Surface errors from regal.parse_module in regal test output

[anderseknert]

It is quite common to have simple mistakes in the policy provided lead to silent failures, and these can be really tedious to debug. It's hard to think of any scenario where you would not want regal.parse_module to fail with a message (the parse error). I'm not sure if it's possible to run something like --strict-builtin-errors but only for selected functions, but something like that would be ideal.

[anderseknert]

I've been pondering this, and I'm now more inclined to explore the model I found out golangci-lint uses for tests. Which is basically that instead of using OPA's test runner, linter rule tests would be written as normal .rego files placed in testdata (or whatever). Each file represents one test case, and expectations/assertions outside of just "this should fail" or "this should not fail" could be described as metadata in the policy. There are certainly a few things that would need to be explored in order to judge whether this model works well for us — not to mention a massive effort. And perhaps we should just try and fix the issue described here first anyway. But this seemed like the best place to note down this idea.