paulk-asert commented on PR #3167: URL: https://github.com/apache/groovy/pull/3167#issuecomment-6601678212
Looked at this for alpha-1 too. The code is confined to `groovy-ginq-sql`, and I ran a differential check of the window functions against the collection provider on H2: `rowNumber`/`nthValue` 0-based, `ntile`, single-row `percentRank`, `lastValue` with the default frame, `lead`/`lag` with defaults, running and whole-partition aggregates, and `sum` with nulls all match. The only differences were numeric types (`double` vs `BigDecimal` for `avg`/`cumeDist`) and two cases where the collection provider itself throws (pre-existing, raised separately). Before merging: - Rebase once #3155 lands; the Windows red here is the post-build cache scrub (`rm: Device or resource busy`), fixed on master in 71552661eb, not a test failure (`BUILD SUCCESSFUL in 30m 5s`). macOS is the known stall. - The user guide still says "The main current restriction in strict mode is window functions" and lists window functions as a `mixedMode` example; this PR doesn't touch the `.adoc`. GEP-26's coverage and deferred tables also still list window functions as deferred. - Your description says auto-merge is paused until you say otherwise, so I'll leave the timing to you. From my side, with the docs updated it's fine for alpha-1. -- 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]
