On Wed Nov 27 03:02:56 2024 +0000, Eugene McArdle wrote:
Would it be preferable to gate those tests behind a version check (similar to how it's done in [1])? That way the new tests are only run when we expect them to succeed, and it won't affect behaviour under older windows versions. @adamrehn for visibility [1] https://gitlab.winehq.org/wine/wine/-/blob/wine-9.22/dlls/advapi32/tests/reg...
preferred way is not to gate the tests with version... it's usually not that simple to create the relationship between the new feature and a version number (we don't have all possible windows version available under testbot) and to maintain it over time
in your case, as you want to play a number of test cases, perhaps the best way forward is to have a first test to detect if the new behavior is present, and if so, then run all the tests that require that new feature