xiangfu0 commented on PR #19836: URL: https://github.com/apache/pinot/pull/19836#issuecomment-6161290936
Yes. The first consumer is [StarTree #4759](https://github.com/startreedata/startree-pinot/pull/4759), stacked on [StarTree #4701](https://github.com/startreedata/startree-pinot/pull/4701): the Java-to-Arrow OBJECT row/result writers in `ArrowMseConversions` and `LiteralValueBatchAdaptor`. They call `getSerializedSize`, reserve the cell, then delegate payload encoding to the codec for Roaring, UltraLogLog and the four primitive lists; codecs returning `-1` keep the existing byte-array fallback. Multi-row columns reuse a borrowed buffer view, refresh it after Arrow growth and clear references in `finally`. Single-cell writes retain the existing uncached setter. The [benchmark](https://github.com/startreedata/startree-pinot/pull/4759#issuecomment-6118737941) is historical StarTree `f4a5094` / OSS `8a900a1` conversion/allocation evidence; it has not been rerun on the rebased heads and does not establish end-to-end query speedup. -- 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]
