jin-won-park opened a new pull request, #18923: URL: https://github.com/apache/iceberg/pull/18923
Truncate cuts strings to a width in code points, but the `startsWith` and `notStartsWith` projections compared the prefix's UTF-16 length with that width. With `truncate[2](s)`, `startsWith(s, "😀")` was projected to `s_trunc = "😀"`, which pruned the partition `"😀a"` even though its row `"😀abc"` matches. The strict projection of `notStartsWith` made the opposite mistake and accepted that partition. Both projections now compare code point counts. Closes #18922 Test plan: both predicates, inclusive and strict, for a prefix that is shorter than the width in code points but not in chars, and for one that matches the width in code points but is longer in chars. --- **AI Disclosure** - Model: Claude Opus 5.5 - Platform/Tool: Claude Code - Human Oversight: fully reviewed - Prompt Summary: Code reading of the partition transform code. The agent found this case and wrote the fix and tests; I reran the reproduction and reviewed the change. -- 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]
