> > such as re-presenting the MAIL TO with a valid sender if you want to
> > (or configurable to just use the alternative valid sender in the first
> > case, might make more sense).
>
> What happens if that machine tries to do a callout on the now "valid"
> sender?
I've been watching the thread. In this case the burden I feel would be on
the person doing the original call out.
Here's how I see how it should happen:
<< smtp from client
<< mail from:<client@here>
>> ok
<< rcpt to:<recipient@there>
** Do a callout here
>> smtp to client's server
>> mail from:<>
<< No address found
>> reset
<< ok
>> mail from:<postmaster@there>
<< ok
>> rcpt to:<client@here>
** they do a callout here for postmaster
<< smtp from client's server back to "there"
<< mail from:<postmaster@here>
>> ok
<< rcpt to:<postmaster@there>
** Unconditionally accept for postmaster or
** immediate reject. No call out.
>> ok
<< call out ok, continue
>> quit
>> call out ok, continue
<< data
>> give me the rest ending with a single dot.
<< ...
>> ok
<< quit
>> ok and closing.
If allowed to use any address than a local address that does callbacks, then
there would be a loop. If they don't do call backs, I don't see how a loop
could possibly happen.
--
Lab tests show that use of micro$oft causes cancer in lab animals
