Hi! We're working on switching over to the MeltanoLabs tap-postgres and have a question about

_sdc_extracted_at

. We sometimes use this ts to determine what rows came on the most recent sync. In the transferwise tap this works because the select is performed in a transaction. When testing we noticed that does not seem to be the case with the meltano varaint. I think this is the relevant line of code https://github.com/MeltanoLabs/tap-postgres/blob/e614378d49a516b597960fedc8736bd3f7d5a290/tap_postgres/client.py#L245C9-L245C48 -- we found other queries in the sdk that use the

begin

in the

with

statement to open a transaction (link) So I guess I have two questions: 1. is this intended behavior or a bug? 2. If the answer to #1 is "intended" can you suggest a pattern for dealing with hard-deletes up stream OR if the upstream table lacks a PK so it can't merge. P.S. We are syncing to snowflake using the pipelinewise target (switching one at a time) and this is for full-table replication Edit: I was totally wrong about the transaction thing. It looks like pipelinewise just gets the

current time

once and passes it to all of the records instead of getting the current timestamp for each record. So I guess this is just a different design. Still open to ideas but clearly isn't a bug