andygrove opened a new issue, #4835: URL: https://github.com/apache/datafusion-comet/issues/4835
`first_value` / `last_value` over a `RANGE` frame that uses a literal `PRECEDING`/`FOLLOWING` offset fall back to Spark today. See `spark/src/main/scala/org/apache/spark/sql/comet/CometWindowExec.scala` (around lines 404-420), which records "FIRST_VALUE with RANGE frame and literal offset is not supported" / "LAST_VALUE with RANGE frame and literal offset is not supported". Comet's stateless native RANGE frame cannot reproduce Spark's monotonic lower-bound iterator semantics under non-ANSI overflow, so it falls back rather than risk an incorrect result. Documented as a current limitation in the window function compatibility guide. -- 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]
