dsander-memcache-client 1.7.7.pre

data/FAQ.rdoc

@@ -0,0 +1,31 @@
+= Memcache-client FAQ
+
+== Does memcache-client work with Ruby 1.9?
+
+Yes, Ruby 1.9 is supported.  The test suite should pass completely on 1.8.6 and 1.9.1.
+
+
+== I'm seeing "execution expired" or "time's up!" errors, what's that all about?
+
+memcache-client 1.6.x+ now has socket operations timed out by default.  This is to prevent
+the Ruby process from hanging if memcached or starling get into a bad state, which has been
+seen in production by both 37signals and FiveRuns.  The default timeout is 0.5 seconds, which
+should be more than enough time under normal circumstances.  It's possible to hit a storm of
+concurrent events which cause this timer to expire: a large Ruby VM can cause the GC to take
+a while, while also storing a large (500k-1MB value), for example.
+
+You can increase the timeout or disable them completely with the following configuration:
+
+  Rails:
+  config.cache_store = :mem_cache_store, 'server1', 'server2', { :timeout => nil } # no timeout
+
+  native:
+  MemCache.new ['server1', 'server2'], { :timeout => 1.0 } # 1 second timeout
+
+
+== Isn't Evan Weaver's memcached gem faster?
+
+The latest version of memcached-client is anywhere from 33% to 100% slower than memcached in various benchmarks.  Keep in mind this means that 10,000 get requests take 1.8 sec instead of 1.2 seconds.
+In practice, memcache-client is unlikely to be a bottleneck in your system but there is always going
+to be an overhead to pure Ruby.  memcache-client does have the advantage of built-in integration into 
+Rails and should work on non-MRI platforms: JRuby, MacRuby, etc.

data/History.rdoc

@@ -0,0 +1,241 @@
+= 1.7.6 (2009-11-03)
+
+* Reworked socket timeout code due to several customer complaints about timeouts not
+  working 100% of the time since 1.7.3.
+* Add option to configure the namespace separator string, for interop with Perl
+  which does not use a separator character:
+    MemCache.new(servers, :namespace_separator => '')
+* Move to jeweler and gemcutter for RubyGem support.
+
+= 1.7.5 (2009-09-09)
+
+* Fix ruby warnings (josh)
+* Make megabyte value size limit optional since Tokyo Tyrant can accept values larger than 1MB.
+  Use :check_size => false to disable the size check. (jsl)
+* Ruby 1.9 support for recent I/O changes.
+* Fix duplicate value marshalling on server error. (rajiv)
+* Added option :autofix_keys (disabled by default) to replace long keys with md5 hashes (sd)
+
+= 1.7.4 (2009-06-09)
+
+* Fix issue with raising timeout errors.
+
+= 1.7.3 (2009-06-06)
+
+* Remove SystemTimer support, refactor I/O to use nonblocking operations.  Speeds up
+  performance approx 100%. Timeouts basically have no overhead now!  (tenderlove)
+* Update load logic to support SystemTimer running in Ruby Enterprise Edition.  Thanks
+  to splattael on github for the comment.
+
+= 1.7.2 (2009-04-12)
+
+* Rollback socket timeout optimization.  It does not work on all operating systems
+  and was a support headache.
+
+= 1.7.1 (2009-03-28)
+
+* Performance optimizations: 
+  * Rely on higher performance operating system socket timeouts for low-level socket 
+    read/writes where possible, instead of the (slower) SystemTimer or (slowest,
+    unreliable) Timeout libraries.
+  * the native binary search is back!  The recent performance tuning made the binary search 
+    a bottleneck again so it had to return.  It uses RubyInline to compile the native extension and 
+    silently falls back to pure Ruby if anything fails.  Make sure you run:
+    `gem install RubyInline` if you want ultimate performance.
+  * the changes make memcache-client 100% faster than 1.7.0 in my performance test on Ruby 1.8.6: 
+    15 sec -> 8 sec.
+* Fix several logging issues.
+
+= 1.7.0 (2009-03-08)
+
+* Go through the memcached protocol document and implement any commands not already implemented:
+  - cas
+  - append
+  - prepend
+  - replace
+  
+  Append and prepend only work with raw data since it makes no sense to concatenate two Marshalled
+  values together.  The cas functionality should be considered a prototype.  Since I don't have an
+  application which uses +cas+, I'm not sure what semantic sugar the API should provide.  Should it
+  retry if the value was changed?  Should it massage the returned string into true/false?  Feedback
+  would be appreciated.
+
+* Add fetch method which provides a method very similar to ActiveSupport::Cache::Store#fetch,
+  basically a wrapper around get and add. (djanowski)
+
+* Implement the flush_all delay parameter, to allow a large memcached farm to be flushed gradually.
+
+* Implement the noreply flag, which tells memcached not to reply in operations which don't
+  need a reply, i.e. set/add/delete/flush_all.
+
+* The only known functionality not implemented anymore is the <flags> parameter to the storage
+  commands.  This would require modification of the API method signatures.  If someone can come
+  up with a clean way to implement it, I would be happy to consider including it.
+
+= 1.6.5 (2009-02-27)
+
+* Change memcache-client to multithreaded by default.  The mutex does not add significant
+  overhead and it is far too easy, now that Sinatra, Rails and Merb are all thread-safe, to
+  use memcache-client in a thread-unsafe manner.  Remove some unnecessary mutexing and add
+  a test to verify heavily multithreaded usage does not act unexpectedly.
+
+* Add optional support for the SystemTimer gem when running on Ruby 1.8.x.  This gem is
+  highly recommended - it ensures timeouts actually work and halves the overhead of using
+  timeouts.  Using this gem, Ruby 1.8.x is actually faster in my performance tests
+  than Ruby 1.9.x.  Just "gem install SystemTimer" and it should be picked up automatically.
+
+= 1.6.4 (2009-02-19)
+
+* Remove native code altogether.  The speedup was only 10% on Ruby 1.8.6 and did not work
+  on Ruby 1.9.1.
+
+* Removed memcache_util.rb from the distribution.  If you are using it, please copy the code
+  into your own project.  The file will live in the github repository for a few more months
+  for this purposes. http://github.com/mperham/memcache-client/raw/7a276089aa3c914e47e3960f9740ac7377204970/lib/memcache_util.rb
+
+* Roll continuum.rb into memcache.rb.  The project is again a single Ruby file, with no dependencies.
+
+= 1.6.3 (2009-02-14)
+
+* Remove gem native extension in preference to RubyInline.  This allows the gem to install
+and work on JRuby and Ruby 1.8.5 when the native code fails to compile.
+
+= 1.6.2 (2009-02-04)
+
+* Validate that values are less than one megabyte in size.
+
+* Refactor error handling in get_multi to handle server failures and return what values
+  we could successfully retrieve.
+
+* Add optional logging parameter for debugging and tracing.
+
+* First official release since 1.5.0.  Thanks to Eric Hodel for turning over the project to me!
+  New project home page: http://github.com/mperham/memcache-client
+
+= 1.6.1 (2009-01-28)
+
+* Add option to disable socket timeout support.  Socket timeout has a significant performance 
+  penalty (approx 3x slower than without in Ruby 1.8.6).  You can turn off the timeouts if you 
+  need absolute performance, but by default timeouts are enabled. The performance
+  penalty is much lower in Ruby 1.8.7, 1.9 and JRuby.  (mperham)
+
+* Add option to disable server failover.  Failover can lead to "split-brain" caches that 
+  return stale data. (mperham)
+
+* Implement continuum binary search in native code for performance reasons.  Pure ruby
+  is available for platforms like JRuby or Rubinius which can't use C extensions. (mperham)
+  
+* Fix #add with raw=true (iamaleksey)
+
+= 1.6.0
+
+* Implement a consistent hashing algorithm, as described in libketama.
+  This dramatically reduces the cost of adding or removing servers dynamically
+  as keys are much more likely to map to the same server.
+
+  Take a scenario where we add a fourth server.  With a naive modulo algorithm, about
+  25% of the keys will map to the same server.  In other words, 75% of your memcached
+  content suddenly becomes invalid.  With a consistent algorithm, 75% of the keys
+  will map to the same server as before - only 25% will be invalidated. (mperham)
+
+* Implement socket timeouts, should fix rare cases of very bad things happening
+  in production at 37signals and FiveRuns.  (jseirles)
+
+= 1.5.0.5
+
+* Remove native C CRC32_ITU_T extension in favor of Zlib's crc32 method.
+  memcache-client is now pure Ruby again and will work with JRuby and Rubinius.
+
+= 1.5.0.4
+
+* Get test suite working again (packagethief)
+* Ruby 1.9 compatiblity fixes (packagethief, mperham)
+* Consistently return server responses and check for errors (packagethief)
+* Properly calculate CRC in Ruby 1.9 strings (mperham)
+* Drop rspec in favor of test/unit, for 1.9 compat (mperham)
+
+= 1.5.0.3 (FiveRuns fork)
+
+* Integrated ITU-T CRC32 operation in native C extension for speed.  Thanks to Justin Balthrop!
+
+= 1.5.0.2 (FiveRuns fork)
+
+* Add support for seamless failover between servers.  If one server connection dies,
+the client will retry the operation on another server before giving up.
+
+* Merge Will Bryant's socket retry patch.
+http://willbryant.net/software/2007/12/21/ruby-memcache-client-reconnect-and-retry
+
+= 1.5.0.1 (FiveRuns fork)
+
+* Fix set not handling client disconnects.
+http://dev.twitter.com/2008/02/solving-case-of-missing-updates.html
+
+= 1.5.0
+
+* Add MemCache#flush_all command.  Patch #13019 and bug #10503.  Patches
+  submitted by Sebastian Delmont and Rick Olson.
+* Type-cast data returned by MemCache#stats.  Patch #10505 submitted by
+  Sebastian Delmont.
+
+= 1.4.0
+
+* Fix bug #10371, #set does not check response for server errors.
+  Submitted by Ben VandenBos.
+* Fix bug #12450, set TCP_NODELAY socket option.  Patch by Chris
+  McGrath.
+* Fix bug #10704, missing #add method.  Patch by Jamie Macey.
+* Fix bug #10371, handle socket EOF in cache_get.  Submitted by Ben
+  VandenBos.
+
+= 1.3.0
+
+* Apply patch #6507, add stats command.  Submitted by Tyler Kovacs.
+* Apply patch #6509, parallel implementation of #get_multi.  Submitted
+  by Tyler Kovacs.
+* Validate keys.  Disallow spaces in keys or keys that are too long.
+* Perform more validation of server responses.  MemCache now reports
+  errors if the socket was not in an expected state.  (Please file
+  bugs if you find some.)
+* Add #incr and #decr.
+* Add raw argument to #set and #get to retrieve #incr and #decr
+  values.
+* Also put on MemCacheError when using Cache::get with block.
+* memcache.rb no longer sets $TESTING to a true value if it was
+  previously defined.  Bug #8213 by Matijs van Zuijlen.
+
+= 1.2.1
+
+* Fix bug #7048, MemCache#servers= referenced changed local variable.
+  Submitted by Justin Dossey.
+* Fix bug #7049, MemCache#initialize resets @buckets.  Submitted by
+  Justin Dossey.
+* Fix bug #6232, Make Cache::Get work with a block only when nil is
+  returned.  Submitted by Jon Evans.
+* Moved to the seattlerb project.
+
+= 1.2.0
+
+NOTE: This version will store keys in different places than previous
+versions!  Be prepared for some thrashing while memcached sorts itself
+out!
+
+* Fixed multithreaded operations, bug 5994 and 5989.
+  Thanks to Blaine Cook, Erik Hetzner, Elliot Smith, Dave Myron (and
+  possibly others I have forgotten).
+* Made memcached interoperable with other memcached libraries, bug
+  4509.  Thanks to anonymous.
+* Added get_multi to match Perl/etc APIs
+
+= 1.1.0
+
+* Added some tests
+* Sped up non-multithreaded and multithreaded operation
+* More Ruby-memcache compatibility
+* More RDoc
+* Switched to Hoe
+
+= 1.0.0
+
+Birthday!
+

data/LICENSE.txt

@@ -0,0 +1,28 @@
+Copyright 2005-2009 Bob Cottrell, Eric Hodel, Mike Perham.
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+1. Redistributions of source code must retain the above copyright
+   notice, this list of conditions and the following disclaimer.
+2. Redistributions in binary form must reproduce the above copyright
+   notice, this list of conditions and the following disclaimer in the
+   documentation and/or other materials provided with the distribution.
+3. Neither the names of the authors nor the names of their contributors
+   may be used to endorse or promote products derived from this software
+   without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS
+OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
+LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
+OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
+OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
+EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+

data/README.rdoc

@@ -0,0 +1,51 @@
+= memcache-client
+
+A ruby library for accessing memcached.
+
+Source:
+
+http://github.com/mperham/memcache-client
+
+== Installing memcache-client
+
+Just install the gem:
+
+  $ sudo gem install memcache-client
+
+== Using memcache-client
+
+With one server:
+
+  CACHE = MemCache.new 'localhost:11211'
+
+Or with multiple servers:
+
+  CACHE = MemCache.new %w[one.example.com:11211 two.example.com:11211]
+
+
+== Tuning memcache-client
+
+The MemCache.new method takes a number of options which can be useful at times.  Please
+read the source comments there for an overview.  If you are using Ruby 1.8.x and using
+multiple memcached servers, you should install the RubyInline gem for ultimate performance.
+
+
+== Using memcache-client with Rails
+
+Rails 2.1+ includes memcache-client 1.5.0 out of the box.  See ActiveSupport::Cache::MemCacheStore
+and the Rails.cache method for more details.  Rails 2.3+ will use the latest memcache-client
+gem installed.
+
+
+== Questions?
+
+memcache-client is maintained by Mike Perham and was originally written by Bob Cottrell, 
+Eric Hodel and the seattle.rb crew.
+
+Email::        mailto:mperham@gmail.com
+Twitter::      mperham[http://twitter.com/mperham]
+WWW::          http://mikeperham.com
+
+If my work on memcache-client is something you support, please take a moment to
+recommend me at WWR[http://workingwithrails.com/person/10797-mike-perham].  I'm not
+asking for money, just a electronic "thumbs up".

data/Rakefile

@@ -0,0 +1,39 @@
+# vim: syntax=Ruby
+require 'rubygems'
+require 'rake/rdoctask'
+require 'rake/testtask'
+
+begin
+  require 'jeweler'
+  Jeweler::Tasks.new do |s|
+    s.name = "dsander-memcache-client"
+    s.summary = s.description = "A Ruby library for accessing memcached."
+    s.email = "mperham@gmail.com"
+    s.homepage = "http://github.com/mperham/memcache-client"
+    s.authors = ['Eric Hodel', 'Robert Cottrell', 'Mike Perham']
+    s.has_rdoc = true
+    s.files = FileList["[A-Z]*", "{lib,test}/**/*", 'performance.txt']
+    s.test_files = FileList["test/test_*.rb"]
+    s.rubyforge_project = 'seattlerb'
+  end
+
+rescue LoadError
+  puts "Jeweler not available. Install it for jeweler-related tasks with: sudo gem install jeweler"
+end
+
+
+Rake::RDocTask.new do |rd|
+	rd.main = "README.rdoc"
+	rd.rdoc_files.include("README.rdoc", "FAQ.rdoc", "History.rdoc", "lib/memcache.rb")
+	rd.rdoc_dir = 'doc'
+end
+
+Rake::TestTask.new do |t|
+  t.warning = true
+end
+
+task :default => :test
+
+task :rcov do
+  `rcov -Ilib test/*.rb`
+end

data/VERSION.yml

@@ -0,0 +1,5 @@
+--- 
+:major: 1
+:minor: 7
+:patch: 7
+:build: pre

data/lib/continuum_native.rb

@@ -0,0 +1,41 @@
+module Continuum
+
+  class << self
+
+    # Native extension to perform the binary search within the continuum
+    # space.  There's a pure ruby version in memcache.rb so this is purely
+    # optional for performance and only necessary if you are using multiple
+    # memcached servers.
+    begin
+      require 'inline'
+      inline do |builder|
+        builder.c <<-EOM
+        int binary_search(VALUE ary, unsigned int r) {
+            int upper = RARRAY_LEN(ary) - 1;
+            int lower = 0;
+            int idx = 0;
+            ID value = rb_intern("value");
+
+            while (lower <= upper) {
+                idx = (lower + upper) / 2;
+
+                VALUE continuumValue = rb_funcall(RARRAY_PTR(ary)[idx], value, 0);
+                unsigned int l = NUM2UINT(continuumValue);
+                if (l == r) {
+                    return idx;
+                }
+                else if (l > r) {
+                    upper = idx - 1;
+                }
+                else {
+                    lower = idx + 1;
+                }
+            }
+            return upper;
+        }
+        EOM
+      end
+    rescue Exception => e
+    end
+  end
+end