Path: utzoo!attcan!uunet!decwrl!ucbvax!CS.NIU.EDU!rickert
From: rickert@CS.NIU.EDU (Neil Rickert)
Newsgroups: comp.mail.sendmail
Subject: Re: RFC-compliant UUCP mail (fwd)
Message-ID: <9007261520.AA11205@cs.niu.edu>
Date: 26 Jul 90 15:20:47 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Lines: 101


   In article: 730@logicon.com, Jeff Makey  <Makey@Logicon.COM>
writes:
>
>If I am not clear, I want to output
>
>    From host.domain!user  Wed Jul 25 16:39:22 1990 remote from mysite
>    From: user@host.domain
>
>instead of
>
>    From host.domain!user  Wed Jul 25 16:39:22 1990 remote from mysite
>    From: mysite!host.domain!user
>
>or
>
>    From user@host.domain  Wed Jul 25 16:39:22 1990 remote from mysite
>    From: user@host.domain

 If you use sendmail 5.64+/IDA-1.3.4 you have two choices.

 If you use the UUCP-A mailer, you will get:

    From host.domain!user  Wed Jul 25 16:39:22 1990
    From: user@host.domain

 If you use the UUCP mailer you will get:

    From host.domain!user  Wed Jul 25 16:39:22 1990 remote from mysite
    From: host.domain!user

 The choice of mailer is specified in the DBM lookup MAILERTABLE.  There
may be some variations from these formats depending on whether 'host.domain'
is the same host as 'mysite' and on what you have in UUCPXTABLE for
'host.domain'.  The variations only affect the UUCP mailer.

 The 'remote from mysite' is omitted from the output with the UUCP-A
mailer, because otherwise the receiving 'rmail' will translate the
address into 'mysite!host.domain!user' with presumably redundant routing.

 The versions of sendmail/IDA and the IDA config package are available from
uxc.cso.uiuc.edu via anonymous ftp, with name
	/pub/sendmail.5.64+IDA-1.3.4.tar.Z

 My thanks to Paul Pomes (paul@uxc.cso.uiuc.edu) for helping me test my
major rewrite of the config package, and for making it accessible via ftp.

 There have been some changes in the package over the last two weeks
specifically related to UUCP mail formats, so anyone who picked it up
earlier will need to get the latest version.  The changes included a
code change (in parseaddr.c if I remember) needed to fully benefit.

  ---- Further comments on the IDA sendmail, and RFC1123 ----

In article <ENAG.90Jul25163050@slembe.ifi.uio.no>, 
	enag@ifi.uio.no (Erik Naggum)
 (in response to my earlier comments) writes:

>Instead of complaining, and showing us all kinds of broken systems,
>why don't you try to make things according to standard?  That means,

 That is what I have been trying to do.  That is why I put a lot of
effort into what was almost a complete rewrite of the IDA sendmail config
package.  It now probably comes closer to adhering to standards than
most systems.

 However, it interprets 'a!b%c' as 'a -> c -> b'.  This is probably not
in disagreement with reasonable interpretations of RFC1123, since it can
be claimed that 'a!b%c' is a purely UUCP address, not an Internet address,
so RFC1123 does not apply.

 It currently interprets 'a!b%c@d' as 'd -> a -> c -> b'.  According
to some interpretations this is contrary to RFC1123.  Never mind.  If you
prefer the interpretation 'd -> c -> a -> b' you need only remove the
comment from one rewrite rule in ruleset #9.  Unless the domain 'd'
of the '@d' portion is your own, it usually will not matter which
interpretation you use, for although the address is fully parsed, including
the 'local part', it is parsed in a way which allows complete restoral to
the original format in the final ruleset #4.

>for instance, that you sit down and try to grok the intentions behind
>the specs, instead of comparing the standard to your broken system
>which just happens to work because others follow the robustness
>principle:
>
>	"Be liberal in what you accept, and
>	 conservative in what you send."
>
 Where is the conservatism in interpretations of RFC1123 which wish to
output mail using the '%-hack' in a way which is inconsistent with the
interpretation of the majority of those existing mailers which use both
'!' and '%' ?  Better to make the RFC822 source routes, with all their
ugliness, work.  Better to demand that all routes have exclusively '!' and
'@' or exclusively '%' and '@'.  There is too long a history of usage of
'%' where 'a!b%c' means 'a -> c -> b' for their to be any reasonable hope
of reliably reversing this interpretation.

=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
  Neil W. Rickert, Computer Sci Dept, Northern Illinois U., DeKalb IL 60115
  InterNet, unix: rickert@cs.niu.edu              Bitnet, VM: T90NWR1@NIUCS
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=