geopandas / pyogrio

Vectorized vector I/O using OGR
https://pyogrio.readthedocs.io
MIT License
272 stars 22 forks source link

Use Arrow pandas types by default when `use_arrow=True`? #303

Closed kylebarron closed 7 months ago

kylebarron commented 1 year ago

One question here is whether we want to set pandas to use the arrow types by default when use_arrow is True? In that case, since the data starts as arrow it would speed up the pandas conversion considerably

Thoughts either way? (I figured for visibility this should be its own issue)

Originally posted by @kylebarron in https://github.com/geopandas/pyogrio/issues/241#issuecomment-1741795111

jorisvandenbossche commented 1 year ago

My general stance is that we should follow the default data types as in pandas. And at the moment, the arrow dtypes are not the default in pandas, and so we should also not make them the default.

We could document better how you can easily obtain that yourself, though, now we have the keyword from https://github.com/geopandas/pyogrio/pull/273