Nikolas Schröter

@LFDanLu I will try to tackle this issue while working on #7240 👍 I don't think introducing another prop for this is the right way to go - instead we...

+1 I would love to be able to prune extraneous packages, especially in a workspace 👍

@snowystinger Gut feeling is probably the [style implementation](https://github.com/adobe/react-spectrum/blob/075f2edf057c6c20fd03d5449528ed65d10e6521/packages/%40react-aria/collections/src/Document.ts#L364) of `ElementNode` not having been adjusted for `Activity` also applying `display: none`, causing it to think we are in a suspended render.

@chirokas Hm, do you know why `setProps` isn't called in the mount phase? Docs for `Activity` mention effects not being mounted, but `setProps` is invoked in a callback ref.

@devongovett I'm not sure I understand why DOM nodes are not being mounted. I thought the point of `Activity` was that the [DOM is still present to later be restored](https://react.dev/reference/react/Activity#restoring-the-dom-of-hidden-components)?

@chirokas Thanks for digging that up! I see that it's basically just an `OffscreenComponent` under the hood, so equivalent of a component entering `Suspense` with the intention of reusing the...

@raix The original reason for injecting a style wasn't performance related but rather to address https://github.com/adobe/react-spectrum/issues/7993, which can only be solved by styling or injecting a meta tag.

@plesiecki Is there a specific reason why you are using `@testing-library/react` over Vitest's interactivity api in browser mode?

@plesiecki Ok, then as a workaround, you can also use keyboard interactions from `vitest/browser` instead of a firing pointer events. The PR is nonethless still worth considering 👍

This should be using `useKeyboard` instead of native event listeners, otherwise we leak the event upwards.