cleborys
commented
5 years ago Currently iterating over a list of drivers in each test. This is not ideal, as sometimes a single browser might fail (e.g. password length in Firefox vs Chrome), and then the tests should point at a specific browser.
There are solutions to create "test matrices" / "parametrized tests" (with different combinations of different parameters like which webdriver to use), but I need to look into that first.
It should be possible to write tests abstractly / independently of which browser is used in the webdriver and then pass the webdriver configuration as a fixture.
Ideally, there is an abstract test which is never run and several concrete tests inheriting from it that utilize different browser fixtures (chrome, firefox,...). These should be the tests that are actulaay run and it should be easy to see which browser fails, if only a specific browser fails.