guillesd

Hi @swasheck this is my fault, core nightly is still not pointing at 0.3-dev. Also, you need DuckDB 1.4-dev to check. We are going to push 0.3-dev to core nightly...

We have had a similar issue in the past and it was actually not a DuckLake issue but an issue with the metadata within the parquet file! Could you provide...

I will try to create a reproducer with a spark generated parquet file to make this a more "generic" issue. It may be that the code path for `add_data_files` expects...

It queries what it needs with range requests! This is a very common pattern, particularly when there is only one writer and multiple readers!

This is not related to permissions, it is just not supported to pass an uri like `ducklake:...` as a database. The same happens with postgres or anything but duckdb database...

You are absolutely right (sorry for sounding like chatGPT)! I think the init operation does not finish before the attach starts (so maybe they are concurrent operations), which they should...

hey @hlinander I did manage to reproduce this. The thing is this is a bit of a hit and miss because you can always tune retry (https://ducklake.select/docs/stable/duckdb/usage/configuration) and sometimes the...

I meant right after you create the table (in your script this would be in the `setup_environment` function) however it doesn't matter because the issue is still valid! This *"as...

Hi @gsueur! Could you provide a concrete example pointing to an s3 dataset? (It can be public). Also provide the concrete minor version (have you tried this with 1.4.3 concretely?)....

@gsueur well to me that is the expected behaviour. A limit clause is not a sample, it is deterministic. In your example query the order by needs to happen before...