Question

I was wondering what the differences are between using a Ruby gem to include a JavaScript library into a Rails project versus including the JavaScript file yourself.

I do know that you can lock down a version of the js library using bundler or alternatively, update all of your js libraries through bundler with one command instead of manually coping and pasting js files into your project. How much overhead does the Rails::Engine class add and is this the only way to get the asset files from a gem available in a Rails app? Are there any other notable differences that shows which way is “the best” way to include these libraries?

Answer 1

Great post from olaf_from_norweden on reddit with the answer:

Let’s walk through what the Sass port of the Twitter Bootstrap gem
actually does:

https://github.com/thomas-mcdonald/bootstrap-sass/blob/master/lib/bootstrap-sass.rb

When the bootstrap gem is evaluated on Rails initialization (when you
boot up the server), it uses require to load various parts of itself
depending on the environment.

If the Compass Sass framework gem is
loaded, it tells Compass the Rails asset folder paths to use.

If you’re using a Rails version with the asset pipeline, it loads an
empty Rails::Engine.

Engines are just Rails apps that you can mount
onto other Rails apps. When Rails encounters an engine during
initialization, it looks up the paths that the engine specifies to
load relevant files.

In this case, the gemspec tells Rails during initialization about the .scss/.png/.js files in its included vendor directory.

In development mode, Rails will recompile these assets on
every request along with the other assets in the main application.

In production, the bootstrap gem isn’t even loaded because it goes in the
:assets group of your Gemfile.

Hope that makes sense. Not the best or
most salient explanation. And even if the gem was loaded in
production, Rails doesn’t go hunting for assets directories unless
it’s compiling assets. So it doesn’t even matter.