Github user zero323 commented on the issue:
https://github.com/apache/spark/pull/16535
@holdenk I admit this is probably not the most common issue :) but taken
together, all cases when we could fail fast, and instead we let it happen on
JVM or worker interpreter, significantly degrade user experience. Both Py4J
tracebacks and stuff we get from the workers, interleaved with many pages of
JVM junk, seem to be completely unreadable for many users.
So my opinion is that if we can fail fast (without triggering Spark job),
fail cheap (without significant runtime overhead or unnecessary limitations)
and provide meaningful feedback (if exactly know what type is expected) then we
should. I wouldn't bother with this alone, but if are already messing with UDF
it is just a good opportunity to make things a little bit better.
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
