DanielLeens commented on PR #12578: URL: https://github.com/apache/seatunnel/pull/12578#issuecomment-6414891410
Thanks @SEZ9 for going through this carefully. Answering both questions against the unchanged head `ef53a70c6f72` (code reading only, nothing executed locally): - **F4, cross-overload on the same row:** `testSetFieldInvalidatesCachedBytesSize` (`SeaTunnelRowTest.java:253`). Its first half warms the cache with `getBytesSize(rowType)`, calls `setField`, then asserts the recomputed value through the untyped `getBytesSize()` and again through the typed overload. Its second half does the reverse on a fresh row (untyped pre-warm, `setField`, typed read). - **F3/F5, the truncated sentence:** the end of @hutiefang76's bullet reads "...check that explicitly calling the parent setter invalidates the estimate afterward." That is what the tail of both limitation tests asserts: `testDirectFieldMutationDoesNotInvalidateCachedBytesSize` (`:280`) writes through `getFields()`, asserts the stale cached size is still returned, then calls `row.setField(0, row.getField(0))` (`:292`) and asserts the size is recomputed; `testNestedFieldMutationDoesNotInvalidateParentCachedBytesSize` (`:297`) mutates the child row, asserts the parent keeps its stale size, then calls `parent.setField(0, child)` (`:314`) and asserts the recomputed value. The production side is unchanged as well: `setField` assigns the field and only writes `size = 0` under `if (size != 0)` (`SeaTunnelRow.java:52-60`), with the contract in its Javadoc. For status, nothing new from my side. My source-level conclusion is unchanged (no code-side blocker) and the head has not moved. The remaining gate is the PR-bound check: the apache-side `Notify test workflow` is still failed and no Build is attached to this head, so a maintainer with write access needs to retry that notification workflow. The fork run (attempt 3) being green is useful evidence for this exact commit but does not replace it. -- 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]
