|
igorsk,
in a perfect world where all code is flawless, I'd agree with you -- but I know better.
I've seen in my career where a driver NOT loaded slowed or stalled other processes. Usually these are the obvious bugs -- rarely make it past unit test and never past an integrated test -- but there could be a very subtle bug that costs an extra few microseconds every call that could add up to a tiny, tiny lag. it would be hard to see it without comparing "with" and "without" the driver active.
I agree with your assessment that the symptoms don't make sense, but cwappy seems to have done all due diligence. As a devout empiricist, I firmly believe when a test is correctly executed then the test isn't wrong because the results don't match the theory.
|