* 3.2.1 * Fix SentinelConnectionPool to work in multiprocess/forked environments. * 3.2.0 * Added support for `select.poll` to test whether data can be read on a socket. This should allow for significantly more connections to be used with pubsub. Fixes #486/#1115 * Attempt to guarentee that the ConnectionPool hands out healthy connections. Healthy connections are those that have an established socket connection to the Redis server, are ready to accept a command and have no data available to read. Fixes #1127/#886 * Use the socket.IPPROTO_TCP constant instead of socket.SOL_TCP. IPPROTO_TCP is available on more interpreters (Jython for instance). Thanks @Junnplus. #1130 * Fixed a regression introduced in 3.0 that mishandles exceptions not derived from the base Exception class. KeyboardInterrupt and gevent.timeout notable. Thanks Christian Fersch. #1128/#1129 * Significant improvements to handing connections with forked processes. Parent and child processes no longer trample on each others' connections. Thanks to Jay Rolette for the patch and highlighting this issue. #504/#732/#784/#863 * PythonParser no longer closes the associated connection's socket. The connection itself will close the socket. #1108/#1085 * 3.1.0 * Connection URLs must have one of the following schemes: redis://, rediss://, unix://. Thanks @jdupl123. #961/#969 * Fixed an issue with retry_on_timeout logic that caused some TimeoutErrors to be retried. Thanks Aaron Yang. #1022/#1023 * Added support for SNI for SSL. Thanks @oridistor and Roey Prat. #1087 * Fixed ConnectionPool repr for pools with no connections. Thanks Cody Scott. #1043/#995 * Fixed GEOHASH to return a None value when specifying a place that doesn't exist on the server. Thanks @guybe7. #1126 * Fixed XREADGROUP to return an empty dictionary for messages that have been deleted but still exist in the unacknowledged queue. Thanks @xeizmendi. #1116 * Added an owned method to Lock objects. owned returns a boolean indicating whether the current lock instance still owns the lock. Thanks Dave Johansen. #1112 * Allow lock.acquire() to accept an optional token argument. If provided, the token argument is used as the unique value used to claim the lock. Thankd Dave Johansen. #1112 * Added a reacquire method to Lock objects. reaquire attempts to renew the lock such that the timeout is extended to the same value that the lock was initially acquired with. Thanks Ihor Kalnytskyi. #1014 * Stream names found within XREAD and XREADGROUP responses now properly respect the decode_responses flag. * XPENDING_RANGE now requires the user the specify the min, max and count arguments. Newer versions of Redis prevent ount from being infinite so it's left to the user to specify these values explicitly. * ZADD now returns None when xx=True and incr=True and an element is specified that doesn't exist in the sorted set. This matches what the server returns in this case. #1084 * Added client_kill_filter that accepts various filters to identify and kill clients. Thanks Theofanis Despoudis. #1098 * Fixed a race condition that occurred when unsubscribing and resubscribing to the same channel or pattern in rapid succession. Thanks Marcin Raczyński. #764 * Added a LockNotOwnedError that is raised when trying to extend or release a lock that is no longer owned. This is a subclass of LockError so previous code should continue to work as expected. Thanks Joshua Harlow. #1095 * Fixed a bug in GEORADIUS that forced decoding of places without respecting the decode_responses option. Thanks Bo Bayles. #1082 * 3.0.1 * Fixed regression with UnixDomainSocketConnection caused by 3.0.0. Thanks Jyrki Muukkonen * Fixed an issue with the new asynchronous flag on flushdb and flushall. Thanks rogeryen * Updated Lock.locked() method to indicate whether *any* process has acquired the lock, not just the current one. This is in line with the behavior of threading.Lock. Thanks Alan Justino da Silva * 3.0.0 BACKWARDS INCOMPATIBLE CHANGES * When using a Lock as a context manager and the lock fails to be acquired a LockError is now raised. This prevents the code block inside the context manager from being executed if the lock could not be acquired. * Renamed LuaLock to Lock. * Removed the pipeline based Lock implementation in favor of the LuaLock implementation. * Only bytes, strings and numbers (ints, longs and floats) are acceptable for keys and values. Previously redis-py attempted to cast other types to str() and store the result. This caused must confusion and frustration when passing boolean values (cast to 'True' and 'False') or None values (cast to 'None'). It is now the user's responsibility to cast all key names and values to bytes, strings or numbers before passing the value to redis-py. * The StrictRedis class has been renamed to Redis. StrictRedis will continue to exist as an alias of Redis for the forseeable future. * The legacy Redis client class has been removed. It caused much confusion to users. * ZINCRBY arguments 'value' and 'amount' have swapped order to match the the Redis server. The new argument order is: keyname, amount, value. * MGET no longer raises an error if zero keys are passed in. Instead an empty list is returned. * MSET and MSETNX now require all keys/values to be specified in a single dictionary argument named mapping. This was changed to allow for future options to these commands in the future. * ZADD now requires all element names/scores be specified in a single dictionary argument named mapping. This was required to allow the NX, XX, CH and INCR options to be specified. * ssl_cert_reqs now has a default value of 'required' by default. This should make connecting to a remote Redis server over SSL more secure. Thanks u2mejc * Removed support for EOL Python 2.6 and 3.3. Thanks jdufresne OTHER CHANGES * Added missing DECRBY command. Thanks derek-dchu * CLUSTER INFO and CLUSTER NODES respones are now properly decoded to strings. * Added a 'locked()' method to Lock objects. This method returns True if the lock has been acquired and owned by the current process, otherwise False. * EXISTS now supports multiple keys. It's return value is now the number of keys in the list that exist. * Ensure all commands can accept key names as bytes. This fixes issues with BLPOP, BRPOP and SORT. * All errors resulting from bad user input are raised as DataError exceptions. DataError is a subclass of RedisError so this should be transparent to anyone previously catching these. * Added support for NX, XX, CH and INCR options to ZADD * Added support for the MIGRATE command * Added support for the MEMORY USAGE and MEMORY PURGE commands. Thanks Itamar Haber * Added support for the 'asynchronous' argument to FLUSHDB and FLUSHALL commands. Thanks Itamar Haber * Added support for the BITFIELD command. Thanks Charles Leifer and Itamar Haber * Improved performance on pipeline requests with large chunks of data. Thanks tzickel * Fixed test suite to not fail if another client is connected to the server the tests are running against. * Added support for SWAPDB. Thanks Itamar Haber * Added support for all STREAM commands. Thanks Roey Prat and Itamar Haber * SHUTDOWN now accepts the 'save' and 'nosave' arguments. Thanks dwilliams-kenzan * Added support for ZPOPMAX, ZPOPMIN, BZPOPMAX, BZPOPMIN. Thanks Itamar Haber * Added support for the 'type' argument in CLIENT LIST. Thanks Roey Prat * Added support for CLIENT PAUSE. Thanks Roey Prat * Added support for CLIENT ID and CLIENT UNBLOCK. Thanks Itamar Haber * GEODIST now returns a None value when referencing a place that does not exist. Thanks qingping209 * Added a ping() method to pubsub objects. Thanks krishan-carbon * Fixed a bug with keys in the INFO dict that contained ':' symbols. Thanks mzalimeni * Fixed the select system call retry compatibility with Python 2.x. Thanks lddubeau * max_connections is now a valid querystring argument for creating connection pools from URLs. Thanks mmaslowskicc * Added the UNLINK command. Thanks yozel * Added socket_type option to Connection for configurability. Thanks garlicnation * Lock.do_acquire now atomically sets acquires the lock and sets the expire value via set(nx=True, px=timeout). Thanks 23doors * Added 'count' argument to SPOP. Thanks AlirezaSadeghi * Fixed an issue parsing client_list respones that contained an '='. Thanks swilly22 * 2.10.6 * Various performance improvements. Thanks cjsimpson * Fixed a bug with SRANDMEMBER where the behavior for `number=0` did not match the spec. Thanks Alex Wang * Added HSTRLEN command. Thanks Alexander Putilin * Added the TOUCH command. Thanks Anis Jonischkeit * Remove unnecessary calls to the server when registering Lua scripts. Thanks Ben Greenberg * SET's EX and PX arguments now allow values of zero. Thanks huangqiyin * Added PUBSUB {CHANNELS, NUMPAT, NUMSUB} commands. Thanks Angus Pearson * PubSub connections that that encounter `InterruptedError`s now retry automatically. Thanks Carlton Gibson and Seth M. Larson * LPUSH and RPUSH commands run on PyPy now correctly returns the number of items of the list. Thanks Jeong YunWon * Added support to automatically retry socket EINTR errors. Thanks Thomas Steinacher * PubSubWorker threads started with `run_in_thread` are now daemonized so the thread shuts down when the running process goes away. Thanks Keith Ainsworth * Added support for GEO commands. Thanks Pau Freixes, Alex DeBrie and Abraham Toriz * Made client construction from URLs smarter. Thanks Tim Savage * Added support for CLUSTER * commands. Thanks Andy Huang * The RESTORE command now accepts an optional `replace` boolean. Thanks Yoshinari Takaoka * Attempt to connect to a new Sentinel if a TimeoutError occurs. Thanks Bo Lopker * Fixed a bug in the client's `__getitem__` where a KeyError would be raised if the value returned by the server is an empty string. Thanks Javier Candeira. * Socket timeouts when connecting to a server are now properly raised as TimeoutErrors. * 2.10.5 * Allow URL encoded parameters in Redis URLs. Characters like a "/" can now be URL encoded and redis-py will correctly decode them. Thanks Paul Keene. * Added support for the WAIT command. Thanks https://github.com/eshizhan * Better shutdown support for the PubSub Worker Thread. It now properly cleans up the connection, unsubscribes from any channels and patterns previously subscribed to and consumes any waiting messages on the socket. * Added the ability to sleep for a brief period in the event of a WatchError occuring. Thanks Joshua Harlow. * Fixed a bug with pipeline error reporting when dealing with characters in error messages that could not be encoded to the connection's character set. Thanks Hendrik Muhs. * Fixed a bug in Sentinel connections that would inadvertantly connect to the master when the connection pool resets. Thanks https://github.com/df3n5 * Better timeout support in Pubsub get_message. Thanks Andy Isaacson. * Fixed a bug with the HiredisParser that would cause the parser to get stuck in an endless loop if a specific number of bytes were delivered from the socket. This fix also increases performance of parsing large responses from the Redis server. * Added support for ZREVRANGEBYLEX. * ConnectionErrors are now raised if Redis refuses a connection due to the maxclients limit being exceeded. Thanks Roman Karpovich. * max_connections can now be set when instantiating client instances. Thanks Ohad Perry. * 2.10.4 (skipped due to a PyPI snafu) * 2.10.3 * Fixed a bug with the bytearray support introduced in 2.10.2. Thanks Josh Owen. * 2.10.2 * Added support for Hiredis's new bytearray support. Thanks https://github.com/tzickel * POSSIBLE BACKWARDS INCOMPATBLE CHANGE: Fixed a possible race condition when multiple threads share the same Lock instance with a timeout. Lock tokens are now stored in thread local storage by default. If you have code that acquires a lock in one thread and passes that lock instance to another thread to release it, you need to disable thread local storage. Refer to the doc strings on the Lock class about the thread_local argument information. * Fixed a regression in from_url where "charset" and "errors" weren't valid options. "encoding" and "encoding_errors" are still accepted and preferred. * The "charset" and "errors" options have been deprecated. Passing either to StrictRedis.__init__ or from_url will still work but will also emit a DeprecationWarning. Instead use the "encoding" and "encoding_errors" options. * Fixed a compatability bug with Python 3 when the server closes a connection. * Added BITPOS command. Thanks https://github.com/jettify. * Fixed a bug when attempting to send large values to Redis in a Pipeline. * 2.10.1 * Fixed a bug where Sentinel connections to a server that's no longer a master and receives a READONLY error will disconnect and reconnect to the master. * 2.10.0 * Discontinuted support for Python 2.5. Upgrade. You'll be happier. * The HiRedis parser will now properly raise ConnectionErrors. * Completely refactored PubSub support. Fixes all known PubSub bugs and adds a bunch of new features. Docs can be found in the README under the new "Publish / Subscribe" section. * Added the new HyperLogLog commanads (PFADD, PFCOUNT, PFMERGE). Thanks Pepijn de Vos and Vincent Ohprecio. * Updated TTL and PTTL commands with Redis 2.8+ semantics. Thanks Markus Kaiserswerth. * *SCAN commands now return a long (int on Python3) cursor value rather than the string representation. This might be slightly backwards incompatible in code using *SCAN commands loops such as "while cursor != '0':". * Added extra *SCAN commands that return iterators instead of the normal [cursor, data] type. Use scan_iter, hscan_iter, sscan_iter, and zscan_iter for iterators. Thanks Mathieu Longtin. * Added support for SLOWLOG commands. Thanks Rick van Hattem. * Added lexicographical commands ZRANGEBYLEX, ZREMRANGEBYLEX, and ZLEXCOUNT for sorted sets. * Connection objects now support an optional argument, socket_read_size, indicating how much data to read during each socket.recv() call. After benchmarking, increased the default size to 64k, which dramatically improves performance when fetching large values, such as many results in a pipeline or a large (>1MB) string value. * Improved the pack_command and send_packed_command functions to increase performance when sending large (>1MB) values. * Sentinel Connections to master servers now detect when a READONLY error is encountered and disconnect themselves and all other active connections to the same master so that the new master can be discovered. * Fixed Sentinel state parsing on Python 3. * Added support for SENTINEL MONITOR, SENTINEL REMOVE, and SENTINEL SET commands. Thanks Greg Murphy. * INFO ouput that doesn't follow the "key:value" format will now be appended to a key named "__raw__" in the INFO dictionary. Thanks Pedro Larroy. * The "vagrant" directory contains a complete vagrant environment for redis-py developers. The environment runs a Redis master, a Redis slave, and 3 Sentinels. Future iterations of the test sutie will incorporate more integration style tests, ensuring things like failover happen correctly. * It's now possible to create connection pool instances from a URL. StrictRedis.from_url() now uses this feature to create a connection pool instance and use that when creating a new client instance. Thanks https://github.com/chillipino * When creating client instances or connection pool instances from an URL, it's now possible to pass additional options to the connection pool with querystring arguments. * Fixed a bug where some encodings (like utf-16) were unusable on Python 3 as command names and literals would get encoded. * Added an SSLConnection class that allows for secure connections through stunnel or other means. Construct and SSL connection with the sll=True option on client classes, using the rediss:// scheme from an URL, or by passing the SSLConnection class to a connection pool's connection_class argument. Thanks https://github.com/oranagra. * Added a socket_connect_timeout option to control how long to wait while establishing a TCP connection before timing out. This lets the client fail fast when attempting to connect to a downed server while keeping a more lenient timeout for all other socket operations. * Added TCP Keep-alive support by passing use the socket_keepalive=True option. Finer grain control can be achieved using the socket_keepalive_options option which expects a dictionary with any of the keys (socket.TCP_KEEPIDLE, socket.TCP_KEEPCNT, socket.TCP_KEEPINTVL) and integers for values. Thanks Yossi Gottlieb. * Added a `retry_on_timeout` option that controls how socket.timeout errors are handled. By default it is set to False and will cause the client to raise a TimeoutError anytime a socket.timeout is encountered. If `retry_on_timeout` is set to True, the client will retry a command that timed out once like other `socket.error`s. * Completely refactored the Lock system. There is now a LuaLock class that's used when the Redis server is capable of running Lua scripts along with a fallback class for Redis servers < 2.6. The new locks fix several subtle race consider that the old lock could face. In additional, a new method, "extend" is available on lock instances that all a lock owner to extend the amount of time they have the lock for. Thanks to Eli Finkelshteyn and https://github.com/chillipino for contributions. * 2.9.1 * IPv6 support. Thanks https://github.com/amashinchi * 2.9.0 * Performance improvement for packing commands when using the PythonParser. Thanks Guillaume Viot. * Executing an empty pipeline transaction no longer sends MULTI/EXEC to the server. Thanks EliFinkelshteyn. * Errors when authenticating (incorrect password) and selecting a database now close the socket. * Full Sentinel support thanks to Vitja Makarov. Thanks! * Better repr support for client and connection pool instances. Thanks Mark Roberts. * Error messages that the server sends to the client are now included in the client error message. Thanks Sangjin Lim. * Added the SCAN, SSCAN, HSCAN, and ZSCAN commands. Thanks Jingchao Hu. * ResponseErrors generated by pipeline execution provide addition context including the position of the command in the pipeline and the actual command text generated the error. * ConnectionPools now play nicer in threaded environments that fork. Thanks Christian Joergensen. * 2.8.0 * redis-py should play better with gevent when a gevent Timeout is raised. Thanks leifkb. * Added SENTINEL command. Thanks Anna Janackova. * Fixed a bug where pipelines could potentially corrupt a connection if the MULTI command generated a ResponseError. Thanks EliFinkelshteyn for the report. * Connections now call socket.shutdown() prior to socket.close() to ensure communication ends immediately per the note at https://docs.python.org/2/library/socket.html#socket.socket.close Thanks to David Martin for pointing this out. * Lock checks are now based on floats rather than ints. Thanks Vitja Makarov. * 2.7.6 * Added CONFIG RESETSTAT command. Thanks Yossi Gottlieb. * Fixed a bug introduced in 2.7.3 that caused issues with script objects and pipelines. Thanks Carpentier Pierre-Francois. * Converted redis-py's test suite to use the awesome py.test library. * Fixed a bug introduced in 2.7.5 that prevented a ConnectionError from being raised when the Redis server is LOADING data. * Added a BusyLoadingError exception that's raised when the Redis server is starting up and not accepting commands yet. BusyLoadingError subclasses ConnectionError, which this state previously returned. Thanks Yossi Gottlieb. * 2.7.5 * DEL, HDEL and ZREM commands now return the numbers of keys deleted instead of just True/False. * from_url now supports URIs with a port number. Thanks Aaron Westendorf. * 2.7.4 * Added missing INCRBY method. Thanks Krzysztof Dorosz. * SET now accepts the EX, PX, NX and XX options from Redis 2.6.12. These options will generate errors if these options are used when connected to a Redis server < 2.6.12. Thanks George Yoshida. * 2.7.3 * Fixed a bug with BRPOPLPUSH and lists with empty strings. * All empty except: clauses have been replaced to only catch Exception subclasses. This prevents a KeyboardInterrupt from triggering exception handlers. Thanks Lucian Branescu Mihaila. * All exceptions that are the result of redis server errors now share a command Exception subclass, ServerError. Thanks Matt Robenolt. * Prevent DISCARD from being called if MULTI wasn't also called. Thanks Pete Aykroyd. * SREM now returns an integer indicating the number of items removed from the set. Thanks https://github.com/ronniekk. * Fixed a bug with BGSAVE and BGREWRITEAOF response callbacks with Python3. Thanks Nathan Wan. * Added CLIENT GETNAME and CLIENT SETNAME commands. Thanks https://github.com/bitterb. * It's now possible to use len() on a pipeline instance to determine the number of commands that will be executed. Thanks Jon Parise. * Fixed a bug in INFO's parse routine with floating point numbers. Thanks Ali Onur Uyar. * Fixed a bug with BITCOUNT to allow `start` and `end` to both be zero. Thanks Tim Bart. * The transaction() method now accepts a boolean keyword argument, value_from_callable. By default, or if False is passes, the transaction() method will return the value of the pipelines execution. Otherwise, it will return whatever func() returns. * Python3 compatibility fix ensuring we're not already bytes(). Thanks Salimane Adjao Moustapha. * Added PSETEX. Thanks YAMAMOTO Takashi. * Added a BlockingConnectionPool to limit the number of connections that can be created. Thanks James Arthur. * SORT now accepts a `groups` option that if specified, will return tuples of n-length, where n is the number of keys specified in the GET argument. This allows for convenient row-based iteration. Thanks Ionuț Arțăriși. * 2.7.2 * Parse errors are now *always* raised on multi/exec pipelines, regardless of the `raise_on_error` flag. See https://groups.google.com/forum/?hl=en&fromgroups=#!topic/redis-db/VUiEFT8U8U0 for more info. * 2.7.1 * Packaged tests with source code * 2.7.0 * Added BITOP and BITCOUNT commands. Thanks Mark Tozzi. * Added the TIME command. Thanks Jason Knight. * Added support for LUA scripting. Thanks to Angus Peart, Drew Smathers, Issac Kelly, Louis-Philippe Perron, Sean Bleier, Jeffrey Kaditz, and Dvir Volk for various patches and contributions to this feature. * Changed the default error handling in pipelines. By default, the first error in a pipeline will now be raised. A new parameter to the pipeline's execute, `raise_on_error`, can be set to False to keep the old behavior of embeedding the exception instances in the result. * Fixed a bug with pipelines where parse errors won't corrupt the socket. * Added the optional `number` argument to SRANDMEMBER for use with Redis 2.6+ servers. * Added PEXPIRE/PEXPIREAT/PTTL commands. Thanks Luper Rouch. * Added INCRBYFLOAT/HINCRBYFLOAT commands. Thanks Nikita Uvarov. * High precision floating point values won't lose their precision when being sent to the Redis server. Thanks Jason Oster and Oleg Pudeyev. * Added CLIENT LIST/CLIENT KILL commands * 2.6.2 * `from_url` is now available as a classmethod on client classes. Thanks Jon Parise for the patch. * Fixed several encoding errors resulting from the Python 3.x support. * 2.6.1 * Python 3.x support! Big thanks to Alex Grönholm. * Fixed a bug in the PythonParser's read_response that could hide an error from the client (#251). * 2.6.0 * Changed (p)subscribe and (p)unsubscribe to no longer return messages indicating the channel was subscribed/unsubscribed to. These messages are available in the listen() loop instead. This is to prevent the following scenario: * Client A is subscribed to "foo" * Client B publishes message to "foo" * Client A subscribes to channel "bar" at the same time. Prior to this change, the subscribe() call would return the published messages on "foo" rather than the subscription confirmation to "bar". * Added support for GETRANGE, thanks Jean-Philippe Caruana * A new setting "decode_responses" specifies whether return values from Redis commands get decoded automatically using the client's charset value. Thanks to Frankie Dintino for the patch. * 2.4.13 * redis.from_url() can take an URL representing a Redis connection string and return a client object. Thanks Kenneth Reitz for the patch. * 2.4.12 * ConnectionPool is now fork-safe. Thanks Josiah Carson for the patch. * 2.4.11 * AuthenticationError will now be correctly raised if an invalid password is supplied. * If Hiredis is unavailable, the HiredisParser will raise a RedisError if selected manually. * Made the INFO command more tolerant of Redis changes formatting. Fix for #217. * 2.4.10 * Buffer reads from socket in the PythonParser. Fix for a Windows-specific bug (#205). * Added the OBJECT and DEBUG OBJECT commands. * Added __del__ methods for classes that hold on to resources that need to be cleaned up. This should prevent resource leakage when these objects leave scope due to misuse or unhandled exceptions. Thanks David Wolever for the suggestion. * Added the ECHO command for completeness. * Fixed a bug where attempting to subscribe to a PubSub channel of a Redis server that's down would blow out the stack. Fixes #179 and #195. Thanks Ovidiu Predescu for the test case. * StrictRedis's TTL command now returns a -1 when querying a key with no expiration. The Redis class continues to return None. * ZADD and SADD now return integer values indicating the number of items added. Thanks Homer Strong. * Renamed the base client class to StrictRedis, replacing ZADD and LREM in favor of their official argument order. The Redis class is now a subclass of StrictRedis, implementing the legacy redis-py implementations of ZADD and LREM. Docs have been updated to suggesting the use of StrictRedis. * SETEX in StrictRedis is now compliant with official Redis SETEX command. the name, value, time implementation moved to "Redis" for backwards compatability. * 2.4.9 * Removed socket retry logic in Connection. This is the responsbility of the caller to determine if the command is safe and can be retried. Thanks David Wolver. * Added some extra guards around various types of exceptions being raised when sending or parsing data. Thanks David Wolver and Denis Bilenko. * 2.4.8 * Imported with_statement from __future__ for Python 2.5 compatability. * 2.4.7 * Fixed a bug where some connections were not getting released back to the connection pool after pipeline execution. * Pipelines can now be used as context managers. This is the preferred way of use to ensure that connections get cleaned up properly. Thanks David Wolever. * Added a convenience method called transaction() on the base Redis class. This method eliminates much of the boilerplate used when using pipelines to watch Redis keys. See the documentation for details on usage. * 2.4.6 * Variadic arguments for SADD, SREM, ZREN, HDEL, LPUSH, and RPUSH. Thanks Raphaël Vinot. * (CRITICAL) Fixed an error in the Hiredis parser that occasionally caused the socket connection to become corrupted and unusable. This became noticeable once connection pools started to be used. * ZRANGE, ZREVRANGE, ZRANGEBYSCORE, and ZREVRANGEBYSCORE now take an additional optional argument, score_cast_func, which is a callable used to cast the score value in the return type. The default is float. * Removed the PUBLISH method from the PubSub class. Connections that are [P]SUBSCRIBEd cannot issue PUBLISH commands, so it doesn't make sense to have it here. * Pipelines now contain WATCH and UNWATCH. Calling WATCH or UNWATCH from the base client class will result in a deprecation warning. After WATCHing one or more keys, the pipeline will be placed in immediate execution mode until UNWATCH or MULTI are called. Refer to the new pipeline docs in the README for more information. Thanks to David Wolever and Randall Leeds for greatly helping with this. * 2.4.5 * The PythonParser now works better when reading zero length strings. * 2.4.4 * Fixed a typo introduced in 2.4.3 * 2.4.3 * Fixed a bug in the UnixDomainSocketConnection caused when trying to form an error message after a socket error. * 2.4.2 * Fixed a bug in pipeline that caused an exception while trying to reconnect after a connection timeout. * 2.4.1 * Fixed a bug in the PythonParser if disconnect is called before connect. * 2.4.0 * WARNING: 2.4 contains several backwards incompatible changes. * Completely refactored Connection objects. Moved much of the Redis protocol packing for requests here, and eliminated the nasty dependencies it had on the client to do AUTH and SELECT commands on connect. * Connection objects now have a parser attribute. Parsers are responsible for reading data Redis sends. Two parsers ship with redis-py: a PythonParser and the HiRedis parser. redis-py will automatically use the HiRedis parser if you have the Python hiredis module installed, otherwise it will fall back to the PythonParser. You can force or the other, or even an external one by passing the `parser_class` argument to ConnectionPool. * Added a UnixDomainSocketConnection for users wanting to talk to the Redis instance running on a local machine only. You can use this connection by passing it to the `connection_class` argument of the ConnectionPool. * Connections no longer derive from threading.local. See threading.local note below. * ConnectionPool has been comletely refactored. The ConnectionPool now maintains a list of connections. The redis-py client only hangs on to a ConnectionPool instance, calling get_connection() anytime it needs to send a command. When get_connection() is called, the command name and any keys involved in the command are passed as arguments. Subclasses of ConnectionPool could use this information to identify the shard the keys belong to and return a connection to it. ConnectionPool also implements disconnect() to force all connections in the pool to disconnect from the Redis server. * redis-py no longer support the SELECT command. You can still connect to a specific database by specifing it when instantiating a client instance or by creating a connection pool. If you need to talk to multiplate databases within your application, you should use a separate client instance for each database you want to talk to. * Completely refactored Publish/Subscribe support. The subscribe and listen commands are no longer available on the redis-py Client class. Instead, the `pubsub` method returns an instance of the PubSub class which contains all publish/subscribe support. Note, you can still PUBLISH from the redis-py client class if you desire. * Removed support for all previously deprecated commands or options. * redis-py no longer uses threading.local in any way. Since the Client class no longer holds on to a connection, it's no longer needed. You can now pass client instances between threads, and commands run on those threads will retrieve an available connection from the pool, use it and release it. It should now be trivial to use redis-py with eventlet or greenlet. * ZADD now accepts pairs of value=score keyword arguments. This should help resolve the long standing #72. The older value and score arguments have been deprecated in favor of the keyword argument style. * Client instances now get their own copy of RESPONSE_CALLBACKS. The new set_response_callback method adds a user defined callback to the instance. * Support Jython, fixing #97. Thanks to Adam Vandenberg for the patch. * Using __getitem__ now properly raises a KeyError when the key is not found. Thanks Ionuț Arțăriși for the patch. * Newer Redis versions return a LOADING message for some commands while the database is loading from disk during server start. This could cause problems with SELECT. We now force a socket disconnection prior to raising a ResponseError so subsuquent connections have to reconnect and re-select the appropriate database. Thanks to Benjamin Anderson for finding this and fixing. * 2.2.4 * WARNING: Potential backwards incompatible change - Changed order of parameters of ZREVRANGEBYSCORE to match those of the actual Redis command. This is only backwards-incompatible if you were passing max and min via keyword args. If passing by normal args, nothing in user code should have to change. Thanks Stéphane Angel for the fix. * Fixed INFO to properly parse the Redis data correctly for both 2.2.x and 2.3+. Thanks Stéphane Angel for the fix. * Lock objects now store their timeout value as a float. This allows floats to be used as timeout values. No changes to existing code required. * WATCH now supports multiple keys. Thanks Rich Schumacher. * Broke out some code that was Python 2.4 incompatible. redis-py should now be useable on 2.4, but this hasn't actually been tested. Thanks Dan Colish for the patch. * Optimized some code using izip and islice. Should have a pretty good speed up on larger data sets. Thanks Dan Colish. * Better error handling when submitting an empty mapping to HMSET. Thanks Dan Colish. * Subscription status is now reset after every (re)connection. * 2.2.3 * Added support for Hiredis. To use, simply "pip install hiredis" or "easy_install hiredis". Thanks for Pieter Noordhuis for the hiredis-py bindings and the patch to redis-py. * The connection class is chosen based on whether hiredis is installed or not. To force the use of the PythonConnection, simply create your own ConnectionPool instance with the connection_class argument assigned to to PythonConnection class. * Added missing command ZREVRANGEBYSCORE. Thanks Jay Baird for the patch. * The INFO command should be parsed correctly on 2.2.x server versions and is backwards compatible with older versions. Thanks Brett Hoerner. * 2.2.2 * Fixed a bug in ZREVRANK where retriving the rank of a value not in the zset would raise an error. * Fixed a bug in Connection.send where the errno import was getting overwritten by a local variable. * Fixed a bug in SLAVEOF when promoting an existing slave to a master. * Reverted change of download URL back to redis-VERSION.tar.gz. 2.2.1's change of this actually broke Pypi for Pip installs. Sorry! * 2.2.1 * Changed archive name to redis-py-VERSION.tar.gz to not conflict with the Redis server archive. * 2.2.0 * Implemented SLAVEOF * Implemented CONFIG as config_get and config_set * Implemented GETBIT/SETBIT * Implemented BRPOPLPUSH * Implemented STRLEN * Implemented PERSIST * Implemented SETRANGE