Garrett LeSage

> which we feel also helps alleviate an accessibility issue Right, which is one of the several reasons why I strongly feel that a proper fix is to not have...

Still not stale, still a regression, still not fixed in PF5 (where it was introduced in a minor version on an update) and even seems to be a problem in...

> This seems to be really hard to achieve. :cry: Hopefully PatternFly will do something on their end at some point in time. They have an open issue on it...

In-page modals seem fine, however:   So this seems shell-specific.

There's some specific text that's way off the page to the right, which causes the width of the pf-v5-c-backdrop to be much wider than the viewport, which causes this offset...

Ah, the problem is this:  It uses negative margins, and since the right/left was changed to inline-start/inline-end, it must've changed the intend and is stretched in both ways instead...

1. Why are we implementing our own .screenreader-text instead of using https://www.patternfly.org/utility-classes/accessibility/#screen-reader-only? (Answer: We have a focus style to show it too.) 2. Also, more importantly, why are we implementing...

Also: Why did our pixel tests not pick this up at all? I thought we tested these modals and on RTL as well?

This breaks basic functionality in Cockpit (as it moves content offscreen, including switching languages), and it should be simple to fix, so I'm marking it as a release blocker.

I can't reproduce the issues now. We will probably still want to replace it with the PF skiplink, but everything seems better now as of #20561.