Re: [PR] Restore lazy evaluation of fallible CASE [datafusion]

2025-03-25 Thread via GitHub
findepi merged PR #15390: URL: https://github.com/apache/datafusion/pull/15390 -- 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: github-unsubscr...@datafu

Re: [PR] Restore lazy evaluation of fallible CASE [datafusion]

2025-03-24 Thread via GitHub
alamb commented on PR #15390: URL: https://github.com/apache/datafusion/pull/15390#issuecomment-2749299083 Here are my benchmark results. My conclusion is no-discernable-difference ``` group findepi_lazy-case main -

Re: [PR] Restore lazy evaluation of fallible CASE [datafusion]

2025-03-24 Thread via GitHub
alamb commented on code in PR #15390: URL: https://github.com/apache/datafusion/pull/15390#discussion_r2010747252 ## datafusion/sqllogictest/test_files/case.slt: ## @@ -467,7 +467,18 @@ FROM t; [{foo: blarg}] +query II +SELECT v, CASE WHEN v != 0 THEN 10/v ELSE 42 END F

[PR] Restore lazy evaluation of fallible CASE [datafusion]

2025-03-24 Thread via GitHub
findepi opened a new pull request, #15390: URL: https://github.com/apache/datafusion/pull/15390 ## Which issue does this PR close? - Closes https://github.com/apache/datafusion/issues/15384 ## Rationale for this change Commit https://github.com/apache/datafusion/

Re: [PR] Restore lazy evaluation of fallible CASE [datafusion]

2025-03-24 Thread via GitHub
findepi commented on PR #15390: URL: https://github.com/apache/datafusion/pull/15390#issuecomment-2748049344 Alternatively, we could change `Literal::evaluate_selection` to return null scalar when selection is empty. However, for this to be effective, it would require adding evaluate_se