[2164] in Kerberos-V5-bugs
Re: rlogin bugs
daemon@ATHENA.MIT.EDU (Barry Jaspan)
Mon Aug 19 13:01:47 1996
Date: Mon, 19 Aug 1996 13:01:38 -0400
From: "Barry Jaspan" <bjaspan@MIT.EDU>
To: kenh@cmf.nrl.navy.mil
Cc: krb5-bugs@MIT.EDU
In-Reply-To: <199608191650.MAA02797@ginger.cmf.nrl.navy.mil> (message from Ken
Hornstein on Mon, 19 Aug 1996 12:50:53 -0400)
>I think if krlogin fails for any reason, you never want to execute UCB
>rlogin.
I disagree. At least falling back to UCB rlogin when you get ECONNREFUSED
makes it easy to have your users use the same binary when connecting to
Kerberos and non-Kerberos hosts.
If a site really wants fall-back functionality, it can be implemented
with a very short shell script instead of the much larger and more
complicated C code. If the shell script wants to behave differently
based on different error conditions, then krlogin can be modified to
exit with a more meaningful exit status.
I do not think that fall-back functionality is a good idea, however.
A user or site that wants security should know it is getting security
every time it runs a program. If falling back to a non-secure program
is necessary, that decision should be explicit and un-missable. A
warning displayed on the screen is not sufficient, as it might not be
noticed or might vanish before the user sees it.
Barry
