Thus you might ask: why did it take so long to actually implement this? Well, simply because it's really hard to to implement, much harder than it might sound on the surface at first:
in order to make the logic work correctly we need to ensure two things: first, the live log output must be fully established by the time the start operation is enqueued. That's relatively easy to make sure. But the other thing is that we must continue the log output until the the start operation completed…