Re: [exim-dev] [PATCH] Implement redirect

On Thu, Jun 29, 2006 at 06:35:27PM +0200, David Saez Padros wrote:
> Hi !!
>
> >> How many MTAs accept a 551 redirect when they are sending mail on to
> >> another MTA?
> >> Of those that don't, what happens? Dropped mail, bounced?
> >
> > Exim bounces. It does not distinguish between 5xx responses.
>
> If no MTA supports it and giving 551 codes will end in lossing mail
> which is the purpose of implementing it ?

Not losing, if it's not supported it'll just bounce. The purpose is to tell
the sender MTA something equivalent to: "Please send this mail to foo@???
instead. If you can't do that, then bounce it to the sender address, so that
she can re-send it to foo@??? manualy".

I don't pretend that this is suitable for everyone. Just that, for some people,
it can be very useful.

Besides, in the long term, this should get better. After all, 551 is a standard
response defined in RFC 2821. I don't see why any MTA wouldn't want to support
it in the sender side.

> If exim implements 551
> codes it must implement both parts (client/server) and must have a
> way to recognize if remote client supports it or not (via some
> propietary EHLO extension)

I don't think either of these is essential. Bouncing mail because user is
actualy somewhere else is not worse than, say, bouncing it because there's no
such user. In the first case, at least, the sender will know what to do.

That said, I have interest in implementing the sender part for Exim.

> and what about using RFC 1035 experimental 'MR' (Mail Rename) 'MB'
> (Mailbox) and 'MG' (Mail Group) DNS records ?

That seems to serve a different purpose (affecting mail sent by some senders,
rather than mail sent by any of them).

--
Robert Millan