On 1/7/26 1:40 PM, Dan Bale via RPG400-L wrote:
Can you answer Charles' question?
I did. "First" and "last" are by key, so no, there's nothing keyed
before D'2017-02-04'.
The first record in the PF (which is the last one in the descending
LF) is keyed D'2017-02-04.' The last record in the PF (which is the
first in the LF) is keyed 2028-12-30.
And this being a test program for a bizarre problem, the source code is
constantly being changed. For example, I just now tried adding an H-spec
with an explicit DATFMT(*ISO). No change in behavior.
Now THIS is somewhat interesting. I took your suggestion and added
DUMP(A) statements. Here's what I got:
For the first READ (after SETGT on JAN31), it shows (taking out the hex):
Key Value:
*2017-02-04 *
Input Buffer:
*2017-02-042017-02-05022017Y00000*
*0000000001 *
but still (squeezing out spaces and hex):
SM_ID PACKED(15,0) 000000000000000.
SM_OK CHAR(1) ' '
SMCARRYOVR DATE(10) '0001-01-01'
SMLASTDATE DATE(10) '0001-01-01'
SMMONTH PACKED(2,0) 00.
SMYEAR PACKED(4,0) 0000.
The dump after the second READ (with SETGT on FEB04), and the third read
(with SETGT on FEB05), it gets:
Key Value:
*2017-03-04 *
Input Buffer:
*2017-03-042017-03-05032017Y00000*
*0000000002 *
But the fields are still as before.
And the last READ (with SETGT on APR25), it gets:
Key Value:
*2017-05-06 *
Input Buffer:
*2017-05-062017-05-07052017Y00000*
*0000000004 *
So it's reading the records, and given that it's currently doing SETGTs,
it's getting the right records, but for some reason, scatter-gather
isn't working!
--
JHHL
(Over 30 years and I've never put a DUMP statement into a program.
Didn't even notice it was available.)
midrange.com
