review.state.done.PRs

Remove tabindex's synthetic click event feature

[OK] The idea of having a single point to register event-handlers when an element is activated, by any means (click, keyboard, other...) is great, but no browsers implemented it, so it was removed. Therefore this PR doesn't affect accessibility in practice. It could be good (as was expressed at the time) if such a feature were to be added.

Refine the inputmode attribute, and expose it on contenteditable

[OK] This PR is about making the `inputmode` attribute apply to any element that has `contenteditable` set—not just text input controls. Any accessibility concerns that applied to `inputmode` would apply here, but I don't see any extra concerns that would be brought about by adding this attribute to more elements (focusing etc. would still apply, but are out-of-scope). As browser support for `inputmode` is relatively new, I'll mention my thoughts on it: the software keyboard displayed would need to align with the user's locale and allow them to enter all needed characters, but this would affect all users, so is not accessibility-specific. Finally: there are some follow-up bugs about situations the user can get into where they don't have the ability to enter characters they might like to (#3478) and to request additional inputmodes (#4307). Again, those issues will affect all users who are using software keyboards, and therefore are not accessibility-specific.

Add the autocapitalize attribute

[OK] Whether the text being entered via the on-screen keyboard is capitalised or not is exposed to screen-readers through the keyboard, so this is accessible (it can also save all users time). The default iOS keyboard makes this apparent to users *before* they type a letter. The default Android keyboard makes this apparent to users *as* they type letters. Whilst this means that in some cases, users will not konw that capitalisation is going to occur when they type until they *have* typed, this still seems accessible because (1) the way the capitalisation is exposed to users is beyond the scope of the spec and (2) it's expected that web content will only set the attribute when it's generally expected to be of use and maybe even anticipated by the user.

Adjust the definition of the main element and various examples

[OK] This is a positive change for accessibility and was a step on the way to the next PR.

Restrict the main element to be used once per document

[OK] This is a positive change for accessibility; agreed by all of the screen-reader users who commented, and by several accessibility consultants.

Tweak how accesskey on legend works

[OK] This is about making an accesskey set on a legend take the user to the first interactive child of the legend, or of the parent fieldset (if applicable). This seems to align well with what the user is likely wanting to do.

Main part of the content has a dedicated element

[OK] This applies the changes in 1dec930 to an area that was previously missed.