Ritchie Vink

Binary search on sorted arrays, similar to numpy `searchsorted`.

Add streaming converters

3

Now that we can memory map larger than RAM ipc/feather files, it would be great if we could convert datasets in a streaming fashion. I'd want to start with: *...

Simply access the public attributes.

perf(python,rust): Use faster compression backend for parquet

3

This uses `zlib-ng` and `lz4_flex` as backend in favor of the defaults. #4259

See: https://github.com/pola-rs/polars/issues/3974#issuecomment-1197566767

```python df = pl.DataFrame({ "cars": ["audi", "vw", None], "animals": ["cat", "dog", None], }) df.select([ pl.all(pl.col(pl.Utf8) ]) ``` ``` --------------------------------------------------------------------------- DuplicateError Traceback (most recent call last) Input In [17], in ()...

We should make the list iterator indexable and then split `arr.unique` on all threads.

github actions: run action if the spec file has changed

1

This can prevent reverts on master if the spec turns out to be invalid.

Python: deprecate most indexing operations.

3

Which ones should we still allow?

implement Arrow2's odbc reader and writers

6

We now have native ODBC support upstream. This has to be exposed in polars similarly to existing IO readers and writers.