Fritz Larco

try now, this branch hasn't passed tests yet, so there may be bugs.

Closing, feel free to re-open.

Yea, deletes are a bit tricky. The way Hightouch does this currently is actually inefficient because it does a full select of the source data each time, whereas sling adds...

This is added. Please see https://docs.slingdata.io/examples/database-to-database/incremental#delete-missing-records-soft-hard. Closing.

@dhanasekar-ravindran-E3673 can you try with env var `SLING_DUCKDB_COMPUTE=false`?

hey @colinmollenhour, that's something only [supported](https://duckdb.org/docs/stable/data/partitioning/partitioned_writes#partitioned-writes) via duckdb. Actually, can you try with `SLING_DUCKDB_COMPUTE=true` (or just omitting the env var) with `v1.4.24`? I think the hang up issue should be...

Hey @metalshanked , this is being worked on. Should come out soon.

See https://docs.slingdata.io/sling-platform/platform/self-hosting

Hi, thanks for the write-up and details. I've put some draft changes in the `parquet.go` file, but thinking over, I'm not too sure how to provide the input. I'm reluctant...

I like `timestampz(-0400)` 👍 . But then that's at the target side, which conflicts with the logical representation of the `columns` options, which is currently at `source_options`. I've been grappling...