ables@catwoman.ACA.MCC.COM (King Ables) (09/08/88)
From article <15805@shemp.CS.UCLA.EDU>, by wales@valeria.cs.ucla.edu: > > It seems, then, that the best thing for us to do is to add an "Oi" line > to our Sendmail configuration file. However, before I go ahead and do > this, I would like to double-check with the net readership in order to > find out whether there are any hidden problems with doing this. For > instance, will "Oi" mess up the "period is EOT" behavior of Berkeley > "mail"? (Or is this done by separate logic in Berkeley "mail" itself?) It's been my experience that many mailers that connect to me depend on being able to end their messages with a .<CR> line. In fact, don't most SMTP listeners instruct you to end with a .<CR> after the DATA command? I don't think I'd want to use Oi... -king ARPA: ables@mcc.com UUCP: {gatech,nbires,seismo,ucb-vax}!cs.utexas.edu!pp!ables
wales@valeria.cs.ucla.edu (Rich Wales) (09/09/88)
In article <15805@shemp.CS.UCLA.EDU> I suggested the possibility of
adding an "Oi" line to our Sendmail configuration file, in order to
allow users to send mail with lines consisting of only a period.
It turns out that this is a *BAD* idea in the current Sendmail. If you
do put "Oi" in your sendmail.cf, it will mess up your SMTP daemon.
(Look in the module "collect.c", for occurrences of the global Boolean
variable "IgnrDot" -- which is set by "Oi" in the configuration file,
and/or by a "-oi" command-line argument. If "IgnrDot" is set, it will
bypass special "dot" processing, even when Sendmail is in SMTP mode.)
My personal feeling is that SMTP mode ought to be made to override "Oi"
in the Sendmail code. And since I assumed this was utterly obvious, I
ended up misreading the relevant part of the Sendmail source when I
first looked at it. (Sigh.)
Now that I realize that "Oi" will break SMTP, the only reasonable thing
to do is *not* to put an "Oi" line in our sendmail.cf after all -- but
rather to modify our user agent program to use a "-oi" command-line
argument when it invokes Sendmail.
Thanks to John Owens and Chris Torek for pointing the problem out to me.
-- Rich Wales // UCLA Computer Science Department // +1 (213) 825-5683
3531 Boelter Hall // Los Angeles, California 90024-1596 // USA
wales@CS.UCLA.EDU ...!(uunet,ucbvax,rutgers)!cs.ucla.edu!wales
"Spiff's hyper-freem drive malfunctions! The aliens close in!"andrew@cc.brunel.ac.uk (Andrew Findlay) (09/09/88)
In article <15805@shemp.CS.UCLA.EDU> wales@CS.UCLA.EDU (Rich Wales) writes: >.... >Further investigation revealed (not surprisingly) that our Sendmail con- >figuration file did *not* have the "Oi" configuration option set. As >best I can figure out from reading the documentation and the source, the >"Oi" option disables the "period is EOT" logic and causes Sendmail to >treat a line with only a period exactly like any other line of text. > >It seems, then, that the best thing for us to do is to add an "Oi" line >to our Sendmail configuration file. However, before I go ahead and do >this, I would like to double-check with the net readership in order to >find out whether there are any hidden problems with doing this. For If you set Oi in the config file it will apply to *all* messages received by Sendmail. This will probably wreck the SMTP function, so your incoming Internet/Ethernet mail will stop working. The 'correct' solution is to include the flag '-oi' in the command line when your user-interface program invokes Sendmail. We had a similar problem with our (UK-specific) Coloured-Book mail system. All was fine until one day a message came in that had a single dot alone on a line. Sendmail took that to be end-of-message and closed the pipe. The FTP system that had received the message file took that as an error and in five minutes time it tried the whole thing again... The '-oi' flag fixed the problem. The 'period-is-EOT' behaviour in Berkeley Mail is a separate issue, and is configurable in .mailrc. Andrew -- --------------------------------------------------------------------- | From Andrew Findlay at Brunel University, Uxbridge, UB8 3PH, UK | | Andrew.Findlay@brunel.ac.uk phone: +44 895 74000 x2512 | ---------------------------------------------------------------------