Page caching is an approach to caching where the entire action output of is stored as a HTML file that the web server can serve without going through Action Pack. This is the fastest way to cache your content as opposed to going dynamically through the process of generating the content. Unfortunately, this incredible speed-up is only available to stateless pages where all visitors are treated the same. Content management systems – including weblogs and wikis – have many pages that are a great fit for this approach, but account-based systems where people log in and manipulate their own data are often less likely candidates.

Specifying which actions to cache is done through the caches_page class method:

class WeblogController < ActionController::Base
  caches_page :show, :new
end

This will generate cache files such as weblog/show/5.html and weblog/new.html, which match the URLs used that would normally trigger dynamic page generation. Page caching works by configuring a web server to first check for the existence of files on disk, and to serve them directly when found, without passing the request through to Action Pack. This is much faster than handling the full dynamic request in the usual way.

Expiration of the cache is handled by deleting the cached file, which results in a lazy regeneration approach where the cache is not restored before another hit is made against it. The API for doing so mimics the options from url_for and friends:

class WeblogController < ActionController::Base
  def update
    List.update(params[:list][:id], params[:list])
    expire_page :action => "show", :id => params[:list][:id]
    redirect_to :action => "show", :id => params[:list][:id]
  end
end

Additionally, you can expire caches using Sweepers that act on changes in the model to determine when a cache is supposed to be expired.

Namespace
Methods
C
E
Instance Public methods
cache_page(content = nil, options = nil, gzip = Zlib::BEST_COMPRESSION)

Manually cache the content in the key determined by options. If no content is provided, the contents of response.body is used. If no options are provided, the url of the current request being handled is used. Example:

cache_page "I'm the cached content", :controller => "lists", :action => "show"

Source: show | on GitHub 

# File actionpack/lib/action_controller/caching/pages.rb, line 166
def cache_page(content = nil, options = nil, gzip = Zlib::BEST_COMPRESSION)
  return unless self.class.perform_caching && caching_allowed?

  path = case options
    when Hash
      url_for(options.merge(:only_path => true, :format => params[:format]))
    when String
      options
    else
      request.path
  end

  if (type = Mime::LOOKUP[self.content_type]) && (type_symbol = type.symbol).present?
    extension = ".#{type_symbol}"
  end

  self.class.cache_page(content || response.body, path, extension, gzip)
end
expire_page(options = {})

Expires the page that was cached with the options as a key. Example:

expire_page :controller => "lists", :action => "show"

Source: show | on GitHub 

# File actionpack/lib/action_controller/caching/pages.rb, line 147
def expire_page(options = {})
  return unless self.class.perform_caching

  if options.is_a?(Hash)
    if options[:action].is_a?(Array)
      options[:action].each do |action|
        self.class.expire_page(url_for(options.merge(:only_path => true, :action => action)))
      end
    else
      self.class.expire_page(url_for(options.merge(:only_path => true)))
    end
  else
    self.class.expire_page(options)
  end
end