At Wed, 3 Dec 2003 17:09:35 +1100,
Bob Smart wrote:
>
> > This should be nearly the first router (for me it comes after system
> > aliases, and an "eximfilter" router I use to sort out mailing
> > lists before they reach tmda)
> >
[snip]
>
> If this all worked I guess it should go in the FAQ. But I can't
> see how it can work. We want tmda DELIVERY to just return as in
> qmail and have exim look at other exim routers below tmdaprocess
> when tmda doesn't want to impede the progress of the message.
> However for this to work tmdaprocess's driver (accept) would
> have to return PASS based on the status of the tmda-filter.
> I don't think it ever does that. The proposal is OK if tmda
> does its own final delivery. However that means messages to users
> with a ~/.tmda/config are never subject to .forward or aliases.
> Of course we can put tmdaprocess last, but then you lose tmda when
> you forward your mail elsewhere. Can we improve the condition to
> see whether the mail has already been through tmda?
>
> Or maybe I missing something?
No, your summary is pretty accurate. Basically this does not buy you
much over a pipe in a .forward file, except that many sites do not
permit pipes in .forwards, and it is one less user configuration item
to screw up.
I guess it never occured to me that there would be messages that TMDA
would not "want to impede the progress of". So yes, TMDA is doing the
final delivery. That is after all what the DA stands for :-)
What you are proposing makes sense from the exim point of view of
course, but it does mean (I think) changing the model of TMDA
filtering, since if you look at the options in the filter file, none
of them involve passing the message on to someone else to deliver.
I suppose something like
DELIVERY="|/bin/true"
might work to skip doing the delivery and just return 0.
So now I think I finally understand why Kirill Miazine built a perl
daemon :-)
db
_____________________________________________
tmda-users mailing list ([EMAIL PROTECTED])
http://tmda.net/lists/listinfo/tmda-users
