web-platform-tests
Update accname/name/comp_host_language_label.html with hidden label tests
I'm a little confused about this one. I thought that there was existing, explicit language that a hidden label does not provide a valid name for an input, but these tests seem to suggest that browsers support the opposite. What am I missing here? I could use a brief chat on this one, I think.
Yes, and I am concerned that this diff greatly increases the Chrome/Gecko failure percentages of a file included in an Interop 2024 Focus Area. Perhaps we can start by adding any new tests that fail to a new .tentative file (so they don't affect the Interop score), and then consider the subtests one by one. In particular the promise rejections are likely b/c the test is calling for computedlabel on unrendered elements, which may be exposing a legitimate implementation detail that should not be tested in the meantime.
