adriangb commented on issue #20195: URL: https://github.com/apache/datafusion/issues/20195#issuecomment-4164457702
> [@gene-bordegaray](https://github.com/gene-bordegaray) asked me in the meeting why a CASE + hash was decided to route dynamic filters, [@adriangb](https://github.com/adriangb) you might remember better than me, but I think it was to avoid keeping a hash table with partition info + the whole build side hash table (which could cause memory issues). There were some discussions [here](https://github.com/apache/datafusion/pull/17529#discussion_r2353830615) and [here](https://github.com/apache/datafusion/issues/17171#issuecomment-3313002604) I guess the question depends on what the alternative is. As far as I can remember / think of the two alternatives would be to try every partition's hash table (poor build side performance and questionable probe side performance) or to make a combined hash table with all values (poor memory use). -- 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]
