The Cost of Frameworks


Hello, Paul Lewis wrote this great comparative analysis of JS Frameworks and Vanilla JS; The Cost of Frameworks. He used the TodoMVC as a baseline. I noticed he used Ember v1.10.0-beta.3 for his comparison. He said he used that instead of 2.20 because thats the latest thats on TodoMVC.

Do you think 2.20 would result in any better performance?


I don’t know about performance, but version 2.0 is 430kb (100kb once compressed).

I think comparing sizes without compressing is pointless, as it tends to penalize frameworks that use descriptive class/method names, which, although take more bytes, get compressed better. And in your production environment, you will always compress your code.

Also, comparing the sizes without comparing the features is kinda irrelevant. Of course Backbone is much smaller, but if want to do the same stuff you do with Ember or Angular, you need to add a lot of additional libraries.

On the other hand, he does have a point: if all you need is included in a small library such as Backbone, you might be better off using Backbone. When it comes to tools, “which tool is the best?” is not the right question. The right question is “which tool for which job?”

His final conclusion is pretty obvious: not using frameworks/libraries will always end up creating a more technically efficient result. This has been true since the very beginning of computing. We trade runtime efficiency for deadline efficiency. Otherwise, we would handcraft everything in assembly language from the ground up.


significantly better, but also I wouldn’t call that comparative analysis great, I’d call it mediocre.


Tom Dale wrote an excellent response: