wgtmac commented on PR #49855: URL: https://github.com/apache/arrow/pull/49855#issuecomment-4815100647
I made the option default off only to avoid changing the general `PreBuffer()` contract. Some callers may expect the cache to stay alive until the next `PreBuffer()` or reader destruction. For the Dataset `GetRecordBatchGenerator` path, I think we should enable this automatically when `pre_buffer=true`. That path reads row groups once, so it has a clear safe point to evict after each row group is decoded. -- 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]
