README.md in alephant-publisher-queue-2.3.1 vs README.md in alephant-publisher-queue-2.4.0

@@ -9,10 +9,11 @@
 - JRuby 1.7.8+
 - An AWS account, with:
   - S3 bucket.
   - SQS Queue.
   - Dynamo DB table.
+  - Elasticache (if using the "revalidate" pattern)
 
 ## Migrating from [Alephant::Publisher](https://github.com/BBC-News/alephant-publisher)
 
 Add the new gem in your Gemfile:
 
@@ -89,23 +90,23 @@
 
 ```
 {{ content }}
 ```
 
-## Usage
+## Usage (standard setup - non-revalidate)
 
 ```ruby
 require "alephant/logger"
 require "alephant/publisher/queue"
 
 module MyApp
   def self.run!
     loop do
       Alephant::Publisher::Queue.create(options).run!
-    rescue => e
-      Alephant::Logger.get_logger.warn "Error: #{e.message}"
     end
+  rescue => e
+    Alephant::Logger.get_logger.error "Error: #{e.message}"
   end
 
   private
 
   def self.options
@@ -147,9 +148,114 @@
 S3 Path:
 
 ```
 S3 / bucket-id / example-s3-path / renderer-id / foo / 7e0c33c476b1089500d5f172102ec03e / 1
 ```
+
+## Usage (revalidate pattern)
+
+```ruby
+require "addressable/uri"
+require "alephant/logger"
+require "alephant/publisher/queue"
+
+module MyApp
+  class UrlGenerator
+    class << self
+      # This function is called to generate the URL to be requested as
+      # part of the rendering process. The return must be a URL as a string.
+      def generate(opts)
+        "http://example.com/?#{url_params(opts)}"
+      end
+
+      private
+
+      def url_params(params_hash)
+        uri = Addressable::URI.new
+        uri.query_values = params_hash
+        uri.query
+      end
+    end
+  end
+
+  class HttpResponseProcessor
+    class << self
+      # This function is called upon a successful HTTP response.
+      #
+      # Use it to modify or process the response of your HTTP request
+      # as you please, but there is one rule - the return value MUST
+      # be a JSON object.
+      def process(opts, status, body)
+        # our response is already JSON, pass it through
+        body
+      end
+
+      # If you wish to vary your revalidate TTL on a per-endpoint (or
+      # other logic) basis, you can do it here - simply return an Integer
+      # value.
+      #
+      # If nil is returned the 'revalidate_cache_ttl' config setting on the
+      # broker will be used as the default TTL, otherwise the default in
+      # 'alephant-broker' will be used.
+      def self.ttl(opts)
+        # 30s revalidate time for all
+        30
+      end
+    end
+  end
+
+  def self.run!
+    loop do
+      Alephant::Publisher::Queue.create(options, processor).run!
+    end
+  rescue => e
+    Alephant::Logger.get_logger.error "Error: #{e.message}"
+  end
+
+  private
+
+  def self.processor
+    Alephant::Publisher::Queue::RevalidateProcessor.new(options, UrlGenerator, HttpResponseProcessor)
+  end
+
+  def self.options
+    Alephant::Publisher::Queue::Options.new.tap do |opts|
+      opts.add_queue(
+        :aws_account_id => 'example',
+        :sqs_queue_name => 'test_queue'
+      )
+      opts.add_writer(
+        :lookup_table_name    => 'lookup-dynamo-table',
+        :s3_bucket_id         => 'bucket-id',
+        :s3_object_path       => 'example-s3-path',
+        :view_path            => 'path/to/views'
+      )
+      opts.add_cache(
+        :elasticache_config_endpoint => 'example',
+        :elasticache_cache_version   => '100',
+        :revalidate_cache_ttl        => '30'
+      )
+    end
+  end
+end
+```
+
+Add a message to your SQS queue, with the following format (`id`, `batch_id`, `options`):
+
+```json
+{
+  "id": "renderer_id",
+  "batch_id": null,
+  "options": {
+    "id": "foo",
+    "type": "chart"
+  }
+}
+```
+
+This will then make a HTTP GET request to the configured endpoint (via `UrlGenerator`), process the response (via `HttpResponseProcessor`), render and store your content.
+
+You will not ordinarily need to push messages onto SQS manually, this will be handled via the broker in real use.
 
 ## Preview Server
 
 [Alephant Preview](https://github.com/BBC-News/alephant-preview) allows you to see the HTML generated by your templates, both standalone and in the context of a page.