Erin Drummond

@MikeWallis42 did you manage to figure out a MRE for this issue?

SQLMesh doesn't create a view called `sqlmesh.model_snapshots` in its state database. Is this something specific to your environment?

Oh, nice, Athena has some kind of "native" DataFrame provided by the library? In that case, [_fetch_native_df()](https://github.com/TobikoData/sqlmesh/blob/751c38dfdd19589a94bb81c2d701c6c995f18071/sqlmesh/core/engine_adapter/base.py#L2078) should be implemented on `AthenaEngineAdapter` to return a `pd.DataFrame` that is backed by...

Nope, uv has never been the default, it just happens to be what that user was using. Generally what we do is something like: ``` $ python -m venv sqlmesh-env...

The clustering key is a performance optimization and does not appear to have any effect on locking. I ran a test that went like: - Create table (id, day) clustered...

In that case, if we can't relax SQLMesh's consistency guarantees then we cant support concurrent batch evaluation on Snowflake unless they add some kind of new feature that allows parts...

DuckDB treats unquoted identifiers as lowercase and SQLMesh makes this explicit at the normalization step. When specifying the columns, does it let you quote the column names? eg: ``` select...

As indicated in the slack thread (repeating here for convenience), right now SQLMesh does not have a way to specify column constraints at table creation, we can only run ALTER...

@mykalmax do you know what might cause this?

Can you check the browser developer tools for errors loading resources? It's possible that sandboxing this in a devcontainer is causing some URLs to be generated incorrectly