Github user rdblue commented on the issue:
https://github.com/apache/spark/pull/20647
> I've rolled back unnecessary style-only changes, but leave the one that
clean up unused imports.
Sounds good. Thank you!
> If we think this is the right equality for DataSourceV2Relation, it
should also be the right equality for scan and streaming relation. They should
be consistent.
I think it is closer to the right equality, but the change relies on the
equality of the reader and not of the configuration for that reader. I think it
should rely on the reader configuration and not the reader itself, which is
unreliable. There is no requirement that a reader implements an equals method
that is appropriate for use in planning and we use equals to determine whether
rules have reached a fixed-point.
This is probably fine for the scan, but I think it isn't right for the
streaming relation. Since it is more correct, lets go with it for this commit.
If you don't mind, lets open a follow-up issue to get equality right. I'm happy
to if you want.
Overall, I just had a couple more comments and then I think this is ready
to go.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
