For context on the SmoothTwoColumn test.

On this branch, the values argument passed into SmoothTwoColumn.evaluate_all

> /home/mike/workspace/datafusion-python/dev/python/tests/test_udwf.py(143)evaluate_all()
-> results = []
(Pdb) values
[<pyarrow.lib.Int64Array object at 0x7fff8ec772e0>
[
  0,
  1,
  2,
  3,
  4,
  5,
  6
]]

On main, it contains both a and b columns:

> /home/mike/workspace/datafusion-python/main/python/tests/test_udwf.py(143)evaluate_all()
-> values_a = values[0]
(Pdb) values
[<pyarrow.lib.Int64Array object at 0x7fffe8dcb700>
[
  0,
  1,
  2,
  3,
  4,
  5,
  6
], <pyarrow.lib.Int64Array object at 0x7fffe8dcbd60>
[
  7,
  4,
  3,
  8,
  9,
  1,
  6
]]
(Pdb) c

Do you need help troubleshooting the test?

@timsaucer - yes, if you wouldn't mind.

Here's where I'm at.

MulticolumnWindowUDF::partition_evaluator gets called with the correct arguments

RUST: evaluating partition_evaluator_args: PartitionEvaluatorArgs { input_exprs: [Column { name: "a", index: 0 }, Column { name: "b", index: 1 }], input_types: [Int64, Int64], is_reversed: false, ignore_nulls: false }

But then RustPartitionEvaluator::evaluate_all only receives the one array

RUST: evaluating evaluate_all with 1 value arrays: [PrimitiveArray<Int64>
[
  0,
  1,
  2,
  3,
  4,
  5,
  6,
]]

My next step would be to dig deeper into the upstream machinery, so I'd appreciate if you could take a quick sanity check before that.

Ok, I think this is ready to merge. @Michael-J-Ward do you want to look over the changes I added in to account for the new string views?

@timsaucer - thanks for finishing this up!

I think it's good to go. Especially since this 1st upgrade PR is primarily for the dependencies.