Github user viirya commented on the issue:
https://github.com/apache/spark/pull/19938
They are possibly passed in split functions. But it is hard to image a case
we will change their values in the functions. In SparkSQL, the output from a
child operator are just used as input to evaluate new output in the parent
operator. We don't use the output as mutable statuses.
The possible problematic case is when we create a global variable in an
operator/expression codegen and use this global variable to carry mutable
status (e.g., the condition meeting status in casewhen) during the evaluation
of the op/expr. Then it is possibly we pass it and modify it in split functions.
If we don't change the values, seems to me this change just creates
redundant local variables.
This doesn't cause any harm at all. So I feel no strong option for this.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
