Bug 16966 – Add benchmark/tests for querySelectorAll

• Component: WebKit Website
• Status: RESOLVED FIXED

[David Smith]

weinig and I still owe a blog post on qSA, and everyone loves easy to interpret benchmark charts, so I modified the mootools test/benchmark in the following ways:

1) more iterations, so we can actually see variance
2) try/catch around each test so that one exception won't break every test after it
3) removed tests of selectors that webkit doesn't implement yet, so as to not throw off the resulting total
4) added a column for the native querySelectorAll implementation

[David Smith]

Created attachment 18598 [details]
the aforementioned benchmark

[David Smith]

Adding a URL to a copy of it on my site.

[Adam Roben (:aroben)]

Comment on attachment 18598 [details]
the aforementioned benchmark

What's the license covering the SlickSpeed benchmark? I think we'd need a LICENSE file in the slickspeed directory making this clear.

I also wonder whether this would be better placed in the WebKitTools directory?

[Darin Adler]

Comment on attachment 18598 [details]
the aforementioned benchmark

Seems OK to put this on the website.

It'd be a little better if there was something that indicated what was changed from the original slickspeed test. Presumably there's a reason we're hosting our own copy and it would be best to be clear on why.

[David Smith]

(In reply to comment #4)
> (From update of attachment 18598 [details] [edit])
> Seems OK to put this on the website.
> 
> It'd be a little better if there was something that indicated what was changed
> from the original slickspeed test. Presumably there's a reason we're hosting
> our own copy and it would be best to be clear on why.
> 

Sure. I'll add an explanatory paragraph about the changes before committing.