pitrou commented on issue #49715: URL: https://github.com/apache/arrow/issues/49715#issuecomment-4682682282
I remember @xhochy (or was it someone else?) pointing out that in many workloads, floating-point values are actually well-compressible using dictionary encoding. Ideally this would be adaptative, but that might require a significant architectural refactoring. One thing is that we have a max dictionary page size setting, but we could also have a max relative dictionary cardinality setting. The idea being that if the dictionary cardinality is greater than a given fraction (e.g. 10%) of the total number of encoded values, we would switch to PLAIN or BYTE_STREAM_SPLIT. -- 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]
