Re: [MI400] More fun with nullmaps

[Gene_Gaunt@xxxxxxxxxxxxxxx]

Thanks James.
If you block records (UFCB offset byte 184, bit 2, is on), notice your
11-byte gap will contain this record feedback information after a READ:

BIN(2)  member number
BIN(4)  RRN within member
CHAR(1)  flags: duplicate key in file, etc.
BIN(4)  zero

For some reason (I don't know why) the OS writes blanks into these 11 bytes
if you don't block records (if UFCB offset byte 184, bit 2, is off).

If you make your file a keyed file, and change the UFCB by replacing that
parameter ID value 60 (arrival sequence) with parameter ID value 53 (key
feedback), and if you use record blocking (UFCB offset byte 184, bit 2, is
on), and have NO variable-length records, then the pointer in the input
buffer (not the pointer in the UFCB) after a READ points to:

field data
BIN(2)  member number
BIN(4)  RRN within member
CHAR(1)  flags:  duplicate key in file, etc.
BIN(4) zero
key fields
key null byte map
field null byte map

So 2+4+1+4=11.  Clear?