asubiotto opened a new pull request, #27402: URL: https://github.com/apache/datafusion/pull/27402
## Which issue does this PR close? - Closes #27376. ## Rationale for this change `LIKE`/`ILIKE` on run-end encoded Utf8 columns casts them to `Utf8`, which expands every run. Dictionary-encoded Utf8 columns already skip this cast. With apache/arrow-rs#12154 the Arrow kernels match each run once, so run-end encoded columns can skip the cast too. Draft until apache/arrow-rs#12154 merges and DataFusion upgrades to an Arrow release that includes it. ## What changes are included in this PR? Type coercion and `LikeExpr` keep an outer `RunEndEncoded` over `Utf8` or `Dictionary(_, Utf8)` uncast when the pattern coerces to `Utf8`. Arrow requires the pattern to have the same type as the values. Other layouts and `SIMILAR TO` still cast. ## What is the testing strategy for this PR? New cases in `regexp/regexp_like.slt` check results and plans for LIKE/ILIKE on `REE<Utf8>` and `REE<Dictionary<Int32, Utf8>>`, and check that other layouts still cast. ## Are there any user-facing changes? No. Plans for these columns no longer show a `CAST`. AI assistance was used for this PR. I reviewed the whole change before opening 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
