Implement the arr subnamespace

[etiennebacher]

Close #710

Still WIP:

• add tests
• double-check when strings are parsed as column names and when they're not as there are some inconsistencies with the list subnamespace
• cleanup docs (mostly for datatype)

[etiennebacher]

@eitsupi R CMD check fails for ubuntu/macos but not for Windows. I'm on windows for now, can you reproduce these panics on either ubuntu or macos? My bad, it fails everywhere but I still can't reproduce locally, can you try?

[eitsupi]

https://github.com/pola-rs/r-polars/actions/runs/7844014358/job/21405536688?pr=790#step:11:62

   thread '<unnamed>' panicked at /home/runner/.cargo/git/checkouts/polars-b0d90607192fd414/f3c4cc5/crates/polars-arrow/src/bitmap/immutable.rs:156:24:
  range end index 2 out of range for slice of length 1

It looks like a bug in Rust itself that has not yet been fixed in the latest release. rust-lang/rust#119352

[etiennebacher]

Given this leads to panicks (and segfaults if it runs twice), we should probably remove arr$min and arr$max for now and reimplement them when rust-polars 0.38.0 is released. Do you agree?

[eitsupi]

Do you agree?

Or I think it could be classified as nightly functions instead of doing a deletion.

r-polars/src/rust/Cargo.toml

Lines 11 to 17 in 025e3be

	[features]
	default = []
	full_features = ["simd", "sql", "disable_limit_max_threads"]
	disable_limit_max_threads = []
	simd = ["polars/simd"]
	sql = ["polars/sql"]
	rpolars_debug_print = []

This actually has nothing to do with simd, but could you make sure to check the simd feature anyway?
I would like to organize the features later (#300).

[etiennebacher]

Not sure I understood correctly, is 097da1a what you had in mind?

[eitsupi]

It is a little different; it does not need to be handled on the Rust side.
Please wait a bit while I fix it.

[etiennebacher]

Thanks for the fix, I'll try to finish this PR during the weekend

[eitsupi]

@etiennebacher Could you update the NEWS in a follow up PR?