I have done some researches but I couldn't find any benchmark that realistically test what real apps are doing. As mentioned in https://blog.mozilla.org/nnethercote/2012/08/24/octane-minus-v8/ we shouldn't be testing what we think is a representative of a real app; we should test the real app. Here, I propose to create a benchmark based on demo apps various JS frameworks have.
Created attachment 211971 [details] First cut
I think we need a better name! How about calling it DOMBench?
(In reply to comment #2) > I think we need a better name! Yes! > How about calling it DOMBench? That might be too generic?
(In reply to comment #3) > (In reply to comment #2) > > I think we need a better name! > > Yes! > > > How about calling it DOMBench? > > That might be too generic? But doesn't it describe what this benchmark is meant to do? If in the future we think it's not good enough, we can add things to it. Ideally, a DOM benchmark ought to measure how real programs use the DOM, JS, and other parts of the engine. This is what this benchmark aspires to do, and it feels like it does it better than other benchmarks. So, to me, DOMBench feels like an accurate description.
RS=me except for the name.
Comment on attachment 211971 [details] First cut RS=me except for the name. Land it after we have a better name. I recommend DOMBench.
DOMBench sounds good.
@rniwa - do we need this bug or Speedometer 3 takes care of it now? :-)
This is clearly fixed now.