lidavidm opened a new pull request #10987:
URL: https://github.com/apache/arrow/pull/10987


   This removes fill_null in favor of coalesce. 
   
   A few points:
   - This leaves behind the Python aliases for convenience.
   - This is about 25% slower than fill_null for the (Array, Scalar) case for 
fixed-width types. As far as I can see, this is mostly because the handwritten 
fill_null only unboxes the scalar once (and does so cheaply), while the generic 
implementation here unboxes the scalar on every loop (the compiler doesn't 
manage to hoist it) *and* UnboxScalar is relatively expensive (requiring a 
virtual call). I tried to make UnboxScalar less expensive but couldn't avoid 
undefined behavior/incorrect results for the DayTimeIntervalType case.
   - This specializes all the two-argument cases for fixed-width types but it 
might not really be worth it except for the (Array, Scalar) case.


-- 
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]


Reply via email to