Cory Grinstead

Are you able to share minimal `.parquet` file that produces this bug?

@ritchie46 This seems like a bug in`df.rechunk()` I'm not familiar with the logic that determines when it `should_rechunk`, but for this specific dataset, `should_rechunk` returns `false` so the `rechunk` doesn't...

@seongs1024 as a workaround, you can use `as_single_chunk_par` or `as_single_chunk` before writing to json. ```rust fn convert_df_to_json_data(df: &mut DataFrame) -> String { df.as_single_chunk_par(); let mut buffer = Vec::new(); JsonWriter::new(&mut buffer)...

+1 to this. The node bindings support this syntax for select, groupby, .... It seems like a nice way to remove a little bit of boilerplate.

@lucazanna we do support window functions `sum(x) over (partition by y)`. If there is anything that specifically doesn't work, please provide some sql examples & we can look at adding....

@ritchie46, I've been thinking about #8425, and wanted to get your thoughts on the direction you'd like to see out of polars sql support. Maybe this is a bit premature,...

Makes sense, Wasn't sure if there was a clear direction you wanted to head in. I guess it makes sense to do the first one before trying to implement ADBC,...

I've been mostly doing them when issues for specific feature requests come in. I updated the OP with a **COMMUNITY REQUESTED** section that I'll use as the highest priority features....

That makes sense. I'll give that a try. Thanks @ritchie46.

The 'collect_all' function does not modify the plans at all. It simply collects the frames in parallel, so it makes sense that it'd be slower as it now needs to...