These changes let us test the IBM float un/pack logic around implementation selection (i.e. Python or C++) without actually needing any particular implementation installed. We directly test the logic by using mocks. This means our tests are pretty closely bound to the implementation, but this seems like the cleanest way to get 100% statement coverage.
Coverage increased (+0.1%) to 99.824% when pulling f608e0735632d2b79ab3886828eb12263674bf65 on abingham:pack-implemenation-test-mocks into b1791255d125862e984b4ea50a73860701326f99 on sixty-north:master.
Coverage increased (+0.1%) to 99.824% when pulling f608e0735632d2b79ab3886828eb12263674bf65 on abingham:pack-implemenation-test-mocks into b1791255d125862e984b4ea50a73860701326f99 on sixty-north:master.
These changes let us test the IBM float un/pack logic around implementation selection (i.e. Python or C++) without actually needing any particular implementation installed. We directly test the logic by using mocks. This means our tests are pretty closely bound to the implementation, but this seems like the cleanest way to get 100% statement coverage.