We should set up a robust testing pipeline that (1) can be run at install-time (and later on request), and (2) is small enough to be shipped with the repo. I imagine it will take the form of running the beamformer in various modes on a small amount of data whose result we know in advance.
We should set up a robust testing pipeline that (1) can be run at install-time (and later on request), and (2) is small enough to be shipped with the repo. I imagine it will take the form of running the beamformer in various modes on a small amount of data whose result we know in advance.