As a scientific program, we should be more rigorous with our testing, and be sure that if any code modifications we make changes the results, we at least understand why it happened.
We should have two test batteries:
Testing each "basic" function independently, and hopefully automatically.
A more high-level testing where we run the data on "standard brains", and get "standard results". Considering the runtime of the program, this should probably not be done automatically.
These are two somewhat independent projects, which could be interesting for someone getting acquaintanced with the code.
As a scientific program, we should be more rigorous with our testing, and be sure that if any code modifications we make changes the results, we at least understand why it happened.
We should have two test batteries:
These are two somewhat independent projects, which could be interesting for someone getting acquaintanced with the code.