Is there any summary of your results?

[smikula]

This is great — I'm working on improving styling perf right now and it's very helpful to understand what selectors are problematic. Right now it looks like I need to clone your repo and run the benchmarks myself. Do you have any summary of your results? It would be great to add that to the readme or elsewhere in the repo.

[ling1726]

Each scenario folder contains a write up of the results and impact of the selector being tested.

I deliberately did not attach profiles since it could change for different machines/browser versions.

Overall the conclusion would be that any selector with more than one combinator would have worse performance than applying a selector with a single css class