v3.0.0

Breaking Change: Redlock now consistently uses seconds for all time units

In particular:

1. The auto_release_time argument to Redlock.__init__() and synchronize() is now in seconds (was previously milliseconds)
2. Redlock.locked() now returns how much longer you'll hold the lease in seconds (or 0 if you don't hold the lock; and was also previously milliseconds)

What's Changed

• Remove superfluous return statement by @brainix in #626
• In Redlock, consistently measure time in seconds by @brainix in #627

Full Changelog: v2.3.7...v3.0.0

v2.3.7

Optimization: CachedOrderedDict.__init__()

Dramatically improve the performance of CachedOrderedDict.__init__(). Prior to this release, we were making multiple round trips to Redis, one per dict_key. As of this release, we make a single bulk call to Redis.

What's Changed

• Document NextId.init() kwargs in docstring by @brainix in #592
• Upgrade requirements by @brainix in #593
• Comment which Redis versions introduced commands by @brainix in #594
• Use dataclasses to define exception classes by @brainix in #595
• Share the same logger between monkey and base by @brainix in #596
• Upgrade Python to 3.10.2 by @brainix in #597
• Improve test coverage by @brainix in #598
• Check for security vulnerabilities by @brainix in #599
• Upgrade requirements by @brainix in #600
• Check for security vulnerabilities on local dev by @brainix in #601
• Upgrade requirements by @brainix in #602
• Upgrade requirements by @brainix in #603
• Implement RedisSet.contains() using SISMEMBER by @brainix in #604
• Lint on file save with Bandit by @brainix in #605
• Upgrade requirements by @brainix in #606
• Upgrade requirements by @brainix in #607
• Write union types as X | Y by @brainix in #608
• Tighten up type checking and type annotations by @brainix in #609
• Upgrade requirements by @brainix in #610
• Upgrade requirements by @brainix in #611
• Configure Bandit to not warn on asserts in tests by @brainix in #612
• Add emojis to README by @brainix in #613
• Write union types as X | Y by @brainix in #614
• Use protocol instead of making _Clearable an ABC by @brainix in #615
• Upgrade requirements by @brainix in #616
• Rename Base class to Container by @brainix in #617
• Refactor _Common class by @brainix in #618
• Clean up formatting by @brainix in #619
• Upgrade requirements by @brainix in #620
• Improve Redlock usage documentation by @brainix in #621
• Upgrade requirements by @brainix in #622
• Create a proper Pottery JSON encoder by @brainix in #623
• Refactor json.dumps() monkey patch by @brainix in #624
• Dramatically speed up CachedOrderedDict.init() by @brainix in #625

Full Changelog: v2.3.6...v2.3.7

v2.3.6

1. Bug Fix: Properly configure logging for the Pottery library
2. Optimization: Prefer Redis UNLINK over DEL; UNLINK is faster and non-blocking

What's Changed

• Happy new year! by @brainix in #577
• Upgrade requirements by @brainix in #578
• Prefer Redis UNLINK over DEL by @brainix in #579
• Comment which Redis versions introduced commands by @brainix in #580
• Upgrade requirements by @brainix in #581
• Preserve Open-Closed Principle with name mangling by @brainix in #582
• Upgrade requirements by @brainix in #583
• Define ABCs by subclassing abc.ABC by @brainix in #584
• Unit test that ABCs can't be instantiated by @brainix in #585
• Upgrade requirements by @brainix in #586
• Upgrade requirements by @brainix in #587
• Document Redlock.init() kwargs in docstrings by @brainix in #588
• Use isinstance() to provide a hint to Mypy by @brainix in #589
• Don't specify 'utf-8' argument to bytes.decode() by @brainix in #590
• Properly configure logging for the Pottery library by @brainix in #591

Full Changelog: v2.3.5...v2.3.6

v2.3.5

Bug Fix: Don't allow a RedisDeque to equal a RedisList...

...even if they're both on the same Redis instance and have the same key.

Before this release:

>>> from pottery import RedisDeque, RedisList
>>> RedisDeque(key='videos:dicts') == RedisList(key='videos:dicts')
True

As of this release:

>>> from pottery import RedisDeque, RedisList
>>> RedisDeque(key='videos:dicts') == RedisList(key='videos:dicts')
False

What's Changed

• Make minor code cleanups by @brainix in #575
• Don't allow a RedisDeque to equal a RedisList... by @brainix in #576

Full Changelog: v2.3.4...v2.3.5

v2.3.4

Bug Fix: Don't allow RedisDeques to equal Python lists...

...even when they contain the same elements. Observe:

>>> import collections
>>> collections.deque([1, 2, 3]) == [1, 2, 3]
False

What's Changed

• Don't allow RedisDeques to equal Python lists... by @brainix in #574

Full Changelog: v2.3.3...v2.3.4

v2.3.3

Bug Fix: Don't allow RedisDeques to equal RedisLists...

...even when they contain the same elements. Observe:

>>> import collections
>>> collections.deque([1, 2, 3]) == [1, 2, 3]
False

What's Changed

• Dramatically simplify RedisList.eq() by @brainix in #573

Full Changelog: v2.3.2...v2.3.3

v2.3.2

Bug Fix: Properly compare RedisLists on different Redis databases

Prior to this release, list1 == list2 would fail if both were RedisLists but on different Redis databases. This is because we'd try to use the same Redis pipeline for both lists. Fixed in #572.

What's Changed

• Reuse the ._same_redis() helper method by @brainix in #561
• Reorder methods in Primitive ABC to help subclass by @brainix in #562
• Use redis-py's .hscan_iter() and .sscan_iter() by @brainix in #563
• Preserve Open-Closed Principle with name mangling by @brainix in #564
• Simplify code using chunking and aliasing by @brainix in #565
• Name variables more consistently by @brainix in #566
• Use whitespace to make code more clear by @brainix in #567
• Preserve Open-Closed Principle with name mangling by @brainix in #568
• Refactor code in RedisCounter methods for clarity by @brainix in #569
• Simplify RedisDict._populate() by @brainix in #570
• Factor our RedisDict._encode_dict() by @brainix in #571
• Properly compare RedisLists on different Redis dbs by @brainix in #572

Full Changelog: v2.3.1...v2.3.2

v2.3.1

Bug Fix: More robustly test if two Redis clients talk to the same database

Previously, we were comparing the two clients' connection_kwargs, but connection_kwargs contains more than just host, port, and db:

>>> from redis import Redis
>>> redis = Redis()
>>> redis.connection_pool.connection_kwargs
{'db': 0, 'username': None, 'password': None, 'socket_timeout': None, 'encoding': 'utf-8', 'encoding_errors': 'strict', 'decode_responses': False, 'retry_on_error': [], 'retry': None, 'health_check_interval': 0, 'client_name': None, 'redis_connect_func': None, 'host': 'localhost', 'port': 6379, 'socket_connect_timeout': None, 'socket_keepalive': None, 'socket_keepalive_options': None}

This PR allows Pottery to recognize that two Redis clients are connected to the same database even if their socket timeout our retry policies are different.

Bug Fix: Make RedisSet.contains_many() work for non-JSONifyable objects

v2.3.0

New Feature: RedisSimpleQueue

RedisSimpleQueue is a Redis-backed multi-producer, multi-consumer FIFO queue compatible with Python’s queue.SimpleQueue. In general, use a Python queue.Queue if you’re using it in one or more threads, use multiprocessing.Queue if you’re using it between processes, and use RedisSimpleQueue if you’re sharing it across machines or if you need for your queue to persist across application crashes or restarts.

What's Changed

• Design RedisSimpleQueue class by @brainix in #553

Full Changelog: v2.2.2...v2.3.0

v2.2.2

Bug Fix: Import Script from the correct place

Reported in #552, fixed in #554.

What's Changed

• Use common logger object across entire library by @brainix in #547
• Postpone evaluation of annotations by @brainix in #548
• Upgrade requirements by @brainix in #549
• Write HyperLogLog doctests by @brainix in #550
• Consistently name UUID variable by @brainix in #551
• Fix #552: Import Script from the correct place by @brainix in #554

Full Changelog: v2.2.1...v2.2.2