set a variable, then test for it in the
>> PREDATA ACL. If you reject at that point, it won't affect a callout.
>
> This is however a nice idea. I might work that out if i find time ;)
>
> Thanks,
> Sander.
> --
>| > > Isn't this a stupid question?
>| > Isn't this a stupid answer?
>| No, it was another stupid question. This is the stupid answer.
>| 1024D/08CEC94D - 34B3 3314 B146 E13C 70C8 9BDB D463 7E41 08CE C94D



--
Ian Eiloart
IT Services, University of Sussex