RE: Focus visible - "unobscure" bullet

Hi everyone,

I can provide a couple of clarifications:

> Firefox suffering from the focus fully disappearing behind sticky content

In the example, I used scroll-padding to tell the browser to include the sticky header/footer in the focus calculation. Firefox hadn't implemented that, but it should be in version 79:
https://bugzilla.mozilla.org/show_bug.cgi?id=1642922

So I think the method will be cross browser by the time 2.2 is published, BUT, it is not a 100% reliable method.

If there is text in the sticky elements that wraps (e.g. with text spacing), the sticky element could get taller without the scroll-padding changing.


> To then also apply the first bullet (minimum area) and conclude that content fails as there are situations where only a part of a thin focus outline is visible is not something I would expect the average reader of such an SC to do.

There is a nuance here, which is that the focus indicator needs to meet bullets 1-3 regardless of position / scrolling etc. The 4th bullet is for the element, which should be at least partially visible. I.e. a difference between element & focus indicator.

Overall, I *think* that would:

  *   Catch examples where the element is completely obscured (e.g. example 1).
  *   Not catch examples where the whole element is partially obscured (e.g. example 2).
  *   Catch examples where the focus indicator is obscured (e.g. example 3)

We can provide info in the understanding on that aspect, first we need to decide if it's a good idea.


> you would find your meet the SC in some browsers all the time, in other browsers only at certain viewport widths, in still others (Firefox) not at all.

To me that is an argument for considering the element & focus indicator as a unit and ignoring the temporary aspects such as scrolling.


> If the 8px along shortest edge is the chosen method, and if that part of the component is completely obscured, the way that the 4th bullet is written means that this scenario passes.

That's a good point, focus indicators which don't outline or include the whole element are going to suffer from odd edge cases, which we could compound with allowing a partially-showing bullet.

Cheers,

-Alastair