Getting smail with ISC 2.2 to work

[staff]

rolfe at w3vh.UUCP (Rolfe Tessem) writes:
>Dante_A_Nicolello at cup.portal.com writes:
	...
>>I am trying to set up smail to send all my uucp transfers to another
>>system for address revision. I set the SMARTHOST=smart-host in the
>>/etc/default/smail file and set  a smart-host (my_smart_host)!%1
>>in my /usr/lib/uucp/paths file. Doesn't work. Either smail or sendmail
>>or whatever does not forward the mail. It just sends it back.

>I'm glad to hear someone else mention this, if only as a sanity check :-).

>I've spent an incredible amount of time trying to make sendmail/smail work
>right on a new ISC 2.2 system, to no avail.  Smail is just never getting
>invoked, it appears.  Running smail directly in test mode seems to work
>(at least the right headers are generated locally).

I've fought this battle too, on my home machine with Interactive 2.2,
and I have lost!  I'm NO sendmail expert, but from the .cf it seemed
to me that uux would be directly invoked, rather than smail.

The funny thing is that when I replaced the real rmail/smail with shell
scripts confirming that they had been called and then exec'ing the real
programs, things got better - but not FULLY OK.  In the end, I have
installed smail 2.5 from the original sources (no need for sendmail, no
local net in my home).  Oh yes, 'deliver' was also needed, the various
*mail*'s wouldn't work properly as local delivery agents.

One tip which might help with smail:  I think there's a bug in the
release/patchlevel I'm using (maybe in Interactive's, too?), when
/usr/lib/uucp/paths is VERY small, with the binary-search code in
routine getpaths():  it is looking for 'smart-host', goes to the middle
of the paths file, and it is the 'm' in 'smart-host'; from there on, it
goes forwards, finds end-of-file, concludes that 'mart-host' is the LAST
entry, and since that's alphabetically BEFORE 'smart-host', gives up!
Easy to fix in source (no reason for binary lookup in such a SMALL paths
file, just read it all in memory first time and linear-search through
it), and just adding a dummy path to 'zulu' as the last line in paths
seemed to hide the bug too, without changing the getpaths() routine.
-- 
Alex Martelli - CAD.LAB s.p.a., v. Stalingrado 45, Bologna, Italia
Email: (work:) staff at cadlab.sublink.org, (home:) alex at am.sublink.org
Phone: (work:) ++39 (51) 371099, (home:) ++39 (51) 250434; 
Fax: ++39 (51) 366964 (work only; any time of day or night).