imo the resizing test is not useful because it's a useful test of a common operation that needs to be optimized, but because it flexes on every major subsystem of the GUI framework.

Another example is startup time. Time to first frame on screen should be less than 20ms. That doesn't mean time until first content is rendered, but time until _all_ content is rendered (loading dialogs, placeholders, etc are better than nothing but entirely miss the point of being fast).

The second example is why even though I understand why developers pick tauri/electron/webviews/etc I can't get over how fucking slow the startup time is for my own work. None of them could show a blank window in under a second the last time I tried.