On Wed, 18 Mar 2026 18:33:40 GMT, Brian Burkhalter <[email protected]> wrote:
>> Change the `listRoots` sub-test on Windows to check for non-empty >> intersection of actual and expected root sets instead of equality of those >> sets. This is to address what appears to be transient mounting of logical >> drives during test execution. > > Brian Burkhalter has updated the pull request incrementally with two > additional commits since the last revision: > > - 8380204: Change a couple of variable names for clarity > - 8380204: Use File.listRoots and Stream.anyMatch I'm not convinced that the `listRoots` sub-test should not instead simply be removed from the test. On Unix, `listRoots` returns `/` and on Windows the value returned by `GetLogicalDrives`, and neither depends on the path string of the `File` as `listRoots` is static. So all the test is doing on Windows is comparing the result of `GetLogicalDrives` obtained at two different times. In the `ListRoots` test on the other hand, the `java.io` result is compared to the `java.nio.file` result. ------------- PR Comment: https://git.openjdk.org/jdk/pull/30282#issuecomment-4085137254
