alamb opened a new issue #1162:
URL: https://github.com/apache/arrow-datafusion/issues/1162


   **Is your feature request related to a problem or challenge? Please describe 
what you are trying to do.**
   Now that we have a nice ConstantPropagator and Simplifier framework, there 
are additional predicate rewrites we can do to improve query performance 
   
   **Describe the solution you'd like**
   Implement additional algebraic simplification / rewrite rules to 
[constant_folding.rs](https://github.com/apache/arrow-datafusion/blob/master/datafusion/src/optimizer/constant_folding.rs)
 after https://github.com/apache/arrow-datafusion/pull/1153 has been merged
   
   Types of standard rewrites I think it would be good to have:
   ```
   true OR <expr> --> true
   false OR <expr> --> expr
   true AND <expr> --> <expr>
   false AND <expr> --> false
   ```
   
   There is also an interesting one in 
https://github.com/apache/arrow-datafusion/pull/1153 that would rotatate expr 
trees to isolate volatile functions:
   ```
   (rand() + 1) + 2 --> rand() + (1 + 2)
   ``` 
   
   The reason for doing that rewrite is that then the constant evaluator could 
convert to `rand() + 3`
   
   
   **Describe alternatives you've considered**
   There are various automated rewriting rules such as egg / tokomak (drafted 
in https://github.com/apache/arrow-datafusion/pull/1066 and 
https://github.com/apache/arrow-datafusion/pull/441 by @Dandandan  and @pjmore 
) which might be better / more powerful than implementing our own rules 
   
   **Additional context**
   I am sure other query optimization systems have a good set of rewrite rules 
that we could investigate. I bet both postgres and spark have interesting 
optimizations to do.
   


-- 
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