77483 2012-01-31 15:13:11 -0800 [Shadow]: Develop performance benchmark 2019-10-04 22:38:58 -0700 1 1 1 Unclassified WebKit DOM 528+ (Nightly build) Unspecified Unspecified RESOLVED LATER P2 Normal --- 102646 103016 1 dglazkov webkit-unassigned dominicc morrita rniwa oldest_to_newest 546547 0 dglazkov 2012-01-31 15:13:11 -0800 We must have an explicit performance goal of shadow DOM rendering being faster than doing the same thing using plain old DOM (even DOM mutation observers). This means we need to write a benchmark to prove it. Also, we should use this benchmark as an opportunity to track our progress. 546814 1 dominicc 2012-01-31 22:24:03 -0800 I think making shadow DOM perf neutral versus plain old DOM should be the goal. (Why would it get faster?) 547118 2 dglazkov 2012-02-01 09:16:14 -0800 (In reply to comment #1) > I think making shadow DOM perf neutral versus plain old DOM should be the goal. (Why would it get faster?) I would think that Plain-old DOM manipulation is going to be both: a) fairly involved and thus slower b) will have no benefit of just manipulating render boxes and thus slower, too. I think we should aim to be faster. No? 1577173 3 rniwa 2019-10-04 22:38:58 -0700 We'll probably do that as a part of new version of Speedometer.