Tushar7012 commented on issue #20052: URL: https://github.com/apache/datafusion/issues/20052#issuecomment-3819074890
Hi @alamb , I attempted to reproduce the extended test slowdown locally, but the full run fails during compilation due to Windows paging/memory limits, so I could not complete a full extended test run. CI is the intended source of timing data, but the relevant before/after logs around PR #19674 have expired, so compile vs test runtime separation is not currently measurable. Based on the scope of PR #19674, there are two plausible contributors. First, extended tests are likely more affected because they include broader SQL logic and coercion coverage, and this PR added or expanded several struct-heavy SQL logic tests. Second, name-based struct casting and centralized compatibility validation introduce additional validation work per cast, which may accumulate in struct-heavy queries. These are hypotheses only; no regression is confirmed without fresh timing data. Proposed next steps would be to rerun the extended tests on fresh CI with logs preserved so we can extract compile vs test durations; optionally add lightweight phase timing to the extended workflow to capture build vs test time; and, if needed, rerun a smaller struct-heavy subset of extended tests to help isolate impact. If there is a preferred approach or existing timing data I should use, please let me know. Happy to help drive whichever next step makes the most sense. -- 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]
