Re: [Exim] acl_not_smtp

Top Page
Delete this message
Reply to this message
Author: Bruce Richardson
Date:  
To: exim-users
Subject: Re: [Exim] acl_not_smtp
On Thu, May 13, 2004 at 09:36:02PM +0100, Jeremy Harris wrote:
> Bruce Richardson wrote:
> >>OK, wishlist. The current situation seems to leave me with great
> >>difficulty in validating recipients,
> >
> >
> >Are you trying to validate local recipients or others?
>
> Others. Well, all actually.

[snip]
> >Are you planning to reject the entire message if one recipient is wrong?
>
> Yes.
>


And what will you do if there is a delay or timeout validating remote
recipients? Rejecting the message would be crazy: if all recipients
that can be verified are valid, the message really should be accepted
and an NDR issued later for any recipients that do turn out to be
invalid. The alternatives are a) to wait however long it takes to
validate remote recipients - which could be hours or days - or b) to
reject the message even though all the recipients may be valid. Neither
option is at all good.

> >You should be able to do what you want by with the $recipients variable.
> >For example, you could use ${perl{}} or ${run{}} to pass $recipients to
> >a perl function or external script to validate them.
>
> Yes. Ugly, when the basic capabilities are already there in Exim.
> I'd like to have access to them, is all.


What you are attempting is ugly and IMO unwise. Nobody has the right to
stop you but nobody has a duty to expend limited time on adding an
unwise and little-wanted feature. I'd be surprised if I am alone in
thinking that and therefore surprised if it makes the wishlist.

The simplest solution to your problem would be to write a
sendmail-replacement script that reads standard input and tries to
deliver it via SMTP on the loopback interface (this would be quite a
short Perl script). Have your MUA call that.


--
Bruce

Hummingbirds are the only birds that can fly backwards, apart from
ostriches if you punch them hard enough.