Mark
I've added the column key-value pairs to `parquet_metadata` in #9798
With the inclusion of #9793 and #9798 all Parquet meta-data should now be queryable using the following functions: | Function | Metadata | |-----------------------|--------------------------| | parquet_file_metadata | Top-Level File Meta...
This is implemented in #6387 now
Yes that sounds right, see [this commit](https://github.com/duckdb/duckdb/commit/0d8a0f6ba539d779b3069ec9b27662876e80cf49) that made the same change for the DELETE for the same reason/bug.
Can you try this in the latest DuckDB version?
Implemented now
I don't quite understand why the regular `min` and `max` aggregates would not work in your example? Finding the minimum value of `b` for a group `a` is exactly what...
You can use `max_by` or `min_by` for that, e.g.: ```sql select a, max(b), max_by(c, b), min_by(c, b) from foo group by a; ┌───────┬────────┬──────────────┬──────────────┐ │ a │ max(b) │ max_by(c, b)...
Unnest is usable in from clauses now, see #3580
I had a look at this, and this issue appears to come from Pandas itself. DuckDB converts to Pandas by creating masked numpy arrays, and then calling `pd.DataFrame.from_dict`, e.g.: ```sql...