tomz commented on PR #4187:
URL: 
https://github.com/apache/datafusion-comet/pull/4187#issuecomment-4373373885

   @coderfender yes, still reproducing on current `main` (`425f9c98`, post 
arrow bump).
   
   To verify, I checked out unmodified `main` and added only the 
`test_scalar_scalar_returns_scalar` test from this PR (no fix). It fails:
   
   ```
   test test_scalar_scalar_returns_scalar_repro_1615 ... FAILED
   panicked at wide_decimal_binary_expr.rs:
   BUG REPRO #1615: Scalar x Scalar returned Array, not Scalar
   ```
   
   `WideDecimalBinaryExpr::evaluate` on `main` still returns 
`ColumnarValue::Array` (length-1) when both inputs are `Scalar` — exactly the 
condition that triggers the q23 BHJ filter crash, since downstream comparisons 
against full batches then see two `Array` operands with mismatched lengths and 
arrow-ord rejects them with *"Cannot compare arrays of different lengths, got N 
vs 1"*. The arrow version bump didn't change this code path.
   
   With the fix in this PR applied, both new regression tests pass and q23 SF=1 
BHJ no longer crashes.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to