Github user viirya commented on the issue:
https://github.com/apache/spark/pull/18346
Thanks @dbtsai for the comment.
Yeah, I've also tried to let `CodegenFallback` evaluate all its children
under codegen path in parallel with this PR. It works.
Of course the advantage of the alternative approach is the performance.
I didn't submit that approach is because it adds additional APIs to
`CodegenFallback`. So it requires users to slightly re-write their
`CodegenFallback` based expressions to avoid this issue.
Another concern is we don't have too much codgen-only expressions in Spark,
most of them are object expressions. It might be rare to have codegen-only
expressions under `CodegenFallback`.
So I am not sure if it is worth to add APIs into `CodegenFallback`. If we
reach an agreement here, I will submit the alternative approach.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
