Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!ucsd!ucbvax!CAEN.ENGIN.UMICH.EDU!paul
From: paul@CAEN.ENGIN.UMICH.EDU (Paul Killey)
Newsgroups: comp.sys.apollo
Subject: Re: addressees at APOLLo.COM (Apollo's sendmail)
Message-ID: <4d34a015e.000f088@caen.engin.umich.edu>
Date: 5 Oct 90 00:29:28 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 29


compiling 5.61 sendmail on apollos is one thing, but getting
functionality people assume therein is another.

E.g.

What if this sequence fails?
	pw = getpwuid(geteuid());

Well, if it fails in one place, all your outbound mail is from
postmaster, and not who really sent it.  If it fails elsewhere you
reference a null pointer and blow up.  Vanilla sendmail does not
accomodate transient registry failures.  You can bounce mail with "No
such user" when the error is really a temporary failure that should
return a 421 code, or an OK with a retry from the queue later.

In addition to rgy quirks, there is the (different from Unix) apollo
notion of file concurrency control.  Do you have a high tolerance for
"Unknown error mailer error #1" messages when /bin/mail exits with a
status of 1?  Or "text file busy" or whatever when your /usr/spool/mail
file is being accessed simultaneously from 2 different nodes?

If you usually have your alias files automatically updated (have OD in
your .cf file), or just have the bad luck to have sendmail try an alias
lookup while another one is doing a dbminit() on another node, you lose
because dbminit() opens the files for write, even if you end up only
needing to read them.

Etc.