Github user HyukjinKwon commented on the issue:
https://github.com/apache/spark/pull/20959
For usability, the workaround I suggested above has more flexibility. For
example, we can make different operation (e.g, filter) on schema inference
path. They are only few lines.
Schema inference is discouraged in production line. I believe, for example,
just taking 100 records and use the schema makes more sense.
I am not against this option but I am saying I don't feel strong on this
for above reasons.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
