Closed jhellerstein closed 12 years ago
The first point is just message batching, isn't it? i.e., if we allow bunches of network events to co-occur, for similar reasons we might be inclined to allow bunches of clock events to co-occur.
In 08d1a94dac, I added a description of the semantic guarantees we provide for periodics (that is, essentially none). Allowing > 1 periodic event per timestamp seems a natural consequence of allowing event batching in general -- AFAIK there is nothing left to do here. Closing, but reopen if you think we need to do more.