Author: Bill Woodger
Subject: Reply to: Next max
Posted: Fri Jun 17, 2016 2:30 am (GMT 5.5)
Full knowledge of the data is important. Consecutive blanks would be a problem, as well as blanks in the final record of a key.
Still, assuming those don't exist, WHEN=GROUP for the blank, with PUSH of the entire record, RECORDS=2, OUTFIL to omit the blanks, and a BUILD with the slash-operator (get the basic data, (81,80,/,1,80), a little breakdown to copy the required bytes and the literal) subordinate to a WHEN= confirming that the keys are equal should be about there.
So, can there be consecutive blanks, and if so, is there a maximum? And what would you want done? If the final record of a group is blank, what would you want done?
I can't see that the JOINKEYS needs the RESTART, just a plain sequence should do. A test for equal keys could replace the primary key and just use the sequence for the match.
But the important thing first is the data.
Subject: Reply to: Next max
Posted: Fri Jun 17, 2016 2:30 am (GMT 5.5)
Full knowledge of the data is important. Consecutive blanks would be a problem, as well as blanks in the final record of a key.
Still, assuming those don't exist, WHEN=GROUP for the blank, with PUSH of the entire record, RECORDS=2, OUTFIL to omit the blanks, and a BUILD with the slash-operator (get the basic data, (81,80,/,1,80), a little breakdown to copy the required bytes and the literal) subordinate to a WHEN= confirming that the keys are equal should be about there.
So, can there be consecutive blanks, and if so, is there a maximum? And what would you want done? If the final record of a group is blank, what would you want done?
I can't see that the JOINKEYS needs the RESTART, just a plain sequence should do. A test for equal keys could replace the primary key and just use the sequence for the match.
But the important thing first is the data.