peterxcli commented on PR #5771: URL: https://github.com/apache/datafusion-comet/pull/5771#issuecomment-5676851313
Thanks both. All three leftovers are fixed in b7e909bbf, docs and comment only — no runtime or benchmark change. - **Audit page, clock entries.** `hour`, `minute` and `second` now point at `benches/extract_date_part.rs` instead of the deleted `extract_clock_fields.rs`, and the stale "83-96% faster" is replaced with the matched result: every fast-path shape with a baseline of at least 2 us takes 82-94% less time, with offset session zones and dictionary input on the general path and within noise on a repeat. - **Audit page, weekday entries.** `dayofweek` and `weekday` no longer carry "8.4-10.2x". They now say about 9x on flat input with no or sparse nulls, 3-4x at 87.5% nulls, roughly 1.1x on an all-null batch or a cardinality-8 dictionary where the old path already did little work, and about 2x at cardinality 1024 — so the page matches the tables rather than the pre-fix headline. - **`datetime.scala`.** The two Sunday-is-0 comment lines left behind by removing `val DayOfWeek` are gone, so nothing sits above `val DayOfYear` any more. On CI: the new head re-runs the whole workflow, which should get past the expired `native-lib-linux` artifact that stopped the September 14 scan rerun before any tests ran. On the Spark 3.4 `[scans]` crash itself, I'm going on Andy's triage that it is the #5023 libhdfs SIGSEGV. As sunchao noted, the surviving artifact shows a native-thread SIGSEGV at address zero but doesn't name `hdfsThreadDestructor`, and I haven't independently confirmed the cause. If it recurs on this run, I'll attach the crash log to #5023 rather than work around it here. I left the branch at its current merge base rather than pull in `main` again, since it is mergeable and I wanted this push to be only the doc and comment fix. -- 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]
