[C++][Compute] `true_unless_null` kernel output incorrect when run-end encoded array contains null

[pitrou]

Describe the bug, including details regarding any error messages, version, and platform.

See this snippet:

>>> a = pa.array([11,11,11,42,42,5,None])
>>> r = pc.run_end_encode(a)

>>> pc.true_unless_null(a)
<pyarrow.lib.BooleanArray object at 0x7fa1fc54f5e0>
[
  true,
  true,
  true,
  true,
  true,
  true,
  null
]

>>> pc.true_unless_null(r)
<pyarrow.lib.BooleanArray object at 0x7fa1fc4cab60>
[
  true,
  true,
  true,
  true,
  true,
  true,
  true
]

[var-nan]

Hi @pitrou, I can give it a try. Can you assign this issue to me?

[pitrou]

@var-nan Looking forward to your PR!

[var-nan]

Hi @pitrou , I have fixed the bug and tested the code with some manual test cases in a main function.

But I'm having some trouble writing the test case for this fix. Below is the unit test code. The unit test code is crashing (with segmentation fault) in ScalarExecutor::EmitResult() after executing the current span. I'd appreciate any help.

// in scalar_validity_test.cc 
TEST_F(TestBooleanValidityKernels, TrueUnlessNullOnREE) {
  auto run_ends = ArrayFromJSON(arrow::int32(), "[1, 4, 9, 10]");
  auto values = ArrayFromJSON(arrow::int32(), "[1,null,4,null]");
  EXPECT_OK_AND_ASSIGN(auto z, RunEndEncodedArray::Make(10, run_ends, values));
  auto expected = ArrayFromJSON(arrow::Boolean(), "[true, null, null, null, true,true,true,true,true,null]");
  CheckScalarUnary("true_unless_null", z, expected);
}

[pitrou]

Issue resolved by pull request 50269
#50269

[pitrou]

@hadrian-reppas Can you post here so that we can assign this issue to you?

[hadrian-reppas]

Here you go!