zanmato1984 commented on PR #45674:
URL: https://github.com/apache/arrow/pull/45674#issuecomment-2732336668

   > Upon further deliberation I have some doubts whether this should be public.
   > 
   > Correct me if I am wrong, but it seems to be kind of anit-pattern of acero 
used in asof_join and sorted_merge nodes tu use additional external thread to 
do the processing. The joining and and merging should ideally occur in 
InputReceived tasks.
   > 
   > In time series processing something like SerialSequencingQueue with 
backpressure would be better fit.
   > 
   > 
   
   I agree that it seems a bit anti-pattern of asof join and sorted merge for 
using an external thread to do the processing. However imho that's independent 
of whether the concurrent queue can be useful enough for implementing 
customized nodes to be made public. If this queue is necessary for your node, 
then we can/should do it. Or if you feel it is more reasonable and available to 
utilize other queue implementation, then we can defer/close this pr.
   
   No rush and no worry. Taking a step back of making something public is a 
good thing. And take your time to find the best way to implement your node. 
What do you think? @gitmodimo 


-- 
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: github-unsubscr...@arrow.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org

Reply via email to