Github user srowen commented on the issue:
https://github.com/apache/spark/pull/22641
-1 We definitely can't just randomly test a subset of cases we need to test
in order to make things faster. Worse, it makes the failure nondeterministic.
However, if there's a good argument that some combinations that are tested
really are superfluous, let's make a change that avoids those extra tests.
For example: is there a good argument that testing combinations that cover
all options at least once, instead of testing every combination, is probably
sufficient?
That is, if I want to test combinations of options A, B, C and x, y, then
is it probably sufficient here to test, say, (A,x), (B,x), (C,y) rather than
all 6 possibilities?
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
