Sunday, January 13, 2013
dojo-declare - performance
http://jsperf.com/dojo-declare
I would expect some slow down but not this big.
Need to investigate. Any collaboration appreciated.
Construction:
Sunday, January 6, 2013
Wednesday, January 2, 2013
dojo/dom-class vs. native classList
1) html 5 supports classList http://davidwalsh.name/classlist 2) dojo implementation is not using these native methods, and is slow compared to native variants: http://jsperf.com/native-dojo-class-manipulation-v-html5/7 Here is (my) draft of improvement: http://jsperf.com/native-dojo-class-manipulation-v-html5/8 (last method "Has Class - Dojo - with feature detection" and naive has(), contains1() methods) Native support detection issue is also mentioned in bug discussions (currently closed as 'rejected'): http://bugs.dojotoolkit.org/ticket/11465 More, it is marked as "posterity" in latest source code (line 43): https://github.com/dojo/dojo/blob/master/dom-class.js 3) Many HTML5 support detections are still missing in has.js: IMHO, about classList, only pull request from brianforbes, 2 years old (quite ugly): https://github.com/bryanforbes/has.js/commit/c2711e28989c0b838ea5d121915de3dad6952b0b For the beginning, we are comparing: Has Class - Dojo domClass.contains("single_class", "a_class"); Has Class - HTML5 document.getElementById("single_class").classList.contains("a_class"); classList contains with dojo byId dom.byId("single_class").classList.contains("a_class"); Has Class - Dojo - with feature detection domClass.contains1("single_class", "a_class"); http://jsperf.com/native-dojo-class-manipulation-v-html5/8
Subscribe to:
Posts (Atom)