Hi, In our performance testing, we've discovered that the methods in builtin_metrics seem to be using reflection - a lot of our recorded stack traces include methods in builtin_metrics calling Reflector.invokeInstanceMethod, which then uses reflection to call a method on BuiltinBoltMetrics.
We're sending a significant number of tuples round the topology (hundreds of thousands), and we think that this use of reflection, called for every tuple emitted and executed, is taking up a significant amount of time in the topology and causing a drop in performance. Before filing a bug, in what way does that code in builtin_metrics need to use reflection, and how can it be changed to not use reflection? Thanks, SimonC
