kakiuwang-ui commented on issue #16134:
URL: https://github.com/apache/datafusion/issues/16134#issuecomment-5664371906

   Correcting myself: I was wrong to say this should stay open. The remaining 
case is documented and deliberate, and I should have read the option's own 
documentation before posting.
   
   `enable_physical_uncorrelated_scalar_subquery`'s config description already 
states that turning it off "restores pre #21240 behavior, which silently 
produces incorrect results for multi-row subqueries", and the slt section 
covering that path says the same in more detail — "multi-row subqueries 
silently return wrong results ... Those cases are intentionally not covered 
here". So it is an accepted limitation of an escape hatch that is itself slated 
for removal, not an oversight.
   
   The facts in my previous comment still hold — the query does reproduce with 
the option off, by the mechanism I described — but the conclusion does not. 
@buraksenn and @u70b3 are right: the bug this issue reports is fixed on the 
supported path, and closing is appropriate. Sorry for the noise.
   
   The one thing that might still be worth a line: the existing coverage pins 
the cases where both paths agree, so nothing currently fails if the default 
path ever stops raising the cardinality error. `subquery.slt` does already 
assert it for SELECT, WHERE and function-argument positions, so that is covered 
— no action needed unless you want the option-off path pinned to its 
documented-wrong behaviour on purpose, which I would not recommend.


-- 
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