morningman commented on PR #69581: URL: https://github.com/apache/doris/pull/69581#issuecomment-6545357929
Addressed the tenth review round in `89afea9808c`. Per finding: - 4430998187 (P2, sliced-out Decimal256 map keys narrowed before compaction): not changed, see the thread. Entries outside a sliced map's offsets are ordinary values of the entries array, not values hidden by a NULL row, and Arrow's own `ValidateFull()` rejects an entries array whose key does not fit the declared precision, so no valid batch fails here. Map keys cannot be NULL and stay checked over the whole entries array, as the description states. - 4430998189 (P2, negative Lance cache sizes cast to uint64): fixed in `89afea9808c`. Session initialization rejects a negative index or metadata cache size, and, while the data cache is enabled, a negative disk capacity or read block size, before the cast, with an InvalidArgument that names the settings; only Lance scans fail, and a disabled data cache still ignores its sizes. Covered by `LanceSessionManagerTest.RejectsNegativeCacheSizesBeforeCreatingTheSession`. BE unit tests on macOS arm64 (ASAN): the `Lance*` suites, 107 pass; with both checks disabled only the new test fails. clang-format 16, clang-tidy on the changed lines and `check-build-hygiene.sh` pass. The two required checks that failed on the previous head are unrelated to this PR: BE UT stopped with a core dump in `TimeSharingTaskExecutorTest.test_invalid_task_handle_returns_error`, a master test that leaves the executor's thread counts uninitialized (fixed separately in #69915), and P0 failed only in `mtmv_p0.ivm.test_ivm_chained_mtmv_2` with an FE `/ by zero` while this PR changes no FE code. CI is running again on the new head. -- 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]
