Erin Drummond

This is now released in [v0.125.5](https://github.com/TobikoData/sqlmesh/releases/tag/v0.125.5) The docs are here: https://sqlmesh.readthedocs.io/en/stable/integrations/engines/athena/ Take it for a spin and let us know if you encounter any issues!

For Redshift, do you see any problem exposing `DISTKEY` as `partition_by` and removing the "Dist By" concept altogether since it only applies for Redshift?

Thanks for that @MikeWallis42. Is there any chance you can represent the MRE as a set of SQLMesh model definitions instead?

@MikeWallis42 thanks for your patience, i've managed to reproduce this based on your example using the following steps: `seeds/salary.csv` ``` update_datetime,employee_id,salary '2024-07-25T23:12:00',1,15000 '2024-07-25T23:12:00',2,25000 ``` `models/salary.sql` ``` MODEL ( name michael_wallis.salary,...

We've identified this as a bug in sqlglot SQL generation and aim to have a fix up next week. It only affects the Trino dialect, other dialects work as expected

Oh man, this has uncovered a can of worms. [This issue](https://github.com/TobikoData/sqlmesh/issues/1979) and subsequent [PR](https://github.com/tobymao/sqlglot/pull/2852) is the reason for the current behaviour because I was having the opposite problem And this...

Ok, it looks like my issue was correct. The target alias does need to be removed from the `THEN UPDATE SET` part and this is also [documented for Postgres](https://www.postgresql.org/docs/current/sql-merge.html): >...

@wahmedswl the issue is that in Postgres, index names are global. So when SQLMesh creates a new underlying table, it does indeed run your `CREATE INDEX IF NOT EXISTS` statements....

We have not started working on this issue yet. You're welcome to work on it yourself if it's blocking you, we are always happy to accept well-formed PR's

I was unable to reproduce this. Assuming: - a catalog called `hive` that uses the Trino Hive connector - a catalog called `iceberg` that uses the Trino Iceberg connector -...