[reading: Barbara Hambly, "The Emancipator's Wife"]
Everyone else's nasty deadlines seem to have passed, so I got some very useful assistance over the last couple of days on the insanely complicated configuration setup for stress testing Project Y.
We pulled the stress test up in the schedule so that we could get a sense of the performance of the thingI couldn't really predict the performance from the raw code itself. Things look good so far: I had 300 calls running in parallel, which looked to use around 5% of a 2.8GHz CPUso on an optimistic assumption of linearity, that should scale to a few thousand calls on the box. Not bad when O(100) calls was actually a fairly low priority on the original list of deliverables.