RE: Keyed PF

["Henrik Krebs" <hkrebs@xxxxxxxxx>]

Tom Liotta / Martin Booth - we seems to have lost the original track: Are 
records in non-unique-keyed PF recieved in arrival seq. whithin the key? And 
the answer: There is no guarantee for that. 
Then I told that I have never *seen* a case where it actually was not in this 
order, even after a keyed or non-keyed RGZPFM except for the obvious case that 
it is a reuse-deleted file (in which case arival seq. is totally unknown - 
should rather be called rrn-seq.) B.t.w. I could suspect that reuse-deleted 
files behaves different in this aspect. Never seen it. Probably never looked 
for it.
I should of cause have added another obvious situation: After deliberately 
re-ordering the physical sequence of records in the file with GSORT (Don't know 
what it is), FMTDTA (called s/3 reformatting utility on s/38, I think #DSORT on 
s/36, $DSORT on s/3) or in other ways.

But as there is no guarantee for this feature you should not rely on it.

Btw I can recommend FMTDTA as a possibly performence improovement with large 
files, even for creating use-once sorted files. I have seen cases where 
creating a temporary file and reading this is faster than reading an existing 
accesspath! The source for FMTDTA is not easy to read (as ugly as a Cobol 
programmer must find old RPGIII code), but there are commands out there to take 
command parameters as input in stead of FMTDTA-source.
 
Henrik