ehddnr301 opened a new issue, #74551: URL: https://github.com/apache/airflow/issues/74551
The shared prefix matcher derives an upper bound by incrementing a Unicode code point. That bound can exclude matching values when the database collation orders strings differently. For example, on PostgreSQL 16 with libc `en_US.utf8`, a prefix search for `카페` can omit `카페매출집계`. The existing substring-equality check does not help because the range predicates already exclude the row. A separate change is being prepared for Dag list display-name searches. The shared matcher and its other consumers, including XCom Dag display-name searches, still use the range predicates. This issue tracks that remaining scope. The shared fix should include an explicit libc-collation regression test, preserve existing input handling and collation equality semantics, and check the effect on indexed searches across supported databases. Simply dropping Unicode upper bounds can broaden index scans substantially. --- Drafted-by: Codex (GPT-6) (no human review before posting) -- 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]
