ajpotts

I investigated this a bit. It seems like the best solution might be to have an `array_like` type variable, something like this: ``` array_like = Union[numeric_scalars, pdarray, Sequence[pdarray]] ``` and...

The script doesn't really need to be included. It's just to show how I identified the scalars. I could delete it if that is better.

I’d like to propose we combine two ideas: - `weakref.finalize` on the Python side - a simple session-ID mechanism on the server to make Arkouda’s memory cleanup reliable. The server...

I thought it could be triggered from the client side and wouldn't need that in the normal situation. If we want it to be resilient to improper disconnect then that...

I'm not sure that would work well with the data science community, especially with how jupyter notebooks work.

✅ Functions that do have multi-dim tests - utils.shape → has explicit 2-D test. - Indirect: broadcast_dims is exercised by other N-D tests, but not directly. ❌ Functions with no...

util.py: invert_permutation, sparse_sum_help, and map need either guards (restrict to 1-D) or tests for N-D input. Other functions are fine (either ndim-agnostic or already tested). utils.py: Already compliant: supports N-D...

❌ Functions/methods with no direct unit tests - from_return_msg (server-return JSON parsing) - nbytes property - to_ndarray / tolist (conversion functions aren’t explicitly tested, though some doctest coverage exists) -...

In client_dtypes.py 1. BitVector Current: __init__ takes any pdarray of ints and forwards values.ndim, values.shape up to ClientDType. Problem: Will happily accept 2-D/3-D pdarray. Tests: client_dtypes_test.py only covers 1-D creation,...

Even stranger: ```python In [11]: t1 = ak.ones(5, dtype=str) In [12]: t2 = ak.array(['1', '1', '1', '1', '1']) In [13]: t1[0] Out[13]: np.str_('1') In [14]: t2[0] Out[14]: np.str_('1') In [15]:...