1.1.3 (2016-11-22)

• [FIXED] Incorrect message output from parameter null or empty checks.
• [FIXED] Issue where replications would error if the server returned missing revisions.
• [UPGRADED] Upgraded to version 2.7.0 of the cloudant-http library.

1.1.2 (2016-10-20)

• [FIXED] Issue preventing index updates from being persisted that
  impacted query performance

1.1.1 (2016-10-05)

• [UPGRADED] Upgraded to version 2.6.2 of the cloudant-http library.
• [REMOVED] Removed com.google.guava:guava:15.0 dependency. Applications may need to update their own dependency tree to include guava if they are using it directly, but were relying on this library to include it.

1.1.0 (2016-08-24)

• [NEW] Updated to version 2.5.1 of the cloudant-http library. This
  includes optional support for handling HTTP status code 429 Too Many Requests with blocking backoff and retries. To enable the
  backoff add an instance of a Replay429Interceptor with the desired
  number of retries and initial backoff:
  builder.addResponseInterceptors(new Replay429Interceptor(retries, initialBackoff));
  A default instance is available using 3 retries and starting with a
  250 ms backoff: Replay429Interceptor.WITH_DEFAULTS
• [DEPRECATED] DatastoreManager constructors. Use DatastoreManager.getInstance factory methods
  instead to guarantee only a single DatastoreManager instance is created for a given storage
  directory path in the scope of the DatastoreManager class.
• [FIX] Corrected a case where two root nodes with identical revision IDs prevented selection of the
  correct new winning revision.
• [FIX] Added migration on Datastore opening to repair datastores with duplicated revisions or
  attachments caused by an issue when executing concurrent pull replications with the same source and target.
  Note that in rare circumstances for some documents this may result in a different, but corrected,
  winning revision after migration.
• [FIX] Prevent insertion of multiple revisions with the same document
  ID and revision ID. Previously this could occur when executing
  concurrent pull replications with the same source and target. A
  replication encountering this condition will now terminate in the
  ERROR state.
• [NOTE] Due to migrations outlined above, the first time an existing
  Datastore is opened with this version of the library, users may
  experience a longer than usual delay before the Datastore is
  ready.