Re: duplicate key error in identity column

On 15-Jul-2015 23:06 -0600, Birgitta Hauser wrote:

Dan on Wednesday, 15.7 2015 20:51 wrote:

[...] v7r2 has an enhancement to the timestamp that allows 12
digits to be defined for the "microsecond" [...]

As an aside:
Timestamps with up to 12 digits can only be managed in SQL described
tables, views or indexes, but cannot be defined within DDS described
physical or logical files
... but all 12 digits for the micro seconds are only filled when
using the SQL Special Register Current_Timestamp(12).

I suspect "only" approaches hyperbole or is at least misleading; e.g. the CAST scalar supports TIMESTAMP(p) syntax.?

When performing the write or update with native I/O as before only
the first 3 digits are filled

Presumably for when using a HLL timestamp host variable or HLL BIF\OpCode to assign the value; i.e. the implication being that the HLLs may not have added any support for additional precision with their HLL timestamp data type, instead adding support only for what was minimally necessary for RLA [Row-Level Access, aka native I/O]? Native I/O should fully support all precisions, despite what limitations any HLL might have for timestamp support; the I/O buffer would necessarily be adjusted for the increased precision, and any datum would merely be passed to the LIC DB to be validated for which there is unlikely to be any special-case to truncate\round the timestamp values.? If an HLL would allow open\read\update a file with the larger precision, surely the timestamp data does not surreptitiously lose precision; that surely would be a defect?