Adriano dos Santos Fernandes wrote:

>>
>>
>>
>>>Or do you think we should generate the RSE (constructed based on the PK)
>>>in BLR even for REPLACE?
>>>
>>>
>>>
>>This is questionable. I would do it this way (construct the PK references in
>>DSQL), but your suggestion about a new BLR code and engine-level handling
>>also looks good. I really don't know what's better.
>>
>>
>>
>I prefer the engine-level handling, although it's require more studies
>from my side. :-)
>
>
>
>

There should be a new blr verb parallel in syntax to blr_store. There
is no reason to tell the engine about primary keys -- it already knows.