In that case, insert signal off error before and signal on error after
the streamstate.

Regards, 
Richard Schuh 

 

> -----Original Message-----
> From: The IBM z/VM Operating System 
> [mailto:[EMAIL PROTECTED] On Behalf Of Berry van Sleeuwen
> Sent: Wednesday, June 11, 2008 3:29 AM
> To: [email protected]
> Subject: Error handeling (Was: How to stop PIPE STARMON)
> 
> Hello Rob,
> 
> Are you sure this will work in all cases?
> 
> The SIGNAL ON ERROR will react to any non-zero returncode. 
> STREAMSTATE ca= n have returncodes 0, 4 and 8, and it looks 
> to me that all three are acceptable, at least in my case. 
> Only rc=12, stream not connected, is =
> 
> indeed the returncode I will trigger upon. But the SIGNAL ON 
> ERROR will =
> 
> exit the stage in all returncodes other than zero.
> 
> But more in general, there are more commands that will exit a 
> returncode =
> 
> based on their function. WAKEUP for instance. Just a small 
> test confirms =
> 
> this:
> 
> signal on error         
> 'wakeup +00:30 (cons'   
> say rc                  
> error:                  
> exit rc                 
> 
> The result is that wakeup will return rc=2 when the timer 
> expires and 6=
>  
> when the console interrupt is received. In both cases valid 
> returncodes, =
> 
> or perhaps required returncodes, but in both cases the SIGNAL 
> traps the 'error' and skips the "SAY rc" command.
> 
> In such cases it looks like errorhandeling should be coded 
> for every individual event instead of a general error handeling trap.
> 
> Regards, Berry.
> 
> On Mon, 9 Jun 2008 21:38:15 +0200, Rob van der Heij 
> <[EMAIL PROTECTED]> =
> 
> wrote:
> 
> >signal on error
> >do forever
> >   peekto
> >   .. process..
> >   readto
> >   streamstate output
> >end
> >error: return rc * ( rc <> 12 )
> 

Reply via email to