It would be nice to write a test, or at least check once to ensure that the outputs are currently created deterministically.
That is to say, if we run the build once, then run it again without changing the code, and diff each of the outputs before/after, there should be no difference.
Maybe just do a quick check rather than writing a test. Adding a test for this will almost double the time it takes to run tests, as the build will have to be run twice.
Overview
It would be nice to write a test, or at least check once to ensure that the outputs are currently created deterministically.
That is to say, if we run the build once, then run it again without changing the code, and
diff
each of the outputs before/after, there should be no difference.