Re: Adding records to a file

Hello Buck,

You wrote:
>Good alternative.  On S/38, monitoring for a failure was about 100 times
>more expensive than the chain.  I need to update my thinking.

Exceptions are still expensive -- but like all things, expense is relative.  In
this case you are only testing for existence so the SETLL op-code and a test for
%EQUAL will work without the overhead of CHAIN (i.e., copying data in to the
program buffer) nor will SETLL 'corrupt' the buffer thus Rob's concern is
satisfied.

The old guideline of coding to the most expected behaviour still holds.  If
*THRESHOLD of my WRITEs will work then I should monitor for a WRITE failure (via
indicator, %ERROR, %STATUS, or MONITOR).  If *THRESHOLD of my WRITEs will fail
(duplicate key etc.) then I should check before i write (via SETLL or CHAIN).
The threshold value I use is between 2/3rds and 80% and I code accordingly.

Regards,
Simon Coulter.
--------------------------------------------------------------------
   FlyByNight Software         AS/400 Technical Specialists

   http://www.flybynight.com.au/
   Phone: +61 3 9419 0175   Mobile: +61 0411 091 400        /"\
   Fax:   +61 3 9419 0175   mailto: shc@flybynight.com.au   \ /
                                                             X
                 ASCII Ribbon campaign against HTML E-Mail  / \
--------------------------------------------------------------------