nickva commented on PR #4984:
URL: https://github.com/apache/couchdb/pull/4984#issuecomment-1991921571

   > Mind we are switching to the discussion of the Mango rolling execution 
statistics here, which is only a possible application. Note that the 
execution_stats message is only used by Mango, but for some reason, it had to 
be driven through (woven into, pun intended) Fabric.
   
   It is related to correctness that's why I mentioned the other pr. I was just 
pointing out that based on the discussion in the previous pr, and new semantics 
of how on the coordinator we want to account only for rows which participate in 
emitting the response to the user. With that view, having separate messages vs 
one message can lead to unexpected behavior. Because it may be possible for 
another worker to insert their message in between the two messages when 
streaming is about to conclude. In general, there could be a good amount of 
difference between the work done prior to emitting each row sent to the client, 
and all the work done by each worker.
   
   > but for some reason, it had to be driven through (woven into, pun 
intended) Fabric.
   
   This is a side-effect of mango using fabric all_docs and view query as a 
layer. That extra handling of execution stats in all_docs and view query was 
another impetus for having a generic completion metadata. Though that would 
also require a bit more work to make sure it passes through the merge sort 
mechanism.


-- 
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: notifications-unsubscr...@couchdb.apache.org

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

Reply via email to