[BUG-REPORT] DataFrame.astype(pyarrow.DictionaryArray(...)) is broken

[NickCrews]

Description I was trying to cast a DF to a pyarrow schema and ran across this issue. DataFrame[column].astype(pyarrow_type) works fine for many pyarrow types such as float, string, bool. But it doesn't work for pyarrow.lib.DictionaryType. However, if I use vaex's "wrapped" version, it works fine. I didn't explore other dtypes, but perhaps this also reveals a problem with other complex pyarrow types?

See the xfail-ing test PR

(PS is there hope for a future DataFrame.astype() similar to pandas? I'm writing this myself and feels like I'm reinventing the wheel.)

Software information

• Vaex version (import vaex; vaex.__version__): {'vaex': '4.10.0', 'vaex-core': '4.10.0', 'vaex-viz': '0.5.2', 'vaex-hdf5': '0.12.2', 'vaex-server': '0.8.1', 'vaex-astro': '0.9.1', 'vaex-jupyter': '0.8.0', 'vaex-ml': '0.18.0', 'vaex-graphql': '0.2.0'}
• Vaex was installed via: both in my application via pip, and on main in dev environment
• OS: MacOS

[maartenbreddels]

I like the idea of having a a dtype for the dataframe, we already kind of do checks for that in DataFrame.__array__, and we may possibly have some code for this in https://github.com/vaexio/vaex/pull/415 (where we want to know if a dataframe is of a homogenous type, so it can be treated as a 'matrix').

What should astype(some_dict_type) do actually?