v0.12.1

• Fixed bug in rabbitmq.Consumer. The consumer was not joining the worker threads when a connection closed exception (or any other exception really) is has been thrown during the consumption. The result was that the process did not exit because of the remaining running threads.
• Fixed bug in multiproc_utils.spawn_worker_processes. The function was not guaranteed to join all spawn processes all possible cases.

v0.12.0

• Allow more efficient row deletion in flask_signalbus. Now when the "choose_rows" classmethod is defined, a "with-query" join will be used for deleting rows. This is a breaking change, because the type of the objects passed to the send_signalbus_message and send_signalbus_messages methods have changed (from model instances, to SQLAlchemy result rows).

v0.11.10

• In flask_signalbus, run ANALYZE (SKIP_LOCKED) table before starting to flush a table.

v0.11.9

• Fixed tests

v0.11.8

• In flask_signalbus, when the "choose_rows" classmethod is defined, always force custom plan.

v0.11.7

• In flask_signalbus, when the "choose_rows" classmethod is defined, always use nested joins for row selection.

v0.11.6

• In flask_signalbus, when the "choose_rows" classmethod is defined, always use bitmap scans for row selection.

v0.11.5

• Fixed minor issues

v0.11.4

• Allow more efficient row selection in flask_signalbus. Now if the signal class has defined a special classmethod named "choose_rows", instead of a "select-in" query, a "with-query" join will be used for selecting rows.

v0.11.3

• Added process_individual_blocks property to TableScanners. It
  implements an alternative mode of operation, which is to call the
  process_rows method separately for each non-empty page (block) in
  the table.