Contract runtime testing.

[g-r-a-n-t]

What is wrong?

We're currently limited to only testing the public interface of a contract. This can make debugging features that rely on multiple new runtime functions difficult. We should be able to test the runtime directly within our contracts test suite.

How can it be fixed

Create a new testing function that takes the following: an executor, contract fixture path, contract name, and a set of Yul statements to inject into the runtime.

It should look something like this:

test_runtime(
  &executor,
  "my_module.fe",
  "MyContract",
  statements! {
    (if (eq(add(1, 1), 2)) {
    (revert(0, 0))
  }
)

What we ultimately want to do is build a Yul object similar to an actual contract, but it only contains the runtime (no constructor) and the test statements. We would then compile this object to bytecode and execute it. If we hit a revert, the test has failed.

We should be able to create a function similar to runtime::build_with_abi_dispatcher, but named build_with_test_statements that simply inserts test statements into the runtime object. We then compile and execute.