1+ 2.0.2
2+ =====
3+
4+ Bug Fixes
5+ ---------
6+ * Add six to requirements.txt
7+ * Avoid KeyError during schema refresh when a keyspace is dropped
8+ and TokenAwarePolicy is not in use
9+ * Avoid registering multiple atexit() cleanup functions will the
10+ asyncore event loop is restarted multiple times
11+ * Delay initialization of reactors in order to avoid problems
12+ with shared state when using multiprocessing (PYTHON-60)
13+ * Add python-six to debian dependencies, move python-blist to
14+ recommends
15+
16+ 2.0.1
17+ =====
18+ May 28, 2014
19+
20+ Bug Fixes
21+ ---------
22+ * Fix check for Cluster.is_shutdown in in @run_in_executor
23+ decorator
24+
25+ 2.0.0
26+ =====
27+ May 28, 2014
28+
29+ Features
30+ --------
31+ * Make libev C extension Python3-compatible (PYTHON-70)
32+ * Support v2 protocol authentication (PYTHON-73, github #125)
33+
34+ Bug Fixes
35+ ---------
36+ * Fix murmur3 C extension compilation under Python3.4 (github #124)
37+
38+ Merged From 1.x
39+ ---------------
40+
41+ Features
42+ ^^^^^^^^
43+ * Add Session.default_consistency_level (PYTHON-14)
44+
45+ Bug Fixes
46+ ^^^^^^^^^
47+ * Don't strip trailing underscores from column names when using the
48+ named_tuple_factory (PYTHON-56)
49+ * Ensure replication factors are ints for NetworkTopologyStrategy
50+ to avoid TypeErrors (github #120)
51+ * Pass WriteType instance to RetryPolicy.on_write_timeout() instead
52+ of the string name of the write type. This caused write timeout
53+ errors to always be rethrown instead of retrying. (github #123)
54+ * Avoid submitting tasks to the ThreadPoolExecutor after shutdown. With
55+ retries enabled, this could cause Cluster.shutdown() to hang under
56+ some circumstances.
57+ * Fix unintended rebuild of token replica map when keyspaces are
58+ discovered (on startup), added, or updated and TokenAwarePolicy is not
59+ in use.
60+ * Avoid rebuilding token metadata when cluster topology has not
61+ actually changed
62+ * Avoid preparing queries for hosts that should be ignored (such as
63+ remote hosts when using the DCAwareRoundRobinPolicy) (PYTHON-75)
64+
65+ Other
66+ ^^^^^
67+ * Add 1 second timeout to join() call on event loop thread during
68+ interpreter shutdown. This can help to prevent the process from
69+ hanging during shutdown.
70+
71+ 2.0.0b1
72+ =======
73+ May 6, 2014
74+
75+ Upgrading from 1.x
76+ ------------------
77+ Cluster.shutdown() should always be called when you are done with a
78+ Cluster instance. If it is not called, there are no guarantees that the
79+ driver will not hang. However, if you *do * have a reproduceable case
80+ where Cluster.shutdown() is not called and the driver hangs, please
81+ report it so that we can attempt to fix it.
82+
83+ If you're using the 2.0 driver against Cassandra 1.2, you will need
84+ to set your protocol version to 1. For example:
85+
86+ cluster = Cluster(..., protocol_version=1)
87+
88+ Features
89+ --------
90+ * Support v2 of Cassandra's native protocol, which includes the following
91+ new features: automatic query paging support, protocol-level batch statements,
92+ and lightweight transactions
93+ * Support for Python 3.3 and 3.4
94+ * Allow a default query timeout to be set per-Session
95+
96+ Bug Fixes
97+ ---------
98+ * Avoid errors during interpreter shutdown (the driver attempts to cleanup
99+ daemonized worker threads before interpreter shutdown)
100+
101+ Deprecations
102+ ------------
103+ The following functions have moved from cassandra.decoder to cassandra.query.
104+ The original functions have been left in place with a DeprecationWarning for
105+ now:
106+
107+ * cassandra.decoder.tuple_factory has moved to cassandra.query.tuple_factory
108+ * cassandra.decoder.named_tuple_factory has moved to cassandra.query.named_tuple_factory
109+ * cassandra.decoder.dict_factory has moved to cassandra.query.dict_factory
110+ * cassandra.decoder.ordered_dict_factory has moved to cassandra.query.ordered_dict_factory
111+
112+ 1.1.2
113+ =====
114+ May 8, 2014
115+
116+ Features
117+ --------
118+ * Allow a specific compression type to be requested for communications with
119+ Cassandra and prefer lz4 if available
120+
121+ Bug Fixes
122+ ---------
123+ * Update token metadata (for TokenAware calculations) when a node is removed
124+ from the ring
125+ * Fix file handle leak with gevent reactor due to blocking Greenlet kills when
126+ closing excess connections
127+ * Avoid handling a node coming up multiple times due to a reconnection attempt
128+ succeeding close to the same time that an UP notification is pushed
129+ * Fix duplicate node-up handling, which could result in multiple reconnectors
130+ being started as well as the executor threads becoming deadlocked, preventing
131+ future node up or node down handling from being executed.
132+ * Handle exhausted ReconnectionPolicy schedule correctly
133+
134+ Other
135+ -----
136+ * Don't log at ERROR when a connection is closed during the startup
137+ communications
138+ * Mke scales, blist optional dependencies
139+
140+ 1.1.1
141+ =====
142+ April 16, 2014
143+
144+ Bug Fixes
145+ ---------
146+ * Fix unconditional import of nose in setup.py (github #111)
147+
148+ 1.1.0
149+ =====
150+ April 16, 2014
151+
152+ Features
153+ --------
154+ * Gevent is now supported through monkey-patching the stdlib (PYTHON-7,
155+ github issue #46)
156+ * Support static columns in schemas, which are available starting in
157+ Cassandra 2.1. (github issue #91)
158+ * Add debian packaging (github issue #101)
159+ * Add utility methods for easy concurrent execution of statements. See
160+ the new cassandra.concurrent module. (github issue #7)
161+
162+ Bug Fixes
163+ ---------
164+ * Correctly supply compaction and compression parameters in CREATE statements
165+ for tables when working with Cassandra 2.0+
166+ * Lowercase boolean literals when generating schemas
167+ * Ignore SSL_ERROR_WANT_READ and SSL_ERROR_WANT_WRITE socket errors. Previously,
168+ these resulted in the connection being defuncted, but they can safely be
169+ ignored by the driver.
170+ * Don't reconnect the control connection every time Cluster.connect() is
171+ called
172+ * Avoid race condition that could leave ResponseFuture callbacks uncalled
173+ if the callback was added outside of the event loop thread (github issue #95)
174+ * Properly escape keyspace name in Session.set_keyspace(). Previously, the
175+ keyspace name was quoted, but any quotes in the string were not escaped.
176+ * Avoid adding hosts to the load balancing policy before their datacenter
177+ and rack information has been set, if possible.
178+ * Avoid KeyError when updating metadata after droping a table (github issues
179+ #97, #98)
180+ * Use tuples instead of sets for DCAwareLoadBalancingPolicy to ensure equal
181+ distribution of requests
182+
183+ Other
184+ -----
185+ * Don't ignore column names when parsing typestrings. This is needed for
186+ user-defined type support. (github issue #90)
187+ * Better error message when libevwrapper is not found
188+ * Only try to import scales when metrics are enabled (github issue #92)
189+ * Cut down on the number of queries executing when a new Cluster
190+ connects and when the control connection has to reconnect (github issue #104,
191+ PYTHON-59)
192+ * Issue warning log when schema versions do not match
193+
11941.0.2
2195=====
3- In Progress
196+ March 4, 2014
4197
5198Bug Fixes
6199---------
@@ -15,6 +208,8 @@ Bug Fixes
15208 TimestampType to fix sorting of pre-unix epoch dates in CASSANDRA-5723.)
16209* Handle latest table options when parsing the schema and generating
17210 CREATE statements.
211+ * Avoid 'Set changed size during iteration' during query plan generation
212+ when hosts go up or down
18213
19214Other
20215-----
0 commit comments