Error handling in tests is not clear

[radu-matei]

trafficstars

The current way errors are handled in tests is not at all clear. Theoretically, test implementations could return an error, but the test runtime ignores the errors, even when unwrapped.

This could (?) be an issue in the way the Wasmtime import bindings handle expected<_, error>.

[squillace]

@radu-matei do we care still? Who should give us a path forward?

[radu-matei]

We still care, someone has to go through the test setup process and make sure errors are not silently swallowed at some point.

[Porges]

@radu-matei Theoretically, test implementations could return an error, but the test runtime ignores the errors, even when unwrapped.

The cause is probably what I fixed in #18?