Ron Horner writes in response to the question regarding a free/inexpensive
bouncer:

> If your customer sets a variable called HPTIMEOUT, that will do the
>  trick.
>
>                          HPTIMEOUT = 20
>
>  This will cause a session to logoff after 20 mins of just sitting there.
>  But this will effect everyone on the system.

Because I'm quite interested in this myself at the moment, and because I had
forgotten that this variable existed, I've just spent a bit of time this
morning examining Ron's suggestion. The variable is defined in on-line help
as:

HPTIMEOUT   A variable used by the CI that lets a user set timed CI
            reads on $STDIN. A positive value indicates the number
            of minutes the CI waits for input.  If a timed CI read
            expires, the session is logged off.  The initial value
            is zero, which means no timed reads.  Type= W I PL

The "W" means that it's a user-settable (writeable) variable , the "I" means
it assumes an integer value, and the "PL" indicates that it is process local
variable. In general, it sounded ideal. So I created a test account and this
UDC file:

   /t testudcs
   FILE UNNUMBERED
   /l all
       1     logon
       2     option logon
       3     setvar hptimeout 3
       4     **

...and thought that I was in hog heaven.

Unfortunately, I was disappointed to discover that the HPTIMEOUT variable only
has control if the user is resident in the CI itself. If the user is in ANY
form of a program (and I checked about half a dozen different forms of
programs), the user can stay logged on indefinitely, regardless of how long
it's been since he or she actually typed something.

The good news is that the HPTIMEOUT function works precisely as advertised
when the user is sitting at the CI prompt. The bad news is that's
unfortunately not very useful. If the HPTIMEOUT variable were to control
$STDIN(X) for any process that ran under the current logon, I think we've have
a winner, and what many of us need.

Wirt "unless I'm missing something" Atmar