You mean by piling up <div>, <span>, CSS and JavaScript?

Yes, why not? That's no different than piling objects in a GUI widget library to build a tree structure for the layout of a window; except that it has a human-readable string representation.

Sure there is, on a GUI toolkit I can always control at some level how those pixels are drawn, while on a browser I need to hope it does the right thing.

Also there is big performance impact, where piling up divs with CCS requires the right incantations, some browser version specific, for hardware acceleration to kick in.

And in any case, it is impossible to fully replicate the UX of the host platform.

> I can always control at some level how those pixels are drawn

How well does that work when your GUI must be shown in arbitrary devices, with arbitrary screen sizes and arbitrary resolutions? And how well does it reflow when the user changes the viewport size?

In any case, the original post to which I replied was comparing HTML5 to the NeWS platform, which provided a unified UI stack itself, so you wouldn't be able to match the native behavior following that approach either.

That is what layout managers and logical pixels are for.

You are missing the fact that NeWS offered this kind of control.

"The NeWS Book" - https://archive.org/details/bitsavers_sunNeWSThe_11666863