uucp and preserving Reply-to: how?

[mikebw at bilow.bilow.uu.ids.net (Mike Bilow)]

Gary S. Trujillo wrote in a message to Mike Bilow:

> Note also that is illegal to use both a '!' and a '@' in a mail address,
> such as aaaa!bbbb at cccc.

 GST> It's not illegal at all - it's just that the results are
 GST> undefined (even if one knows what will happen specific
 GST> cases), which is quite another matter. 

It isn't illegal for me to make up my own language and try to use it to start
conversations with total strangers, either, but it isn't going to accomplish
anything productive.  What's the practical difference?

> The UUCP side will parse this as "send to aaaa, which will send it to
> bbbb at cccc."  The SMTP side will parse this as "send to cccc, which
> will send it to aaaa!bbbb."

 GST> Well, address interpretation is done by the mail transfer
 GST> agent, not by the UUCP software.

There is a difference between the embedded address and the envelope address. 
While the embedded address is whatever the user types into the message, the
envelope address is supposedly something constructed as a result of parsing
that embedded address.  The envelope address is what must be meaningful to the
mail transfer agent, since it is what will actually move the mail somewhere.

 GST> I'm using Smail 3.1 and
 GST> could probably tweak things to override the default behavior
 GST> (such addresses appear so infrequently that I've never
 GST> bothered figuring out how to do so).

I have never been a fan of smail.  The last I knew, it didn't even understand
MX records returned in response to DNS queries.  I know how sendmail works, and
it works far more reliably if you are willing to put the effort into learning
how to configure it.

 GST> I just know that my
 GST> MTA will always treat "host1!user at host2" as
 GST> "user%host2 at host1" (which resolves to "host1!host2!user" in
 GST> bangpath-speak) and that I can rely on that result

This is a UUCP-flavor interpretation.  An SMTP-flavor interpretation would be
to send it to host2, which would be expected to send it to host1!user.

 GST> - but I
 GST> don't use such addresses (though I do use the bangpath form
 GST> on occasion, when I want routing to be done through a given
 GST> host for one reason or another) because I know that I can't
 GST> count on the same interpretation being placed on such an
 GST> address by the mail system at the receiving end - which
 GST> messes up replies (to say nothing of the fact that the MTA
 GST> on the receiving end will probably re-write the address). 

You will usually get consistent behavior from sticking to a set of operators
which parse either a left-to-right ('!') or right-to-left ('%', '@'), as long
as you don't mix them up.  Bang paths are discouraged these days even for UUCP
use because source routing is discouraged, and most systems have converted to
handling "UUCP" as a fake domain.  That is, users would address mail that would
have been sent to "host1!host2!user" instead to "user at host2.UUCP" under the new
system.

 GST> So I just think that one should not use "host1!user at host2"
 GST> addresses, but they're not formally illegal.  If they were,
 GST> they'd be kicked out altogether.

No, they are not formally illegal in the sense that the mail system will bounce
the message and send back a nasty warning from mailer-daemon.  On the other
hand, you never really know whether the software will decide to send such a
message to the addressee intended or to some obscure host in Siberia.
 
-- Mike