Kyle Barron

Should be wednesday: https://github.com/PyO3/rust-numpy/pull/453

One interesting option is something like https://github.com/rouault/gdal/commit/422220b599546b00dc0860cf59c2493fe5fd6b4a where you can annotate a PR title with `[no ci]` to avoid running CI checks. Sometimes I like having CI run on draft...

Should probably verify that `.free` does in fact work: https://github.com/kylebarron/parquet-wasm/issues/591#issuecomment-2353772203

My personal view in all of these data type issues is that in the long term, it makes sense to adopt `use_arrow=True` as the standard, and deprecate `use_arrow=False`. For one,...

> But this can actually be controlled by a keyword Ah right. That reminded me of https://github.com/geopandas/pyogrio/issues/241#issuecomment-1508995869, so in theory you could use `types_mapper` here for the `pyarrow -> pandas`...

For multi-channel images (assuming each channel has the same data type and dimensions) you could represent that as an array with type [Fixed Shape Tensor](https://arrow.apache.org/docs/format/CanonicalExtensions.html#fixed-shape-tensor).

> what I don't see in a tensor is how to represent that in the PyCapsule interface, unless it's a nested set of fixed-size list (`+w`). Yeah that's it. Plus...

Arrow provides a space for arbitrary key-value metadata on each Field. This is often used by Arrow [extension types](https://arrow.apache.org/docs/format/Columnar.html#format-metadata-extension-types) in the `ARROW:extension:name` and `ARROW:extension:metadata` fields of that metadata, however you...

That depends on whether you want to add implementation-specific metadata or create a whole new _logical type_. I figure the former would be narrower and you could just document that...

In general this is a limitation of Parquet. JSON has three states: a valid value, `null`, and a missing/undefined key. Because Parquet is columnar, the third option does not exist...