alippai commented on issue #5942: URL: https://github.com/apache/arrow-datafusion/issues/5942#issuecomment-1507478686
> #5997 should help with this, the file range logic was partitioning row groups based on the location of their ColumnMetadata, which is normally written at the end of a ColumnChunk, not their actual data. This causes issues because duckdb appears to not write the un-inlined metadata and instead just writes a file offset of 0, causing all the row groups to be placed in the first partition. > > Technically this is probably not spec-compliant of DuckDB, but it's more reliable for us to partition based on where the actual page data is anyway. @Mytherin maybe this is interesting for you -- 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]
