Heck, make that LOOKUP FLOOR with a key of highvalues.
On 03/18/2015 04:30 PM, John P. Hartmann wrote:
I was going to suggest LOOKUP CEILING.
On 03/18/2015 04:27 PM, Rob van der Heij wrote:
Not sure where MVS is in the picture. Are you talking BatchpipeWorks or
such? I'm rusty on what level is there...
Have you considered using 'lookup' to pair the start/stop messages?
You can
even feed the 'stop' message back into lookup to drop the master, so your
keys can be serially monogamous. I used something like that to track
spool
files by RSCS messages on various S&F nodes to check our SLA and identify
which nodes were causing the problems.
Rob
On 18 March 2015 at 16:19, Glenn Knickerbocker <[email protected]> wrote:
On 3/18/2015 10:53 AM, FAULHABER Michael wrote:
"another day another darling", so why do you mind that job1 runs two
days later again? Currently you look for job1, job2 and so on. I
suggest (just for the analysis) to add job and date to one key. So in
the next step you care for the key with the highest phase number.
That's exactly what I did first when I found out I was missing reruns,
but it gave me too much output. I don't care how many days a run takes.
I just want to know the current status of each one.
Oh, and I just noticed the reason I needed that extra record at the
start of the file: SELECT SECOND was in effect when I issued the WRITE.
That has another unwanted effect of writing a null record at EOF.
Instead of adding a record with LITERAL, what I need is to remember to
SELECT FIRST before that WRITE.
¬R
