Several golden outputs of tests have non-decomposed swap in their output.
This means that swap wasn't decomposed for this test.
This is fine for testing e.g. the router.
But it is not fine for testing whether the combination of router, decomposition and rc scheduling works for decomposed swaps because
these end-to-end tests are not done now.
So coverage is reduced.
I agree to have simpler tests and have more unit tests.
But we also need to keep some end-to-end tests of semi-realistic applications; when that means that the golden outputs have to be updated on each change of OpenQL, then that is the problem, not the presence of the test and should another way be found to do those end-to-end tests.
Several golden outputs of tests have non-decomposed swap in their output. This means that swap wasn't decomposed for this test. This is fine for testing e.g. the router. But it is not fine for testing whether the combination of router, decomposition and rc scheduling works for decomposed swaps because these end-to-end tests are not done now. So coverage is reduced. I agree to have simpler tests and have more unit tests. But we also need to keep some end-to-end tests of semi-realistic applications; when that means that the golden outputs have to be updated on each change of OpenQL, then that is the problem, not the presence of the test and should another way be found to do those end-to-end tests.