Re: Why SRS really sucks

On Sun, 2006-04-02 at 20:27 +0200, Stephan Menzel wrote:
> If I only do SRS when the sending domain is SPF protected, what would happen 
> to te other protection forms similar to SPF like Sender ID?
> Those protection schemes also benefit from SRS and I would break them if I 
> only perform SRS when a 'real' SPF record is there.
> 
> So the way I see it, SRS should be done in those cases too, shouldn't it?

That depends. There are _always_ going to be people out there who will
make up random new rules about email, diverging from the known and
accepted norms and deciding unilaterally to reject the mail you send.
But mangling the recipient address unnecessarily would be bad -- there's
useful information in there which you'd be obscuring, so it's best to do
it only when you _need_ to.

My approach is to have a list of recipient domains for which SRS will be
done. I add recipient domains to the list only if they meet all three
criteria below:

 1. They reject mail for spurious non-standard reasons, such as SPF
    failure or their own random inventions (like gmx.de).

 2. They fail to fix this error when I explain it to them.

 3. I still actually care about getting mail through to those particular
    domains, and their users don't just migrate elsewhere after #2.

There aren't actually many domains in my list -- most people don't get
to #2.

In fact, my list of broken domains contains a 'reason' for each domain
-- so some domains have only their own addresses rewritten, some have
all addresses with SPF records rewritten, etc.

-- 
dwmw2