Comments on: JavaScript Performance: Comparing the Atlas and Prototype class idioms http://ajaxian.com/archives/javascript-performance-comparing-the-atlas-and-prototype-class-idioms Cleaning up the web with Ajax Thu, 17 May 2012 07:43:39 +0000 hourly 1 http://wordpress.org/?v=3.3.2 By: Alex Russell http://ajaxian.com/archives/javascript-performance-comparing-the-atlas-and-prototype-class-idioms/comment-page-1#comment-2077 Alex Russell Wed, 30 Nov -0001 00:00:00 +0000 http://test.ajaxian.com/archives/javascript-performance-comparing-the-atlas-and-prototype-class-idioms#comment-2077 This is a frustrating comparison. It appears to have been run inside a browser and not in the command line interpreters for each engine and so is not well isolated. I'm also not sure that it covers any of the important debates about writing quick code in a JS interpreter. The amount of time it takes to instantiate a class is a *pittance* compared to the memory cost difference between the tested approaches, and neither seem to actually hit the linearly-increasing object allocation speed issue on IE. In short, it's a blind benchmark of the wrong attributes of the wrong idioms. DOM performance, regexp speed, string manipulation, and memory size stats actually matter since they dwarf object instantiation cost. I'm afraid that this set of tests is merely a distraction. Regards This is a frustrating comparison.

It appears to have been run inside a browser and not in the command line interpreters for each engine and so is not well isolated.

I’m also not sure that it covers any of the important debates about writing quick code in a JS interpreter. The amount of time it takes to instantiate a class is a *pittance* compared to the memory cost difference between the tested approaches, and neither seem to actually hit the linearly-increasing object allocation speed issue on IE.

In short, it’s a blind benchmark of the wrong attributes of the wrong idioms. DOM performance, regexp speed, string manipulation, and memory size stats actually matter since they dwarf object instantiation cost. I’m afraid that this set of tests is merely a distraction.

Regards

]]>