We now run all the compatibility tests in the following modes:
Normal (no prefix or default tags)
With prefix
With default tags
We now also have tests for measure and distribution in block execution time measuring mode.
There is also a reliability improvement by making sure that we always emit a packet by stubbing rand. That way, we can simply call the StatsD method once, rather than in a loop, and get rid of all the thread and draining logic.
I've simplified the implementation significantly, by stubbing rand (thanks Jean for the idea) so we are sure that every StatsD invocation will actually emit a datagram.
We now run all the compatibility tests in the following modes:
We now also have tests for
measure
anddistribution
in block execution time measuring mode.There is also a reliability improvement by making sure that we always emit a packet by stubbing rand. That way, we can simply call the StatsD method once, rather than in a loop, and get rid of all the thread and draining logic.