🚀 go-pugleaf

In article <AF6AEC139668972BF@bones.demon.co.uk>, Alan Staniforth
<alan@bones.demon.co.uk> writes

>If I were to scan the body DATA for possible bounce targets during recipt 
>of a message and get a hit, given that 554 is given as an allowable 
>failure response to DATA, I suppose the correct way to bounce would be:
>
>554 Transaction failed
>
>sent after receipt of the <CR><LF>.<CR><LF> terminating the DATA.

our experience is that  554 5.7.1 works just fine (see RFC2034 for the
extended codes)

>If this is an acceptable approach,

it is a permanent error, and a valid response to DATA, so it should not
cause any problems with retries

> and if I were to bounce a message at 
>that point, should the MTA handing over the mail to my server return a 
>bounce notification to the sender? (I realise *should* and *will* are two 
>very different entities in this case, but I am curious.)

the client should indeed generate a "bounce" message.

The only downside of the approach is that it obliges you to receive the
entire body before you arrange for it to be rejected.

This of course does allow you to scan a great deal more for stop words
and other giveaway features of mail you do not wish to receive. As it
happens, in the circumstances that we generate a 554, we feel that
having the server download the whole message before generating the
rejection is a positive benefit.

-- 
richard                      richard.clayton    @    T U R N P I K E .com
                                                     tel: +44 1306 732300
"Assembly of Japanese bicycle require great peace of mind" quoted in ZAMM