Re: 5250 and un-do

There is an MI TOD concept, the "Standard Time Format", and it is a 64-bit value. It changed at 6.1 of the OS, before that, the 1st 48 bits represented the number of 8-microsecond units of time. The remaining bits, 49-63, guaranteed uniqueness.

At 6.1, the time unit went down to 1 microsecond, and the 1st 51 bits represented the number o 1-microsecond units of time. This provided better granularity to the microsecond level, The uniqueness bits were those from 52-63.

I believe this uniqueness goes across the system, so that you get up to 4,095 events within the same microsecond - I think - the uniqueness values are non-zero, if I read things correctly.

Anyone, please correct me as needed.

Vern

On 12/21/2017 9:35 AM, Buck Calabro wrote:

On 12/21/2017 10:01 AM, Luis Rodriguez wrote:

Mmmm.. IIRC, in the latest DB2 versions (7.3 at least, don’t know if
available in 7.1/7.2 latest TRs) if you define a TIMESTAMP column it will
include a sequence that ensures the no duplication of the data?

Not near an IBM I (or the manual) right now, but that should avoid the
requirement for nanoseconds precision.

Thanks for reminding me about this. The reference is
Database > Programming > SQL programming > Data manipulation language >
Retrieving data using the SELECT statement

https://www.ibm.com/support/knowledgecenter/ssw_ibm_i_73/sqlp/rbafydtts.htm

'When the CURRENT TIMESTAMP special register or a variable with the
TIMESTAMP data type is used with a precision greater than 6, the
timestamp value is a combination of the system clock and uniqueness
bits. The uniqueness bits are assigned in an ascending order. Therefore,
comparison operations for timestamps with any precision will represent
an accurate order of when the timestamps were assigned.'

I myself tend to think of GENERATE_UNIQUE for a guaranteed unique value,
and I think of TIMESTAMP for... well, time.