README in memcached-1.2 vs README in memcached-1.2.1

@@ -52,14 +52,14 @@
   $cache.set 'test', value
   $cache.set 'test2', value
   $cache.get ['test', 'test2', 'missing']
     #=> {"test" => "hello", "test2" => "hello"}
 
-You can set a counter and increment it:
+You can set a counter and increment it. Note that you must initialize it with an integer, encoded as an unmarshalled ASCII string:
 
   start = 1
-  $cache.set 'counter', start, 0, false
+  $cache.set 'counter', start.to_s, 0, false
   $cache.increment 'counter' #=> 2
   $cache.increment 'counter' #=> 3
   $cache.get('counter', false).to_i #=> 3
 
 You can get some server stats:
@@ -71,14 +71,21 @@
   $cache.set 'test', nil
   $cache.get 'test' #=> nil
   $cache.delete 'test'
   $cache.get 'test' #=> raises Memcached::NotFound
 
-== Legacy applications
+== Pipelining
 
-There is a compatibility wrapper for legacy applications called Memcached::Rails.
+Pipelining updates is extremely effective in <b>memcached</b>, leading to more than 25x write throughput than the default settings. Use the following options to enable it:
 
+  :no_block => true,
+  :buffer_requests => true,
+  :noreply => true,
+  :binary_protocol => false
+
+Currently #append, #prepend, #set, and #delete are pipelined. Note that when you perform a read, all pending writes are flushed to the servers.
+
 == Threading
 
 <b>memcached</b> is threadsafe, but each thread requires its own Memcached instance. Create a global Memcached, and then call Memcached#clone each time you spawn a thread.
 
   thread = Thread.new do
@@ -88,9 +95,13 @@
     cache.get 'example'
   end
 
   # Join the thread so that exceptions don't get lost
   thread.join
+
+== Legacy applications
+
+There is a compatibility wrapper for legacy applications called Memcached::Rails.
 
 == Benchmarks
 
 <b>memcached</b>, correctly configured, is at least twice as fast as <b>memcache-client</b> and <b>dalli</b>. See BENCHMARKS[link:files/BENCHMARKS.html] for details.