Github user mgaido91 commented on the issue:
https://github.com/apache/spark/pull/22641
@dilipbiswal you're perfectly right. The point is: let's assume we
introduce a bug which causes returning always SNAPPY. Your test will pass about
2/3 times out of 6. So it is quite likely that the test automation would fail
detecting the issue.
In the case you referenced, I was taking 50 out of 600 possible values. 50
is a number quite high anyway, so it enforces well enough that we support
multiple combinations. Here the cardinality is much lower, so we are not
enforcing that we are generalizing well.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
