Wait-freedom is a term from distributed computing. The algorithm is unavailable if a stream (or a distributed node) is able to correctly shut down, even if all incoming from other flows are delayed / lost at any time.
If you care about consistency, you cannot guarantee freedom of expectation (provided that you always want to finish correctly, that is, guarantee accessibility). This follows from the CAP theorem [1], since freedom of expectation essentially implies the inadmissibility of partitions.
[1] http://en.wikipedia.org/wiki/CAP_theorem
source
share