Those who dislike header re-writing should look away now. I warned you.
san-loss is to follow.
Do any exim users make use of the "feature" of Exim to re-write headers to
replace the local-parts of addresses with different names? The example in
spec.txt (for 1.62 at least) suggests:
*@hitch.book.fict ${lookup{$1}dbm{/etc/realnames}\
{$value}fail}@??? bcfrF
to replace e.g. fred@??? with F.G.Bloggs@???.
The above seems to have some odd properties since each address is presented in
isolation, and irrespective of where the message came from. So a message from
jim@??? Cc:'d to fred@??? (but not To: him) will have that
header re-written on the way in i.e. it will re-write a header in an
externally generated message.
Assuming I want re-writing but only on the way out what should one do? Am I
being worried over nothing, is the extra re-writing is harmless?
In fact our equivalent of /etc/realnames maps all valid (personal) aliases to
the prefered form for outgoing mail (a quirk of how we build it), so this
would mean that when Cc:'ed you always get headers showing what form the
outgoing mail would come from. (some people may think this is neat!)
If this is "correct", shouldn't this also be done for To: headers (or in fact
all headers).
In my case it is of course more complex, I want the re-writing done only for
messages leaving our site. In smail I do this using a transport specific
re-write rule. I can probably hack something very nasty in exim to simulate
this, but should I just bite this bullet and force re-writes everywhere?
[ The reason I've stuck with smail for so long is that I understand the
re-writes code. This summer I finally plan to switch to Exim having run a
pilot server (on a small domain with only ~100 users and no re-writes) for the
past year. ]
-- Jon
