andygrove commented on PR #5874: URL: https://github.com/apache/datafusion-comet/pull/5874#issuecomment-5719748423
Triage note: #5607 covers the same ground for `length` on `BinaryType` and reaches the opposite conclusion. Its current head benchmarked the codegen dispatcher as slower than plain Spark fallback for binary, so it keeps the `Unsupported` gate on all three of `CometLength`, `CometBitLength` and `CometOctetLength`, documents why, and adds a `CometBinaryLengthBenchmark` to back that up. This PR deletes the gate from `CometLength` by moving to `SparkLengthFunc`. They collide on `strings.scala`, on all three of the `length.sql` / `bit_length.sql` / `octet_length.sql` fixtures — where this PR asserts binary runs natively and #5607 asserts `expect_fallback` — and both add a file called `CometBinaryLengthBenchmark.scala`. Native should win if the kernel is Spark-compatible, which would leave #5607 as `bit_length` / `octet_length` plus the benchmark. @adibmbrk, could you two agree the ordering? -- 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]
