SOLVED-[Exim] Remote delivery problem on dialup

Thanks to everyone who helped. I solved my problem and
exim is now able to deliver mail on my dialup machine.

To briefly describe what lead up to the fix and the fix itself:

I first tried setting various variables in mutt (envelope_from, use
ESC-f to set "from" just before sending), and none of these changes
in the MUA seemed to change anything in the envelope used by
the MTA. Mail was still not being delivered.

I tried to put "set untrusted_set_sender" and "untrusted_set_sender" in
/etc/exim.conf, but this threw an error and no mail was sent. (I'm not
sure why this is. This was mentioned by several users, and is mentioned
in the spec. Maybe I'm not putting it in the right place).

In ~/.muttrc I tried setting sendmail=/usr/sbin/exim -oi -oem and
/usr/sbin/exim -f,
but neither really changed things for my situation. Using the -f switch in
combination with the ESC-f option in mutt gave an interesting log entry, a
one-liner that had as the sender "--@erols.com" (this was when I had my
qualify_domain=@erols.com). Of course this would be rejected.
(the -oi -oem setting looks like it could help track down a problem, but I
should probably read up on it a bit more before using it all the time).

It finally worked when checking the log, thinking about the what the isp's
side
of the log looked like (thanks again Gary) and seeing how changing my
qualify_domain changed my log (and presumably how my isp "saw" my
request to forward mail). It was a combination of having my qualify_domain
be the same as my machine listing in the "rewrite configuration" section AND

having the appropriate entry in /etc/email-addresses. My problem was that,
until the end, none of these things were synced to get the mail out. By
the time I added an entry in /etc/email-addresses, I had already changed
the qualify_domain value in my /etc/exim.conf to various things (my host
name
alone, my isp name). So when called to send mail, the qualify_domain
setting
didn't match the name in the rewrite configuration, so the
/etc/email-addresses
wouldn't be called.(?)

The log now shows resolvable addresses for both the "<=" and the "=>"
dialog.
(Before the "<=" part was not a resolvable address, and once my isp caught
it,
they would try to send it back to that address, but because that was not a
resolvable
address (not to mention it's not mine), I would never see the report that
this was
being rejected w/a 550 error.)

So, AFAIK, if I'd just left the default config alone (after picking #2 and
answering
the questions during the installation) and added the appropriate entry to
/etc/email-addresses, I would have been good to go from the beginning.

Thanks again for everyone's help, George

ps - Please cc me for any responses, b/c I'm not signed onto this list.