You should realize that SORT will not be concerned/understand the decimal positions and will treat everything as integer numbers
so ( forgive my cobol approximation, I am not that fluent in COBOL )
if You have a packed variable with two decimals 999.99 it will be stored in three bytes and from the sort point of view You should consider it as a <unorthodox> integer with no decimal
if COBOL understands the value as 123.45 SORT will understand just 12345 !
so just drop the decimals and carry on all the decimal point elucubrations Yourself
just tested with DFSORT ,
using INCLUDE COND=(1,10,PD,NE,+100000.00)
gives a return code 16, comparison field error under the decimal point,
something has not been told here
Bill, the sign does not make any difference, I just tested with both x'c' and x'f' signs and with a +xxx and xxx value,
the result does not change ( the output is correct )
what I do not understand, as i said , is how the TS can complain about bad output when the sort would never work as written!