[codingthebeach]

Nice! I had fun with this, especially comparing Ember's half a meg footprint to Backbone's 20K. One helpful data point could be whether the library supports a (formal) custom build apparatus, since you can often get a sizable reduction that way. Full jQuery UI clocks in at ~242K, but using http://jqueryui.com/download/ you can often strip it down to 20K or less. Similarly with THREE.js and several others. Going in the other direction, a library that seems small (Backbone.js, @ ~20K) may actually end up being relatively heavy because of required external dependencies. A more explicit side-by-side "compare" function would also be nice here, comparing library size along with other metrics like script loader compatibility, maintenance frequency, different flavors of popularity, etc.

[yoran]

It's not very fair to compare Ember's size with Backbone's as Backbone implements just a very small subset of Ember's functionality.

Also, it seems that the sizes are outdated because the latest build of Ember.js is 478KB (see http://builds.emberjs.com/release/ember.min.js).

[stefanrush]

Thank you! I was actually unaware that Ember has separate versions for production and dev. I was using the dev version which I assume has extra features for debugging. I just updated it with the correct version. Sorry Ember...

[yoran]

No worries, thanks for correcting!

[stefanrush]

Thanks for the feedback. Great point on modular libraries. I definitely thought about that during development. I just couldn't find a good way to implement it. Some libraries like Modernizr have billions of possible permutations.

I also like the idea of comparing other metrics. Definitely something I'll look at going forward.

[Nemcue]

Ember is ~75k gzipped, so that comparison is moot.