Support for atomic batches
The tokens only need to be hex encoded when using ByteOrderedPartitioner.
Make sure connection count is decremented as we are disposing connection...
…ions from pool. Remove unused var.
Added incompatible argument exception
Connection.return_to_pool() also goes through this method
Would close #170
Prevent closed connections from ending up in the pool (Mutator uses "return_to_pool"). Since current_conns has already been decremented, the pool would reach a state where a new connection was always opened but could not be put in the queue since it was full.
When a connection is recycled, and the pool is full, the old connection would be disposed twice.
…on would be disposed twice. This would happen because the Queue.Full exception would be raised, and so "conn"" would be the old connection that has already been disposed. This connection would again be disposed due to the Full exception.
Add "include_ttl" parameter that allows column TTL's to be returned as well
Adding column_count and column_reversed support to ColumnFamilyStub.get and ColumnFamityStub.multiget
…olumns before slicing them. Insert order doesn't matter, so changing the dict object used.
…. It can work alongside include_timestamp.