On Mon, 16 May 2011, W B Hacker wrote:
> Chr. von Stuckrad wrote:
> I suspect that what you have left still to do just might be too unusual
> to be a problem Exim expects to have to deal with.
>
> At least not with just a flag or three..
Exactly what I feared to hear :-)
> You could *perhaps*:
> ...[filtermethod]...
> COULD be internal - embedded perl for one...
OK, could try that, but only if I can use perl inside
or smtp-rewrite right-hand-sides (never checked that yet).
...
>> May be I have no chance anyway to find these cases?
>> May be work with a combiation of 'condition' and 'redirect'
>> and transport to another MTA?
>
> Or simply reject.
>
> svn+python mailscript with kerberos have no sacred ignore-RFC permit.
OK, but those are used/added by 'users', so I'd rather not
school every newcommer in the necessary patches/fixes which
do not confrorm to their standard manuals.
> - their job, not the mailadmin's.
... :-)
> Routing to some other MTA doesn't seem to be a solution, as Exim is far
> from the only MTA that would balk at such formats.
No, misunderstanding? ALL mail ALWAYS must go to the Smarthost!
(Question was 'only', how to fix it first)
> EVEN IF 'sendmail' would pass them (not convinced of that), what then
> happens at the far-end target MTA?
Having (and being unable to influence) the 'historical mismatch'
('Usernames' have Mailsaddresses <user@???>, but
Krb-principals (which is 'centralized User-Accounts') always have
(are) "user@UPPERCASEDOMAIN", the central MTA simmply does not accept
all the wrong principal's addresses.
To get the acute problem 'mostly done', we found a 'mostly correct'
but 'very local' solution - the address-generating svn server(s)
REWRITE their <username@DOMAIN> to <username@???>, if and
only if the username is of the form ([a-z][a-z0-9]+), which is
ON THE SPCIFIC HOST 'nearly' always correct. Alternatively we saw we
could work with callouts to the central, then to our local hub, so
that the svn-server's exim would find out, who really exists.
So the 'current big problem' is gone for normal use.
The real problem seemingly can not really be fixed by exim, except by
creating convoluted pathes for the mail or a kind of database-driven
rewrite for all 'possibly known' (user-)addresses.
Thanks, Stucki
--
Christoph von Stuckrad * * |nickname |Mail <stucki@???> \
Freie Universitaet Berlin |/_*|'stucki' |Tel(Mo.,Mi.):+49 30 838-75 459|
Mathematik & Informatik EDV |\ *|if online| (Di,Do,Fr):+49 30 77 39 6600|
Takustr. 9 / 14195 Berlin * * |on IRCnet|Fax(home): +49 30 77 39 6601/
