On Thu, 10 Mar 2022 22:21:28 GMT, Stuart Marks <sma...@openjdk.org> wrote:
> The difficulty with having a loop instead of constants is that the expected > value now needs to be computed. We could probably use `tableSizeFor` to do > this. But this is starting to get uncomfortably close to a testing > antipattern which is to use the code under test as part of the test. If a bug > is introduced into `tableSizeFor`, it could introduce the error into both the > actual value and the expected value, covering up the bug. (This is related to > one of the flaws with the Enum/ConstantDirectoryOptimalCapacity test.) Now we > _hope_ that `tableSizeFor` is correct and tested, but that verges on having > tests depend on each other in a subtle and uncomfortable way. you are correct about this. ------------- PR: https://git.openjdk.java.net/jdk/pull/7431