Monica Dinculescu

ah, that is a big difference. maybe the js to css compile libs make sense instead then

Or just write CSS, who could think about that?

If we can write inline CSS that gets parsed out into "real CSS" by build tools then everyone wins.

That's a whole lot of tooling just to avoid hitting alt+tab.

What is the dev benefit of whiting in style att vs in a\j style sheet? Selectors are hard, or something else?

I write CSS separately and alt+tqb and whatever isn't a problem. What hate the most making class names :-) also error prone when in a team

I don’t really have a problem with CSS in particular, but componentizing everything is sooooo nice. avoids a lot of shared css mess

So this is benchmark testing for in-line styles vs. class names? Not 100% sure I understand what is being shown.

yup. http://foo.style .margin=2px (style prop setter) is way slower than adding a class `with-margin` that has the same style.

but, but , why?

//cc @rmondello

I just ran this benchmark with Safari 10.1 and the current Safari Technology Preview. STP narrows the gap by an order of magnitude.

I can't say when this improvement will make it to Safari proper, but improvements are coming.

I wonder how custom-elements and shadow-dom would effect the results.

With custom elements, you can use a template (and therefore a proper <style> node for styling), so it’s the same as the plain-dom world.

Nice! Also worth noting that many small classes (atomic css) has the best perf of all. Stuff like http://fela.js.org  is killer.

Don't have a proper bench, but did some testing in Chrome, and render and paint went down by half when using atomic css vs BEM-style css.