Ritchie Vink
Ritchie Vink
### Problem description evaluate `is_first` for `struct` dtypes.
The goal is to replace the current Arrow (Large)String type with a string type that allows a union between an inlined small string and an offset to a string that...
### Description This is probably the most executed query on large datasets. We materialize here where we shouldn't. For `parquet` and `ipc` we can use metadata. For `csv` and `(nd)json`...
Ensure that the parquet types always get converted to polars supported data-types. Before this, deeply nested datatypes were still of the wrong type, leading to unneeded casts. All code increase...
The great migration of the string/binary type. All preliminary work is done. This is the final push of #13459
We did a full scan per thread to do radix partitioning. This adds a repartition phase that scatters the keys. Will follow up with benchmarks.
:crossed_fingers: