Excerpts from mail: 13-Apr-93 PAG Instructions, Please Don
[EMAIL PROTECTED] (2947)
> pagsh
> echo -n "klog "
> klog
This definitely does not result in any reasonable behavior. It starts a
new shell that is a bourne shell that is unauthenticated under a new
PAG. When the user exits that shell he gets prompted for `klog
Password:' and ends up authenticating the (possibly) unPAGed shell.
Here's how I use the pagsh command.
I use bash as my login shell, but most any modern bourne shell will work
(except Ultrix). I define a shell function
setpag ()
{
exec pagsh -c "exec $SHELL"
}
Then when I execute the command
$ echo $$
1572
$ setpag
$ tokens
Tokens held by the Cache Manager:
--End of list--
$ echo $$
1572
$
I get a new shell, of the same type I started with, executing under an
unauthenticated PAG, in the same process. Then I
$ klog
Password:
$
And I'm set. I know this is two steps, but I didn't feel like spending
anymore time trying to figure out how to make it one step. But, you
cannot execute this setpag function from within a script, as the new
exec'ed shell starts up in interactive mode and the remainder of the
script is ignored.
I only use this approach to create multiple xterms with different AFS
authentication in each. If you want access to the setpag() function for
login purposes, use the Transarc supplied login command (or build your
own, it's not very hard).
Bill Fithen