RE: Causing (and detecting) a workstation failure

>This leads me to believe that RPG should
>be able to intercept this sort of
>I/O error.  How did we do it before
>CEERTX and scope messages?

I distinctly recall a protracted discussion with IBM support regarding how
S/38 handled I/O error recovery.  If a user powered down the RPG program
would go into a paroxysm if you let the default error handler deal with it.
(remember, there's no display now that the user turned it off, so sending a
message to the user is an issue.)

I strongly suspect that TCP/IP has changed the way the system sees an I/O
error making those device error status codes obsolete (or perhaps valid for
SNA remote sites.)

In the case you're describing you might be better off with a watchdog timer
to detect the lost browser.  Not at all elegant.
  --buck