This document contains various notes for working with Reflow's code base.
It's often useful to run a local standing Reflow server, instead of using -local, e.g., when debugging interactions between the Reflow evaluator and a Reflow server, or debugging work stealing.
First, launch a local Reflow server. (If you launch multiple instances, make sure they have different working directories.)
% reflow serve -addr :9000 -config /tmp/config -dir /tmp/reflowlet
And then use the "static" cluster provider to use them from a Reflow invocation, e.g.,:
% reflow -config /tmp/config -cluster static,localhost:9000 ...
Multiple instances are separated by commas.