On 3/25/25 00:47, Sami Imseih wrote:
I know it was mentioned above by both Michael and Andrei that
AppendJumble should not be exposed. I am not sure I agree with
that. I think it should be exposed, along with
JUMBLE_FIELD, JUMBLE_FIELD_SINGLE and JUMBLE_STRING
and _jumbleList.
It would be helpful if you could provide an example to support the
reasoning behind exposing this stuff. I assume you want to influence the
logic of jumbling, correct? If that’s the case, it might be beneficial
to add a callback to the JumbleState that is triggered during each node
jumbling attempt. This could facilitate moving clocations code and data
into pg_stat_statements and give developers the opportunity to influence
the jumbling process, adding extra meaning to their hashes.
One reason this might be necessary is that generating the same hash for
both a Param node and a Const node could lead to a more generalized
hash, allowing us to group more queries under the same value.
--
regards, Andrei Lepikhov
