On Tue, 2 Dec 2003 13:49:54 +0000 (GMT), Philip Hazel
<ph10@???> wrote:
>On Tue, 2 Dec 2003, Marc Haber wrote:
>> >Well, that's true, I guess. If you want to allow for that I suppose you
>> >need to have some way of remembering whether scanning was completed. For
>> >example, in some database.
>>
>> Yes, but that's bad since the filter can't know whether exim was
>> successfully able to complete final delivery. IMO, this bookkeeping
>> should be done by the MTA.
>
>I think I am now mis-understanding what you are saying, because this no
>longer makes sense to me. If Exim completes the final delivery, the
>message will be deleted and the filter will never run again. What have I
>not understood here?
With the scenario you have outlined, the scanner would have to keep a
database record for a message around that will never be presented to
it again since final delivery was completed. One would need to use a
time-based database cleanup procedure which I consider very ugly.
>> Additionally, the system filter does not have access to the entire
>> message.
>
>No, but a "C system filter" with a different API (the same API as
>local_scan(), as you suggested) would have. I'm not saying that the
>system filter does what you want, just that what you want is in effect
>just a slightly different kind of system filter, which might be easier
>to implement at the start of delivery instead of at the end of
>reception.
OK, I understand now. I thought your message basically was "use the
system filter".
Maybe an ACL to be evaluated after the reception was completed (even
after the acl_data) would be a good idea too.
Greetings
Marc
--
-------------------------------------- !! No courtesy copies, please !! -----
Marc Haber | " Questions are the | Mailadresse im Header
Karlsruhe, Germany | Beginning of Wisdom " | Fon: *49 721 966 32 15
Nordisch by Nature | Lt. Worf, TNG "Rightful Heir" | Fax: *49 721 966 31 29
