jscheffl commented on PR #42410:
URL: https://github.com/apache/airflow/pull/42410#issuecomment-2451613605

   I am not a big fan of floats - maybe because in my job history I had always 
long long discussions about FMA and reproducibility of execution... but 
whatever.
   
   I would not block this but I still feel this is somewhat strange. integers 
have a very wide range in the default and we are talking about potentially a 
couple of thousand tasks we want to set priority. I don't see a point that  
there is a real "functional need" to set priorities exponential or in another 
manner that - with a normal modelling - yo need such large ranges. Anyway that 
would be hard to manage from a functional standpoint. Also I don't see a 
realistic setup where the priority weight strategy because of a very hghe long 
DAG would get to such boundaries.
   And the argument that at a certain level a big big float with +1 is the same 
big big float and no harm... then we could also figure out the smallest bound 
of INT supported by Postgres/MySQL/whatsoever DB and just cap the upper/lower 
boundaries to prevent a roll-over.
   
   It would be more convincing if there is a real problem/use case for such big 
priorities and not caused because DAG develops just add t00000000 many zeroes 
to the priority.


-- 
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