I went into documentation again about date-time stuff. The late Bruce
Vining really knew what this was about. He once told me his wife didn't
let him go to social events, he'd talk only about calendars.
Anyhow, here's a link with some info -
https://www.ibm.com/docs/en/i/7.5?topic=values-date. Internally, a date
is a Scaliger number, the number of days since noon on January 1, 4713
BCE. This is an integer. There's more info here -
https://www.tondering.dk/claus/cal/julperiod.php. The first link
mentions the length in the SQLDA as the length of character
representation - that depends on the format, it says. That might be
closer to what DSPFFD shows.
I created a PF using DDS, 2 date fields, one DATFMT(*YMD), the other
DATFMT(*JUL), and a time field, TIMHMS TIMFMT(*HMS) - DSPFFD reports
lengths as 8 , 6, and 8 resp., with mention of format. SYSCOLUMNS shows
lengths of 4, 4, and 3, and nothing said about format. SYSCOLUMNS is a
view over one or other of the QADB* database cross-reference files,
maybe the view leaves out the format.
Time is represented as 3 bytes, first if hours, then minutes, then seconds.
I ran the following SELECT to see internal form, thinking the hex value
might show me -
select datymd, hex(datymd), datjul, hex(datjul), timhms,
hex(timhms) from vhamberg/testdattim;
The result is here -
DATYMD HEXYMD DATJUL HEXJUL TIMHMS HEXHMS
2023-09-16 00258A2C 2023-09-16 00258A2C 05.51.17 055117
That integer for the hex values for dates is 2,460,204 - that is the
Scaliger number for today. That tondering site above shows it as
2,460,203.91, the decimal part is how much of a day has passed since
noon yesterday, so the value in the table appears to be rounded up. Or
otherwise adjusted.
Now you can invite me to social events - or not! Bruce knew SO much more
than I do!
Regards
Vern
On 9/15/2023 6:49 PM, x y wrote:
On V7R5, the output shows a date data type field has a length of 4 (which
is the actual size of the binary field on the persistent storage device)
and a timestamp data type column has a length of 10 (also the actual binary
size). I haven't checked a time data type yet.
AFAIK this is not consistent with DSPFD/DSPFFD and API output. Is this
variation by design or is it a miss?
midrange.com
